This workshop, hosted by the United Arab Emirates, focused on the theme of 'Beyond Digital' government, exploring how agentic AI and human-centred design are reshaping public services . The UAE's overarching ambition, as outlined in the opening remarks, is to become the first government in the world to run half of its services and operations on agentic AI within two years, a vision referred to as UAE Government 4.0 .
Dina Fares from Digital Ajman presented a 'headless' agentic AI service model, arguing that the next frontier of government is not simply moving services online but enabling services to proactively understand and guide citizens . She demonstrated this through a use case in which an AI agent helped a user renew a commercial licence and negotiate a lease contract across multiple government entities without the user needing to navigate each system separately . Fares cautioned that while AI demos are easy to produce, real implementation must contend with genuine data, rules, integrations, and accountability, noting that only 39% of organisations surveyed by McKinsey in 2025 could prove real ROI on AI projects .
Meshal Abdulla BinHussain from the Ministry of Finance described a three-wave transformation - digitise and integrate, re-engineer services, then automate and infuse AI. This transformation unified 52 federal entities onto shared financial systems . The results included over 52,000 automated transactions, an 87% reduction in processing time, 99.8% reconciliation accuracy, and 50,000 customers served up to nine times faster .
Manal AlAfad from the Telecommunications and Digital Government Regulatory Authority shifted focus to the human dimension, highlighting that 5% of citizens - including elderly people, those with language barriers, and those with cognitive difficulties - remain unable to use digital services . She outlined four principles for a 'human government': accessible by default, human-centred by default, trusted by default, and flexible by default .
The session concluded with the observation that the UAE's journey is not simply about adopting technology but about reimagining government around people, with trust and resilience as its foundation .
Overall Purpose
- The purpose of this UAE workshop is to showcase the UAE's practical journey in deploying agentic AI and digital transformation across government services, sharing real-world use cases from three government entities - Digital Ajman, the General Women's Union, and the Ministry of Finance - with the aim of demonstrating how technology can make government services more human, inclusive, and efficient.
- --
Major Discussion Points
- The UAE's strategic ambition to become the world's first government to run half its services on agentic AI (UAE Government 4.0): The UAE has set a clear national directive to move beyond digital transformation as a goal and treat it as a starting point, with a cabinet mandate to deploy agentic AI across government operations within two years.
- Ajman's agentic AI model, reengineering government services around the customer rather than internal systems: Digital Ajman presented a customer-facing agentic AI service that proactively guides users, checks documents, negotiates on their behalf, and coordinates across multiple government agencies - all without the user needing to understand backend complexity. The key argument was that the real challenge is not building AI demos but proving genuine ROI and maintaining trust, with only 39% of organisations able to demonstrate real returns.
- The Ministry of Finance's phased digital transformation - from legacy systems to live AI use cases with measurable outcomes: The Ministry moved in three waves (digitise and integrate, re-engineer services, then automate and infuse AI), unifying 52 federal entities onto shared financial systems and deploying 26 live AI and automation use cases. Results included an 87% reduction in processing time, 99.8% reconciliation accuracy, and 50,000 customers served up to nine times faster. - The risk of digital exclusion and the need for a 'human government' model that is accessible by default: TDRA's representative challenged the assumption that high digital adoption rates (95%) represent full success, highlighting that the remaining 5% - elderly people, those with language barriers, and those with cognitive difficulties - are systematically left behind. She argued that the future of government must be designed with people, not for them, built on four principles: accessible, human-centred, trusted, and flexible by default. - Governance, trust, and responsible AI deployment as foundational - not optional - elements of agentic AI: Multiple speakers emphasised that AI in government must be traceable, explainable, and policy-aligned. Dina Fares warned that a 'confident wrong answer is more dangerous than a simple 'I don't know,'' and cited Gartner's prediction that over 40% of agentic AI projects will be cancelled by 2027 due to escalating costs, unclear value, or weak risk controls. The Ministry of Finance similarly embedded governance through a Chief AI Officer, an AI Centre of Excellence, and ISO-certified standards.
- --
Overall Tone
- The overall tone of the discussion is confident, aspirational, and pragmatic, with a strong undercurrent of national pride in the UAE's digital achievements. Speakers were enthusiastic about the potential of agentic AI but were notably candid about the difficulties of real-world implementation. This balanced optimism with realism gave the session credibility.
- The tone shifted slightly towards the end, becoming more reflective and humanistic as Manal AlAfad redirected attention from technological achievement to social inclusion, questioning whether governments were truly serving all citizens. The moderator's closing remarks reinforced this shift, summarising the session's thread as being 'not simply about adopting technology but about reimagining government around people' . There were also moments of levity, which kept the atmosphere warm and collegial throughout.
Beyond Digital: UAE Government Workshop on Agentic AI and Human-Centred Public Services
#
Opening and Strategic Vision
The workshop, titled "Beyond Digital," was hosted by the United Arab Emirates as a country session at what appeared to be an international forum in partnership with the ITU . Opening remarks were delivered by His Excellency Majid al-Mismar, who, after a lighthearted moment in which he acknowledged having read from the wrong script , set out the UAE's strategic ambition with clarity. He emphasised that for the UAE, digital transformation was never the end goal but rather the starting point , with the overarching aim of putting government services within easy reach of every citizen, resident, and visitor whilst earning their trust . Under the directive of the UAE's leadership, the country has set out to become the first government in the world to run half of its services and operations on agentic AI within two years - a journey referred to as UAE Government 4.0 . This vision, he explained, carries government beyond the digital toward an institution that does not merely inform decisions but helps to carry them out . The session was structured around three real-world use cases from the General Women's Union, the Department of Digital Ajman, and the Ministry of Finance , with the moderator, Mohamed Bushlaibi, introducing each speaker in turn .
---
#
Digital Ajman: Headless Government and the Agentic AI Frontier
The first substantive presentation was delivered by Dina Fares, Executive Director of Digital Transformation at Digital Ajman, who opened with a provocation: when was the last time a complex government service that required more than one entity truly felt effortless ? She argued that the evolution of digital governance - from counters to websites, from websites to apps, and from apps to unified super-app platforms - had reached a new inflection point . The question was no longer whether a citizen could access a service with minimum effort, but whether the service could proactively go to the citizen, understand their needs, guide and advise them, and help them complete what they needed to do . This shift, she argued, is where agentic AI becomes genuinely powerful - not as a decorative chatbot sitting atop a webpage, but as a real service player .
Ajman's approach was built around the concept of a "headless" service: one in which the intelligence of the service - what Fares called the "service brain," comprising rules, data, processes, systems, policies, and controls - is separated from the user interface . This architecture means the interface can change across channels (app, kiosk, voice, or channels not yet imagined), whilst the underlying service logic remains consistent and strong . Fares was careful to distinguish between AI deployed internally, where employees can tolerate complexity and ambiguity, and AI facing citizens directly, who see only the government and for whom a confusing experience becomes not merely a technical issue but a trust issue .
To illustrate the model in practice, Fares presented a video demonstration (in Arabic with subtitles) following a fictional user named Omar, who needed to renew a commercial licence and renegotiate a lease contract . The agentic AI service automatically checked his documents, identified that his rental contract had expired, facilitated a negotiation with his landlord within the same session, and coordinated across two government agencies - the economic department responsible for trade licences and the municipality responsible for lease contracts - without Omar needing to navigate each system separately . Fares emphasised that what appeared simple on screen was underpinned by significant backend complexity: "the real work works at the background" . She therefore characterised the project not as a technology initiative but as "a service reengineering product, a data readiness project, an integration, a governance project that creates partnerships" .
Fares was notably candid about the challenges of real-world AI implementation. Citing a 2025 McKinsey AI Global Survey, she noted that only 39% of organisations were able to prove real return on investment on their AI projects , arguing that the challenge is not using AI but turning it into measurable value - defined for governments as fewer steps, less waiting, clearer journeys, better accuracy, and more trust . She also referenced a Gartner prediction that over 40% of agentic AI projects would be cancelled by the end of 2027 due to escalating costs, unclear business value, or weak risk controls . Her response to these warnings was not to advocate caution for its own sake but to argue for building better: "innovation without control becomes noise, and control without innovation becomes a delay" . She proposed a traffic-light governance framework - green when the path is clear, yellow when assessment is needed, and red when risk, policy, or trust is questionable - as a means of moving with direction, judgement, and control rather than blindly .
A particularly striking observation was her warning that "a confident wrong answer is more dangerous than a simple 'I don't know'" . In government, she argued, AI cannot merely sound right - it must be right, traceable, explainable, and aligned with policy . Trust, she concluded, is not something added at the end of a project but part of the architecture itself, likening governance controls to wearing a seatbelt in a fast car: not because a crash is planned, but to move more safely and with confidence . The ultimate goal, she stated, is not for the government to say it used AI, but for citizens to say the service was easier than they expected - because when governments become easier, people gain confidence in their institutions .
---
#
General Women's Union: National Strategy and Digital Empowerment
Ghalya AlMannaee, introduced by the moderator as Chairwoman of Strategic [title incomplete in transcript] at the General Women's Union, provided an overview of the UAE's institutional framework for advancing women's empowerment. She described a Cabinet-approved national strategy for tracking the progress of women, supported by over 155 partnerships spanning federal, local, and private sector entities . A national policy for empowering women provides a framework and guidelines for all federal and local entities across the UAE . This policy is structured around several priorities, of which she explicitly named: building strong family cohesion to enhance women's quality of life; integrating women into the workforce, particularly in future sectors; and building accountability and digital faith . Further priorities were mentioned in passing but were not clearly enumerated as distinct items in her remarks.
AlMannaee also highlighted several digital and cultural initiatives. The UAE is deploying virtual reality tools to help new generations learn traditional heritage crafts, including one called SEDU, described as a traditional handicraft . The Medjari platform has been developed to empower productive families in conducting their businesses , whilst the UAE Women Portal serves as the first national archive for Emirati women . A dedicated platform for women in sport has also been established . AlMannaee closed by inviting international partners to engage through official channels to benefit from these initiatives , signalling the UAE's openness to global collaboration on women's empowerment.
---
#
Ministry of Finance: From Legacy Systems to Live AI at Scale
Meshal Abdulla BinHussain, Director of the Digital Transformation Department at the Ministry of Finance, framed his presentation as a story about accountability - how financial services were made faster and more trusted through digital technology and AI, and how others could do the same . He acknowledged that the Ministry's starting point was a familiar one: financial services running on manual reviews, legacy systems, and multi-step approvals that were slow and painful for citizens, businesses, and the government itself . The transformation was driven by a strong mandate comprising a national digital government strategy, a zero government bureaucracy programme, a national AI strategy, and a Cabinet directive to run half of services and processes on agentic AI .
BinHussain described a three-wave transformation approach in which the sequencing was explicitly identified as critical . The first wave involved digitising and integrating: unifying financial systems into one real-time source of truth, with the explicit acknowledgement that "you cannot put AI on scattered data" . The second wave involved re-engineering services around the customer through Services 2.0 and zero-bureaucracy initiatives, cutting redundant steps and co-designing services with the people who actually use them . Only in the third wave was AI introduced - beginning with robotics automation and generative AI, and advancing toward agentic AI that actually acts . This sequencing reflects a disciplined approach to transformation that prioritises foundational infrastructure before advanced technology deployment.
The foundational infrastructure built by the Ministry was substantial. A sovereign data platform, a central data lake, open data capabilities, and advanced analytics were constructed to turn scattered numbers into decisions . Fifty-two federal entities were unified onto shared financial systems, including financial management, budgeting, treasury, digital procurement, and federal properties platforms - all connected through the national digital identity (UAE Pass) and the federal government network . On top of this foundation, 26 live AI and automation use cases were deployed in finance, including a generative AI platform and a legal researcher trained on financial and tax legislation . Governance was embedded through the appointment of a Chief AI Officer, an AI Centre of Excellence, and AI champions within every business department , with adherence to ISO 27001 information security standards and a stated intention to progress toward ISO 42001 for AI management .
The results were presented with extensive quantitative evidence. Over 52,000 manual transactions - spanning payroll, invoices, settlements, and reconciliation - were automated, with processing time reduced by 87% and reconciliation accuracy reaching 99.8% . Fifty thousand customers were served up to nine times faster, and 15,000 working hours were returned to employees for strategic and judgement-based purposes . On digital procurement, vendor registration was reduced from 30 days to one day, and catalogue purchasing from 60 days to just six minutes, with over 130 million AED saved . An in-house intelligent call centre sentiment analysis solution analysed more than 58,000 customer calls, achieving 97.6% first-call resolution, 96.7% customer satisfaction, and 80% of calls answered within 20 seconds . BinHussain noted that for the first time, the Ministry could hear every customer rather than just a sample - a remark that reframed AI not merely as an efficiency tool but as a means of achieving comprehensive citizen listening at scale.
---
#
TDRA: The Human Dimension and the Excluded 5%
The final presentation, delivered by Manal AlAfad, Manager of Digital Adoption Enablement at TDRA, introduced a deliberately different perspective. Where earlier speakers had celebrated high adoption rates and efficiency gains, AlAfad opened by asking the audience to consider not the 95% of digital services that are used, but the 5% that are not . She identified the citizens within that 5% as elderly people who cannot navigate websites and mobile applications, people with language barriers, and people with cognitive challenges related to web design . Her central argument was that for years, governments had measured the success of services by adoption and satisfaction rates without caring about how inclusive or human those services truly were .
AlAfad situated this concern within the broader trajectory of digital government, describing a progression from paper-based services to electronic services, to connected government, to agentic AI government, and ultimately to what she termed a "human government" . She argued that the UAE's leadership focuses on the human dimension, and that the principle is to design with people for people rather than designing for them and forcing them to use it . She described the UAE's federal digital accessibility policy, which has been formally adopted by the federal government and requires services to be designed with users, tested for inclusivity, and audited by TDRA . A structured plan has been put in place for federal entities to re-engineer their services and co-design them with users, including people of determination and elderly people .
AlAfad outlined four principles for a human government: accessible by default, human-centred by default, trusted by default, and flexible by default . Trusted by default means services are secure and built on connected data and connected entities . Flexible by default means services can be designed, fixed quickly, and audited simultaneously . Her contribution offered a complementary perspective on inclusion alongside the session's focus on technological capability - noting that as governments become more technologically mature, there will always be people who cannot use the technology as a regular user, including the speakers themselves in the future . This reorientation from technological achievement to social inclusion gave the session a reflective and humanistic dimension.
---
#
Convergences, Tensions, and Closing Reflections
Across the session, a high degree of consensus emerged on several foundational principles. All speakers affirmed that technology is an instrument to serve citizens rather than an end in itself . Multiple speakers identified trust as a non-negotiable architectural requirement: Fares argued it must be embedded in the design of services , BinHussain institutionalised it through formal governance structures , and AlAfad articulated it as one of the four default principles of human government . Both Fares and BinHussain independently converged on the necessity of solid data and service foundations before AI deployment , and both emphasised that real-world implementation is fundamentally harder than demonstrations suggest .
Meaningful points of contrast also ran through the session. The ambitious top-down directive for the UAE to run half its services on agentic AI within two years sat alongside Fares's candid citation of industry data on project failure rates and her advocacy for a measured, traffic-light approach to risk . BinHussain's extensive quantitative metrics - framed as evidence of success - were complemented by AlAfad's argument that adoption and satisfaction rates do not fully capture the experience of those who are left behind . More broadly, AlAfad's closing contribution introduced an important reminder that technological progress must be considered alongside equity and inclusion .
The moderator, Mohamed Bushlaibi, synthesised these threads in his closing remarks, observing that one clear thread ran through all that had been heard: the UAE's journey is not simply about adopting technology but about reimagining government around people, making services smarter, more inclusive, and more human, with trust and resilience as the foundation . This framing - which incorporated both the technological ambition of the earlier presentations and the humanistic perspective of the final one - provided a coherent conclusion to a session that had moved, progressively and deliberately, from visionary aspiration through practical implementation to a reflective examination of what it truly means to serve all citizens.
UAE's agentic AI ambition - The UAE aims to become the first government to run half of its services and operations on agentic AI within two years, under a directive called UAE Government 4.0
Arg. 1The UAE has set an ambitious target to become the world's first government to operate half of its services and processes on agentic AI within a two-year timeframe. This initiative, called UAE Government 4.0, represents a strategic leap beyond conventional digital transformation. It is driven by a leadership directive and positions the UAE as a global pioneer in AI-powered governance.
Speaker 1 explicitly stated that the UAE has set out to become the first government in the world to run half of its services and operations on agentic AI , with a two-year timeline , and named this journey UAE Government 4.0 .
on: The primary focus of digital government: efficiency and automation versus human inclusion and accessibility
Digital transformation as a starting point - Digital transformation was never the end goal for the UAE; it was the starting point toward a government that actively helps carry out decisions, not just inform them
Arg. 2The UAE frames digital transformation not as a destination but as a foundation upon which more advanced, action-oriented government services are built. The ultimate aim has always been to place government services within easy reach of every citizen, resident, and visitor while earning their trust. The next step is a government that does not merely inform decisions but actively helps to carry them out.
Speaker 1 clarified that for the UAE, digital transformation was never the goal but the starting point , with the aim of putting government services within easy reach of every citizen, resident, and visitor , and described the journey as moving beyond digital toward a government that helps carry out decisions .
on: Technology serves people, not the other way around
Reimagining government around people - The UAE's journey is about reimagining government around people, making services smarter, more inclusive, and more human, with trust and resilience as the foundation
Arg. 1In his closing remarks, the moderator synthesised the session's key theme: that the UAE's digital journey is fundamentally about placing people at the centre of government design. Rather than simply adopting technology, the UAE is reimagining the entire structure of government services to be smarter, more inclusive, and more human. Trust and resilience are identified as the foundational principles underpinning this transformation.
Mohammad Bushlaibi summarised the session by stating that one clear thread ran through all presentations - the UAE's journey is not simply about adopting technology but about reimagining government around people, making services smarter, more inclusive, and more human, with trust and resilience as the foundation .
on: Trust is a foundational, non-negotiable element of AI-powered government services
Proactive service delivery - The next shift in digital governance is moving from services that citizens access to services that proactively go to citizens, understand their needs, and guide them through completion
Arg. 1Dina Fares argues that the evolution of digital government has progressed from counters to websites, then to apps and super apps, but the next fundamental shift is a change in the direction of service delivery. Rather than asking whether citizens can access services with minimum effort, the question becomes whether the service can proactively reach the citizen, understand their needs, and guide them to completion. This represents a paradigm shift from reactive to proactive government service.
Fares traced the evolution of digital governance from counters to websites, apps, and unified platforms , and articulated the new question as whether the service can go first to the customer, understand their needs, guide and advise them, and help them complete what they need . She illustrated this with the case of Omar, where the agentic AI service proactively guided him through renewing his commercial licence and lease contract without him needing to navigate multiple government agencies .
on: Technology serves people, not the other way around
on: The primary focus of digital government: efficiency and automation versus human inclusion and accessibility
Agentic AI as a service player - Agentic AI is not a decorative chatbot but a real service player that understands context, checks requirements, guides customers, and supports action, with value measured by outcomes rather than conversation
Arg. 2Fares distinguishes agentic AI from conventional chatbots, arguing that it functions as a genuine service participant rather than a superficial interface addition. It understands context, verifies documents, guides users, and takes action on their behalf. Crucially, the value of agentic AI lies not in the quality of the conversation it generates but in the tangible outcomes it delivers for the user.
Fares explicitly stated that agentic AI should not be a chatbot sitting on top of a page but a real service player , and described how it understands context, checks requirements, checks documents, guides the customer, and supports action . She emphasised that the value is not in the conversation but in the outcome , and demonstrated this through the Omar video case study where the AI agent completed multiple government processes on his behalf .
Headless service architecture - The 'headless' government service model separates the service intelligence (the 'service brain') from the interface, allowing the same underlying service to be delivered across any channel, present or future
Arg. 3Fares introduces the concept of a 'headless' government service, where the intelligence and logic of the service — the 'service brain' comprising rules, data, processes, systems, policies, and controls — is decoupled from any specific user interface. This architecture means the service can be delivered through an app, a kiosk, voice, or any future channel yet to be imagined, without the underlying service logic needing to change. The interface may evolve, but the service remains consistent and strong.
Fares described the service brain as the set of rules, data, processes, systems, policies, and controls that the customer should not need to understand , and explained that the intelligence should not be locked within any single app, page, or platform, as customers may access services through apps, kiosks, or voice today and through unimagined channels tomorrow . She described Ajman's project as headless, AI-based, and built with a new mindset of provisioning government service .
on: Solid data and systems foundations must precede effective AI deployment
Trust implications of customer-facing AI - Customer-facing AI is fundamentally different from internal AI because customers do not understand government systems or workflows; a confusing experience becomes a trust issue, not merely a technical one
Arg. 4Fares draws a critical distinction between AI deployed for internal government employees and AI that faces citizens directly. Employees can tolerate complexity and understand organisational language, but customers see only the government as a whole and do not understand its internal systems or rules. When the customer-facing experience is confusing, it does not remain a technical problem — it erodes public trust in government itself.
Fares explained that when AI is internal, the user is an employee who understands the language and complexity of the organisation , but when AI faces customers, the situation is entirely different because customers do not see systems, do not know rules, and do not understand workflows - they see only the government . She concluded that if the experience is confusing, it becomes a trust issue, not just a technical issue .
on: Trust is a foundational, non-negotiable element of AI-powered government services
Proving AI's ROI - Only 39% of organisations were able to prove real ROI on AI projects according to a 2025 McKinsey survey, demonstrating that the challenge is not adopting AI but turning it into measurable value
Arg. 5Fares cites a 2025 McKinsey global AI survey to highlight that while AI adoption is widespread, the ability to demonstrate genuine return on investment remains elusive for the majority of organisations. This finding underscores that the real challenge is not the deployment of AI technology itself but the translation of that technology into concrete, measurable value. For governments, this value is defined in terms of fewer steps, less waiting, clearer journeys, better accuracy, and greater trust.
Fares referenced a 2025 McKinsey AI global survey which found that AI is widely adopted but that ROI was not being proven, with only 39% of organisations able to demonstrate real ROI on their projects . She used this to argue that the challenge is not using AI but turning it into value, and for governments that value means fewer steps, less waiting, clear journeys, better accuracy, and more trust .
on: Measurable, real-world outcomes are the true measure of AI value in government
Risk of project failure - Gartner predicts that over 40% of agentic AI projects will be cancelled by 2027 due to escalating costs, unclear business value, or weak risk controls, highlighting the need to build better rather than stop
Arg. 6Fares draws on a Gartner report to warn that a significant proportion of agentic AI projects are at risk of cancellation within a few years, primarily due to uncontrolled costs, a failure to demonstrate business value, and inadequate risk governance. Rather than using this as a reason to halt AI initiatives, she argues it is a call to build more responsibly and effectively. The response should be to improve the quality and governance of AI projects, not to abandon innovation.
Fares cited a Gartner report predicting that over 40% of agentic AI projects will be cancelled by the end of 2027 due to escalating costs, unclear business value, or weak risk controls . She responded to this by stating that the answer is not to stop but to build better, because innovation without control becomes noise and control without innovation becomes a delay .
on: Responsible AI implementation requires governance, control, and accountability alongside innovation
on: Pace of AI deployment: ambition versus caution
Balancing innovation and control - Innovation must be balanced with control using a traffic-light approach: green when the path is clear, yellow when assessment is needed, and red when risk, policy, or trust is questionable
Arg. 7Fares proposes a practical framework for managing the tension between the need to innovate rapidly and the imperative to maintain control and accountability in government AI. Using a traffic-light metaphor, she advocates for a structured approach where progress is permitted when conditions are safe, paused for assessment when uncertainty arises, and halted when risk, policy compliance, or trust is in question. This allows governments to move with speed but not recklessly.
Fares articulated the traffic-light framework, stating that green signals a clear path, yellow means stopping to assess, and red indicates that risk, policy, or trust is questionable . She summarised the principle as moving fast but not blindly - moving with direction, judgement, and control .
on: Responsible AI implementation requires governance, control, and accountability alongside innovation
on: Pace of AI deployment: ambition versus caution
AI accuracy and accountability - In government, AI cannot merely sound right; it must be right, traceable, explainable, and aligned with policy, because a confident wrong answer is more dangerous than admitting uncertainty
Arg. 8Fares stresses that the standards for AI accuracy in government are uniquely high because the consequences of error directly affect citizens' lives and trust in public institutions. An AI system that delivers a confident but incorrect answer is more harmful than one that acknowledges its limitations. Government AI must therefore be not only accurate but also traceable, explainable, and fully aligned with existing policy frameworks.
Fares warned that a confident wrong answer is more dangerous than a simple 'I don't know' , and stated that in government, AI cannot just sound right - it must be right, traceable, explainable, and aligned with policy . She noted that this is where many projects struggle .
on: Trust is a foundational, non-negotiable element of AI-powered government services
on: The role of AI accuracy and accountability: confidence versus correctness
Three-wave transformation approach - The Ministry of Finance moved in three sequential waves—digitise and integrate, re-engineer services around the customer, then automate and infuse AI—with the order of these steps being critical to success
Arg. 1BinHussain describes the Ministry of Finance's transformation as a deliberate, sequenced journey across three waves, emphasising that the order in which these steps were taken was critically important. The first wave involved digitising and integrating systems into a unified source of truth. The second re-engineered services around the customer, eliminating redundant steps, before the third wave introduced automation and AI on top of this solid foundation.
BinHussain outlined the three waves explicitly: first, digitise and integrate by unifying financial systems into one real-time source of truth; second, re-engineer services around the customer and cut redundant steps through Services 2.0 and zero-bureaucracy initiatives; and third, automate and infuse AI, starting with robotics automation and generative AI . He stressed that the order of these waves mattered a lot .
on: Solid data and systems foundations must precede effective AI deployment
on: Pace of AI deployment: ambition versus caution
Data and systems integration as a foundation - Foundational infrastructure, including a sovereign data platform, a central data lake, and 52 federal entities unified onto shared financial systems, was essential before AI could be effectively deployed
Arg. 2BinHussain argues that robust data and systems infrastructure must be established before AI can deliver meaningful results, encapsulated in the principle that 'you cannot put AI on scattered data.' The Ministry built a sovereign data platform and central data lake, and unified 52 federal entities onto shared financial systems, creating a single source of truth. This foundational work was a prerequisite for the effective deployment of AI and automation.
BinHussain stated that you cannot put AI on scattered data and described the construction of a sovereign data platform, a central data lake, open data, and advanced analytics . He noted that 52 federal entities were unified onto shared financial systems including financial management, budgeting, treasury, digital procurement, and federal properties platforms, all connected through the national digital identity and UAE Pass .
on: Solid data and systems foundations must precede effective AI deployment
Measurable outcomes of AI adoption - Concrete results demonstrate significant impact: over 52,000 manual transactions automated, processing time reduced by 87%, reconciliation accuracy at 99.8%, and 50,000 customers served up to nine times faster
Arg. 3BinHussain presents a comprehensive set of quantitative results to demonstrate the tangible impact of the Ministry's AI and automation programme. These metrics span transaction volumes, processing speed, accuracy, and customer service, providing concrete evidence that the transformation has delivered real value. The figures also include improvements in call centre performance and procurement timelines, painting a picture of systemic efficiency gains across the ministry.
BinHussain reported that more than 52,000 manual transactions are now automated across payroll, invoices, settlements, and reconciliation, with processing time reduced by 87% and reconciliation accuracy reaching 99.8% . He also noted that 50,000 customers are served up to nine times faster and 15,000 working hours have been returned to employees for strategic purposes . The call centre AI achieved 97.6% first-call resolution, 96.7% customer satisfaction, and 80% of calls answered within 20 seconds . Vendor registration was reduced from 30 days to 1 day and catalogue purchasing from 60 days to 6 minutes .
on: Measurable, real-world outcomes are the true measure of AI value in government
on: The role of AI accuracy and accountability: confidence versus correctness
AI governance structure - AI governance requires dedicated leadership and structure, including a Chief AI Officer, an AI Centre of Excellence, AI champions in every department, and adherence to international standards such as ISO 27001
Arg. 4BinHussain emphasises that effective AI deployment in government cannot happen without a formal governance structure that assigns clear accountability and embeds AI oversight throughout the organisation. This includes appointing a Chief AI Officer, establishing an AI Centre of Excellence, and designating AI champions within every business department. Adherence to international standards such as ISO 27001 and the emerging ISO 42001 for AI management further ensures that governance is rigorous and internationally recognised.
BinHussain described the governance structure as including a Chief AI Officer responsible for the whole agenda, an AI Centre of Excellence, and AI champions within every single business department . He noted that AI use policies and open data policies are in place, that the ministry follows ISO 27001 information security standards with segregation of duties, and that they are moving towards ISO 42001 for AI management .
on: Responsible AI implementation requires governance, control, and accountability alongside innovation
National women's advancement strategy - The UAE has a national strategy for the advancement of women, approved by the Cabinet, with over 155 partnerships across federal, local, and private sectors to track women's progress
Arg. 1AlMannaee describes a Cabinet-approved national strategy dedicated to tracking and advancing the progress of women in the UAE. The strategy is supported by an extensive network of over 155 partnerships spanning federal government, local government, and the private sector. This broad coalition of partners reflects the UAE's commitment to a whole-of-society approach to women's advancement.
AlMannaee stated that the national strategy for the advancement of women has been approved by the UAE Cabinet and is used for tracking the progress of women, with more than 155 partnerships from federal, local, and private sector entities .
National women's empowerment policy - A national policy for empowering women provides a framework and guidelines for all federal and local entities, with five priorities including integrating women into future sectors and building digital accountability
Arg. 2AlMannaee outlines a national policy for empowering women that serves as a comprehensive framework and set of guidelines for all federal and local entities across the UAE. The policy has five main priorities, which include strengthening family cohesion, integrating women into future-oriented work sectors, and building accountability and digital faith. It also encompasses developing supportive regulations and fostering institutional partnerships.
AlMannaee described the national policy for empowering women as providing a framework and guideline for all federal and local entities in the UAE , with five main priorities: building strong family cohesion to enhance women's quality of life, integrating women in the work field especially in future sectors, building accountability and digital faith, developing partnerships between institutions, and developing supportive regulations .
Digital platforms for women's empowerment - Digital and cultural platforms such as the UAE Women Portal, a platform for women in sport, and VR tools for heritage learning are being deployed to empower women and preserve national identity
Arg. 3AlMannaee highlights a range of digital and cultural initiatives being deployed to empower women and preserve Emirati heritage. These include the UAE Women Portal, described as the first national archive for Emirati women, a dedicated platform for women in sport, and the use of virtual reality to help new generations learn traditional handicrafts. The Medjari platform is also mentioned as a tool to empower productive families in conducting their businesses.
AlMannaee mentioned the use of VR to help new generations learn heritage, specifically referencing a traditional handicraft called SEDU being developed with SEDU . She also described Medjari as a platform to empower productive families for doing their business , the UAE Women Portal as the first national archive for Emirati women , and a platform for women in sport .
The excluded 5% - Whilst 95% of digital services are used by the general population, the remaining 5% represents elderly people, those with language barriers, and those with cognitive or accessibility challenges who are being left behind
Arg. 1AlAfad opens her presentation by challenging the audience to consider who is not being served by digital government services. While 95% of digital services are in use, the remaining 5% represents a population that is systematically excluded — including elderly people who cannot navigate digital interfaces, those with language barriers, and those with cognitive challenges. This framing reorients the conversation from aggregate adoption rates to the human cost of exclusion.
AlAfad posed the question of what happens to the 5% of digital services not being used , and explained that this 5% represents elderly people who cannot navigate websites and mobile apps as regular users, people with language barriers, and those with cognitive issues with website design .
on: The primary focus of digital government: efficiency and automation versus human inclusion and accessibility
Rethinking success metrics - Governments have historically measured success by adoption and satisfaction rates without considering how inclusive or human their services truly are, which must change as technology advances
Arg. 2AlAfad argues that the conventional metrics governments use to evaluate digital services — adoption rates and satisfaction scores — are insufficient because they fail to capture whether services are genuinely inclusive and human. As technology becomes more advanced, the risk of leaving certain populations further behind increases. Governments must therefore broaden their definition of success to include inclusivity and humanity as core measures.
AlAfad stated that for years governments have measured the success of services by adoption and satisfaction, but no one has cared about how inclusive or human services are . She warned that as society becomes more mature and technological, there will be people who cannot use technology as regular users, including the speakers themselves in the future .
on: Defining government success: technology adoption metrics versus inclusivity and human-centredness
Digital accessibility policy - The UAE has issued and enforced a digital accessibility policy adopted by the federal government, requiring services to be designed with users, tested for inclusivity, and audited by TDRA
Arg. 3AlAfad describes a concrete policy intervention by the UAE federal government to address digital exclusion: a digital accessibility policy that has been formally issued and enforced across federal entities. The policy requires that services be designed in collaboration with users, tested for inclusivity, and then measured and audited by TDRA to ensure compliance. Federal government entities have also been given a comprehensive plan to re-engineer their services for people of determination and elderly users.
AlAfad stated that two years ago the UAE issued and enforced a digital accessibility policy adopted by the federal government, which requires services to be tested and designed with the user and then measured and audited by TDRA to assess how inclusive they are . She also noted that a large plan was put in place by federal government entities to re-engineer their services and design them with users for people of determination and elderly people .
on: Responsible AI implementation requires governance, control, and accountability alongside innovation
Four principles of human government - The future of government is a 'human government' built on four principles: accessible by default, human-centred by default, trusted by default, and flexible by default, designing with people rather than for them
Arg. 4AlAfad articulates a vision for the future of government centred on four foundational principles that together define what she calls a 'human government.' These principles — accessibility, human-centredness, trust, and flexibility — are framed as defaults rather than optional features, meaning they should be embedded into the design of all government services from the outset. The overarching philosophy is one of co-design: building with people rather than imposing solutions upon them.
AlAfad shared four principles for human government: accessible by default, human-centred by default, trusted by default (meaning secured and built on connected data and entities), and flexible by default (meaning it can be designed, fixed quickly, and audited) . She described the UAE's leadership as focusing on the human, designing with people for people rather than designing for them and forcing them to use it .
on: Technology serves people, not the other way around
on: The primary focus of digital government: efficiency and automation versus human inclusion and accessibility
Session Knowledge Graph
Speakers · Topics · Arguments · Relationships
All speakers consistently affirm that technology is an instrument to serve citizens rather than an end in itself. Speaker 1 stated that digital transformation was never the goal but the starting point , with the aim of putting services within easy reach of every citizen . Fares argued that the question has shifted from whether citizens can access services to whether services can proactively go to citizens and understand their needs . AlAfad emphasised that the UAE's leadership focuses on the human, designing with people for people rather than designing for them and forcing them to use it . Bushlaibi synthesised this in his closing remarks, stating that the UAE's journey is not simply about adopting technology but about reimagining government around people .
Digital transformation as a starting point - Digital transformation was never the end goal for the UAE; it was the starting point toward a government that actively helps carry out decisions, not just inform them
Reimagining government around people - The UAE's journey is about reimagining government around people, making services smarter, more inclusive, and more human, with trust and resilience as the foundation
Proactive service delivery - The next shift in digital governance is moving from services that citizens access to services that proactively go to citizens, understand their needs, and guide them through completion
Four principles of human government - The future of government is a 'human government' built on four principles: accessible by default, human-centred by default, trusted by default, and flexible by default, designing with people rather than for them
Multiple speakers converge on trust as a core architectural and governance requirement. Fares argued that when customer-facing AI experiences are confusing, it becomes a trust issue rather than merely a technical one , and that in government, AI cannot just sound right - it must be right, traceable, explainable, and aligned with policy . She also described trust as part of the architecture, not something added at the end . BinHussain embedded trust through formal governance structures including a Chief AI Officer, AI Centre of Excellence, and adherence to ISO 27001 . Bushlaibi identified trust and resilience as the foundation of the UAE's entire digital journey .
Trust implications of customer-facing AI - Customer-facing AI is fundamentally different from internal AI because customers do not understand government systems or workflows; a confusing experience becomes a trust issue, not merely a technical one
AI accuracy and accountability - In government, AI cannot merely sound right; it must be right, traceable, explainable, and aligned with policy, because a confident wrong answer is more dangerous than admitting uncertainty
AI governance structure - AI governance requires dedicated leadership and structure, including a Chief AI Officer, an AI Centre of Excellence, AI champions in every department, and adherence to international standards such as ISO 27001
Reimagining government around people - The UAE's journey is about reimagining government around people, making services smarter, more inclusive, and more human, with trust and resilience as the foundation
Speakers across the session agree that AI deployment must be governed responsibly. Fares warned that innovation without control becomes noise and control without innovation becomes a delay , proposing a traffic-light framework to move with direction, judgement, and control . She also cited Gartner's prediction that over 40% of agentic AI projects will be cancelled by 2027 due to weak risk controls , arguing the response must be to build better . BinHussain institutionalised this through formal governance structures including a Chief AI Officer and adherence to ISO 27001 and the emerging ISO 42001 for AI management . AlAfad described the UAE's digital accessibility policy as requiring services to be tested, measured, and audited by TDRA .
Balancing innovation and control - Innovation must be balanced with control using a traffic-light approach: green when the path is clear, yellow when assessment is needed, and red when risk, policy, or trust is questionable
Risk of project failure - Gartner predicts that over 40% of agentic AI projects will be cancelled by 2027 due to escalating costs, unclear business value, or weak risk controls, highlighting the need to build better rather than stop
AI governance structure - AI governance requires dedicated leadership and structure, including a Chief AI Officer, an AI Centre of Excellence, AI champions in every department, and adherence to international standards such as ISO 27001
Digital accessibility policy - The UAE has issued and enforced a digital accessibility policy adopted by the federal government, requiring services to be designed with users, tested for inclusivity, and audited by TDRA
Both Fares and BinHussain emphasise that the value of AI must be demonstrated through concrete, measurable outcomes rather than technological impressiveness. Fares cited a 2025 McKinsey survey finding that only 39% of organisations could prove real ROI on AI projects , and argued that the challenge is not using AI but turning it into value - defined for governments as fewer steps, less waiting, clearer journeys, better accuracy, and more trust . She also stated that the question is not how impressive the technology is but whether the service became easier . BinHussain provided extensive quantitative evidence of impact, including over 52,000 manual transactions automated, processing time reduced by 87%, reconciliation accuracy at 99.8%, and 50,000 customers served up to nine times faster .
Proving AI's ROI - Only 39% of organisations were able to prove real ROI on AI projects according to a 2025 McKinsey survey, demonstrating that the challenge is not adopting AI but turning it into measurable value
Measurable outcomes of AI adoption - Concrete results demonstrate significant impact: over 52,000 manual transactions automated, processing time reduced by 87%, reconciliation accuracy at 99.8%, and 50,000 customers served up to nine times faster
Both Fares and BinHussain stress that robust underlying infrastructure is a prerequisite for meaningful AI deployment. Fares described the 'service brain' - the set of rules, data, processes, systems, policies, and controls - as the essential foundation that allows the interface to change while the service remains strong . BinHussain was explicit that 'you cannot put AI on scattered data' , and described how the Ministry first unified 52 federal entities onto shared financial systems and built a sovereign data platform and central data lake before deploying AI . His three-wave approach placed digitisation and integration as the first and essential step, with AI only introduced in the third wave .
Headless service architecture - The 'headless' government service model separates the service intelligence (the 'service brain') from the interface, allowing the same underlying service to be delivered across any channel, present or future
Data and systems integration as a foundation - Foundational infrastructure, including a sovereign data platform, a central data lake, and 52 federal entities unified onto shared financial systems, was essential before AI could be effectively deployed
Three-wave transformation approach - The Ministry of Finance moved in three sequential waves—digitise and integrate, re-engineer services around the customer, then automate and infuse AI—with the order of these steps being critical to success
Speaker 1, Fares, and BinHussain all share a vision of agentic AI as a transformative force in government operations, aligned with the UAE Government 4.0 directive. Speaker 1 announced the UAE's ambition to become the first government to run half of its services on agentic AI within two years , describing this as moving beyond digital toward a government that helps carry out decisions . Fares articulated how agentic AI transforms the experience into proactive, intelligent, and action-oriented support that understands context, checks requirements, and supports action . BinHussain described the Ministry's progression toward agentic AI that actually acts, building on a foundation of robotics automation and generative AI . All three present agentic AI not as a future aspiration but as an active, structured programme of implementation. Both Fares and AlAfad share a concern for citizens who are left behind or poorly served by digital government systems, and both argue that the design of services must account for the full range of human needs. Fares highlighted that customers do not see systems, rules, or workflows — they see only the government — and that a confusing experience becomes a trust issue . AlAfad extended this concern to the 5% of citizens who cannot use digital services at all, including elderly people, those with language barriers, and those with cognitive challenges . Both argue that governments must move beyond measuring success by adoption rates alone and instead design services that are genuinely human and inclusive . Fares, BinHussain, and AlAfad all share the view that the ultimate purpose of digital and AI transformation is to improve the lived experience of citizens, with each speaker framing this from a different angle. Fares stated that the goal is not to say 'we used AI' but for customers to say 'the service is easier than I expected' , and that when governments become easier, people gain confidence in their government . BinHussain demonstrated this through concrete metrics showing citizens served up to nine times faster and 15,000 working hours returned to employees for strategic purposes . AlAfad articulated this as a shift toward a 'human government' built on four principles — accessible, human-centred, trusted, and flexible by default — designed with people rather than for them . Both AlMannaee and AlAfad share a commitment to ensuring that digital government serves all segments of society, particularly those who may be marginalised or underserved. AlMannaee described a national policy for empowering women with five priorities including integrating women into future sectors and building digital accountability , and highlighted digital platforms such as the UAE Women Portal and Medjari to empower productive families . AlAfad described the UAE's digital accessibility policy requiring services to be designed with users and audited for inclusivity , and outlined a plan to re-engineer services for people of determination and elderly people . Both speakers reflect a shared institutional commitment to inclusive digital design.
In a session hosted by a government proudly showcasing its AI achievements, it is somewhat unexpected that multiple speakers openly acknowledged the significant risks and limitations of AI implementation. Fares stated bluntly that 'the demo is easy, the implementation is hard' , cited McKinsey data showing only 39% of organisations could prove real ROI , and referenced Gartner's prediction that over 40% of agentic AI projects will be cancelled by 2027 . BinHussain implicitly acknowledged the difficulty by emphasising the critical importance of sequencing - digitising and integrating before deploying AI - and the need for formal governance structures . AlAfad challenged the audience to consider the 5% of citizens being left behind even as 95% adoption is celebrated , and questioned whether governments have ever truly measured inclusivity . This candid acknowledgement of implementation challenges across all three practitioner speakers represents an unexpected degree of honest self-reflection within what might otherwise have been a purely promotional showcase.
Despite coming from different government entities with different mandates, both Fares and BinHussain independently converged on the view that AI transformation is fundamentally a service re-engineering exercise rather than a technology project. This is somewhat unexpected given the session's focus on AI and digital innovation. Fares explicitly stated that her project is 'not a technology project - it's a service reengineering product, a data readiness project, an integration and governance project' . BinHussain similarly placed service re-engineering as the second wave of transformation - before AI was introduced - through Services 2.0 and zero-bureaucracy initiatives that cut redundant steps and co-designed services with actual users . Both speakers thus independently arrived at the same conclusion: that AI cannot substitute for the harder work of redesigning services around citizens.
Both Fares and AlAfad, speaking from different institutional perspectives, converged on the idea that governments must design services that are not tied to any specific interface or channel, because future channels cannot yet be imagined. Fares argued that the intelligence of a service should not be locked within any single app, page, or platform, noting that today customers access services through apps, kiosks, or voice, but tomorrow there may be channels not yet imagined . Her 'headless' architecture is a direct response to this uncertainty. AlAfad's principle of 'flexible by default' - meaning services can be designed, fixed quickly, and audited - reflects the same underlying concern. This convergence between a technical architect and a human-centred design advocate on the need for channel-agnostic, future-proof service design is an unexpected area of alignment.
The session demonstrated a remarkably high level of consensus across all speakers on the fundamental principles underpinning the UAE's approach to digital and AI-powered government. All speakers agreed that technology is a means to serve people rather than an end in itself ; that trust is a non-negotiable foundation for AI deployment ; that responsible governance and accountability must accompany innovation ; and that the ultimate measure of success is the quality of the citizen experience rather than the sophistication of the technology . Practitioners Fares and BinHussain additionally converged on the necessity of solid data foundations and service re-engineering before AI deployment . AlMannaee and AlAfad shared a commitment to inclusive design for all segments of society . Notably, all three practitioner speakers were candid about the challenges and risks of AI implementation , which added credibility to the overall narrative.
Speaker 1 sets an ambitious top-down directive for the UAE to become the first government to run half its services on agentic AI within two years , and BinHussain describes a cabinet directive to run half of services and processes on agentic AI , projecting confidence in rapid deployment. Fares, however, introduces a note of caution, citing Gartner's prediction that over 40% of agentic AI projects will be cancelled by 2027 due to escalating costs, unclear business value, or weak risk controls , and advocates for a traffic-light framework where progress is paused or halted when risk or trust is questionable . She explicitly warns that 'the demo is easy, the implementation is hard' and that innovation without control becomes noise . This creates a tension between the leadership-driven urgency expressed by Speaker 1 and BinHussain and the more measured, risk-aware approach advocated by Fares.
UAE's agentic AI ambition - The UAE aims to become the first government to run half of its services and operations on agentic AI within two years, under a directive called UAE Government 4.0
Three-wave transformation approach - The Ministry of Finance moved in three sequential waves—digitise and integrate, re-engineer services around the customer, then automate and infuse AI—with the order of these steps being critical to success
Risk of project failure - Gartner predicts that over 40% of agentic AI projects will be cancelled by 2027 due to escalating costs, unclear business value, or weak risk controls, highlighting the need to build better rather than stop
Balancing innovation and control - Innovation must be balanced with control using a traffic-light approach: green when the path is clear, yellow when assessment is needed, and red when risk, policy, or trust is questionable
BinHussain presents an extensive array of quantitative metrics as evidence of successful transformation, including 52,000 automated transactions, 87% reduction in processing time, 99.8% reconciliation accuracy, 97.6% first-call resolution, and 96.7% customer satisfaction . These figures are framed as demonstrating the value and success of the ministry's AI programme. AlAfad directly challenges this metrics-centric view of success, arguing that for years governments have measured success by adoption and satisfaction without caring about how inclusive or human services truly are . She highlights that 5% of the population - elderly people, those with language barriers, and those with cognitive challenges - are being left behind , and that these groups are invisible in conventional success metrics. This represents a substantive disagreement about what constitutes genuine government success in digital transformation.
Measurable outcomes of AI adoption - Concrete results demonstrate significant impact: over 52,000 manual transactions automated, processing time reduced by 87%, reconciliation accuracy at 99.8%, and 50,000 customers served up to nine times faster
Rethinking success metrics - Governments have historically measured success by adoption and satisfaction rates without considering how inclusive or human their services truly are, which must change as technology advances
Speaker 1, Fares, and BinHussain collectively frame the UAE's digital government journey primarily around efficiency, automation, and agentic AI capability - with Speaker 1 describing the goal as becoming the first government to run half its services on agentic AI , Fares demonstrating how AI agents complete complex multi-agency processes on behalf of citizens , and BinHussain showcasing dramatic efficiency gains . AlAfad, by contrast, reorients the conversation entirely, questioning whether the 95% adoption rate conceals a 5% of citizens who are systematically excluded , and arguing that the entire trajectory from digital government to agentic AI government must culminate in a 'human government' that is accessible by default and designed with people rather than for them . She explicitly notes that 'all of us today and yesterday and the day before talking about agentic AI, AI, digitisation, but no one knows to what extent we come more mature and more technological, there is some people who cannot use even me tomorrow' , directly challenging the dominant framing of the session.
UAE's agentic AI ambition - The UAE aims to become the first government to run half of its services and operations on agentic AI within two years, under a directive called UAE Government 4.0
Proactive service delivery - The next shift in digital governance is moving from services that citizens access to services that proactively go to citizens, understand their needs, and guide them through completion
Measurable outcomes of AI adoption - Concrete results demonstrate significant impact: over 52,000 manual transactions automated, processing time reduced by 87%, reconciliation accuracy at 99.8%, and 50,000 customers served up to nine times faster
The excluded 5% - Whilst 95% of digital services are used by the general population, the remaining 5% represents elderly people, those with language barriers, and those with cognitive or accessibility challenges who are being left behind
Four principles of human government - The future of government is a 'human government' built on four principles: accessible by default, human-centred by default, trusted by default, and flexible by default, designing with people rather than for them
Fares places strong emphasis on the dangers of AI that sounds right but is wrong, warning that 'a confident wrong answer is more dangerous than a simple I don't know' , and stressing that government AI must be traceable, explainable, and aligned with policy . She notes this is where many projects struggle . BinHussain, while presenting a governance structure , focuses predominantly on the impressive performance metrics of his ministry's AI systems - 99.8% reconciliation accuracy, 97.6% first-call resolution - without engaging with the risk of AI errors or the need for explainability. The difference in emphasis reflects a divergence between a cautionary, accountability-first perspective and a results-oriented, performance-first perspective on AI deployment in government.
AI accuracy and accountability - In government, AI cannot merely sound right; it must be right, traceable, explainable, and aligned with policy, because a confident wrong answer is more dangerous than admitting uncertainty
Measurable outcomes of AI adoption - Concrete results demonstrate significant impact: over 52,000 manual transactions automated, processing time reduced by 87%, reconciliation accuracy at 99.8%, and 50,000 customers served up to nine times faster
Given that this is a UAE government showcase session with all speakers representing UAE government entities, it is unexpected that AlAfad would so directly challenge the framing of the entire session. While the other speakers celebrate the UAE's rapid advancement toward agentic AI , AlAfad opens by asking the audience to consider the 5% of citizens who cannot use digital services and explicitly states that 'all of us today and yesterday and the day before talking about agentic AI, AI, digitisation, but no one knows to what extent we come more mature and more technological, there is some people who cannot use even me tomorrow' . In a session designed to showcase UAE achievements, this constitutes an unexpected internal critique, suggesting that the very progress being celebrated may be leaving vulnerable populations further behind. This is particularly striking given the collaborative and celebratory tone set by Speaker 1 and reinforced by BinHussain's extensive metrics .
It is unexpected that Fares, presenting on behalf of a UAE government entity in a session explicitly designed to showcase UAE's AI achievements, would prominently cite external industry reports warning of widespread AI project failure. Speaker 1 opens the session with pride in the UAE's ambition to be the world's first government to run half its services on agentic AI , framing the session as a success story to share with the world . Yet Fares cites a 2025 McKinsey survey showing only 39% of organisations can prove real AI ROI and a Gartner prediction that over 40% of agentic AI projects will be cancelled by 2027 . While Fares frames these as calls to 'build better' rather than stop , the introduction of these cautionary statistics in a showcase session represents an unexpected note of candour that implicitly questions whether the UAE's own ambitious targets may face similar challenges.
Fares makes an unexpected distinction within the AI showcase context by explicitly warning against AI that merely 'sounds right' and emphasising that 'the value is not in the conversation, it's in the outcome' . She cautions that 'a confident wrong answer is more dangerous than a simple I don't know' and that AI must be traceable and explainable . This is unexpected because BinHussain's presentation, while also outcome-focused, highlights the impressive performance of AI systems including a call centre sentiment analysis solution and automation robots running at 96.3% success rate - metrics that implicitly celebrate the conversational and transactional capabilities of AI. Fares's warning that many projects struggle precisely because they prioritise the appearance of intelligence over genuine accountability creates an unexpected tension with BinHussain's confidence in his ministry's AI performance figures.
The session is broadly characterised by alignment on the UAE's strategic direction toward agentic AI and Government 4.0, with all speakers representing UAE government entities and sharing a common institutional context. However, meaningful disagreements emerge across three main axes: (1) the appropriate pace and risk tolerance for AI deployment, with Fares advocating caution against the ambitious timelines set by Speaker 1 and BinHussain ; (2) the definition of success in digital government, with AlAfad challenging the adoption-and-satisfaction metrics favoured by BinHussain in favour of inclusivity and human-centredness ; and (3) the primary focus of transformation, with the majority of speakers emphasising efficiency and automation while AlAfad redirects attention to the 5% of citizens being left behind . These disagreements are largely differences of emphasis and priority rather than fundamental opposition, but they reflect genuine tensions in the UAE's digital transformation agenda between speed and safety, scale and inclusion, and technological ambition and human accountability.
All speakers agree that the ultimate goal of digital government transformation is to serve people better and that technology is a means rather than an end. Speaker 1 states the aim has always been to put government services within easy reach of every citizen, resident, and visitor , and that digital transformation was the starting point rather than the goal . Fares argues the goal is that customers will say 'the service is easier than I expected' and that the future of digital government is 'more of a human government experience' . BinHussain frames the entire journey as being about accountability and making financial services faster and more trusted . AlAfad articulates the vision of a 'human government' designed with people for people . Bushlaibi synthesises this in his closing remarks, stating the UAE's journey is about reimagining government around people . However, they disagree significantly on how to achieve this — whether through rapid agentic AI deployment, sequential foundational waves, headless service architecture, or inclusive accessibility-first design.
Digital transformation as a starting point - Digital transformation was never the end goal for the UAE; it was the starting point toward a government that actively helps carry out decisions, not just inform them Proactive service delivery - The next shift in digital governance is moving from services that citizens access to services that proactively go to citizens, understand their needs, and guide them through completion Three-wave transformation approach - The Ministry of Finance moved in three sequential waves—digitise and integrate, re-engineer services around the customer, then automate and infuse AI—with the order of these steps being critical to success Four principles of human government - The future of government is a 'human government' built on four principles: accessible by default, human-centred by default, trusted by default, and flexible by default, designing with people rather than for them Reimagining government around people - The UAE's journey is about reimagining government around people, making services smarter, more inclusive, and more human, with trust and resilience as the foundation
Both Fares and BinHussain agree that robust foundational infrastructure must precede AI deployment, but they emphasise different aspects of what that foundation entails. BinHussain explicitly states 'you cannot put AI on scattered data' and describes the construction of a sovereign data platform and the unification of 52 federal entities onto shared systems as prerequisites. Fares similarly argues that the real work is in the 'service brain' — the rules, data, processes, systems, policies, and controls behind the interface — and frames the project as 'a service reengineering product, a data readiness project, an integration, a governance project' . However, Fares focuses more on the service logic and governance architecture, while BinHussain emphasises the data infrastructure and systems integration layer, reflecting different but complementary conceptions of what 'foundation' means.
Data and systems integration as a foundation - Foundational infrastructure, including a sovereign data platform, a central data lake, and 52 federal entities unified onto shared financial systems, was essential before AI could be effectively deployed Headless service architecture - The 'headless' government service model separates the service intelligence (the 'service brain') from the interface, allowing the same underlying service to be delivered across any channel, present or future Proving AI's ROI - Only 39% of organisations were able to prove real ROI on AI projects according to a 2025 McKinsey survey, demonstrating that the challenge is not adopting AI but turning it into measurable value
Both Fares and BinHussain agree that AI governance and control are essential, but they approach governance differently. BinHussain describes a formal institutional governance structure — a Chief AI Officer, an AI Centre of Excellence, AI champions in every department, and adherence to ISO 27001 and the emerging ISO 42001 — framing governance as an organisational and compliance matter. Fares, by contrast, frames governance as an architectural and philosophical principle embedded in the service design itself, using the traffic-light metaphor and arguing that trust is 'part of our architecture' , likening it to wearing a seatbelt . Both agree governance is necessary but disagree on whether it is primarily an institutional structure or a design principle.
AI governance structure - AI governance requires dedicated leadership and structure, including a Chief AI Officer, an AI Centre of Excellence, AI champions in every department, and adherence to international standards such as ISO 27001 Balancing innovation and control - Innovation must be balanced with control using a traffic-light approach: green when the path is clear, yellow when assessment is needed, and red when risk, policy, or trust is questionable AI accuracy and accountability - In government, AI cannot merely sound right; it must be right, traceable, explainable, and aligned with policy, because a confident wrong answer is more dangerous than admitting uncertainty
Both Fares and AlAfad agree that the citizen's experience of government services is fundamentally about trust and human experience, not merely technical functionality. Fares argues that when customer-facing AI is confusing, it becomes a trust issue rather than a technical issue , and that the future of digital government is 'more of a human government experience' . AlAfad similarly argues for a 'human government' built on principles of accessibility, human-centredness, trust, and flexibility by default . However, they differ in their focus: Fares concentrates on the majority of citizens who interact with agentic AI services and the need for those services to be accurate and trustworthy , while AlAfad focuses specifically on the excluded minority — the 5% who cannot access digital services at all — arguing that trust and human-centredness must extend to those currently left behind.
Trust implications of customer-facing AI - Customer-facing AI is fundamentally different from internal AI because customers do not understand government systems or workflows; a confusing experience becomes a trust issue, not merely a technical one Four principles of human government - The future of government is a 'human government' built on four principles: accessible by default, human-centred by default, trusted by default, and flexible by default, designing with people rather than for them
- The UAE has set an ambitious target to become the first government in the world to run half of its services and operations on agentic AI within two years, under the UAE Government 4.0 initiative, with digital transformation viewed as a starting point rather than an end goal.
- Agentic AI in government represents a fundamental shift from citizens accessing services to services proactively reaching citizens, understanding their needs, guiding them through processes, and completing actions on their behalf, with value measured by outcomes rather than by the sophistication of the technology itself.
- The 'headless' government service model, as demonstrated by Digital Ajman, separates the service intelligence (the 'service brain' comprising rules, data, processes, and controls) from the user interface, enabling consistent service delivery across any channel, present or future.
- Customer-facing AI carries significantly higher stakes than internal AI because citizens do not understand government systems or workflows; a confusing or incorrect AI response becomes a trust issue rather than merely a technical problem, meaning government AI must be accurate, traceable, explainable, and policy-aligned.
- Proving return on investment remains a critical challenge for AI adoption: only 39% of organisations could demonstrate real ROI according to a 2025 McKinsey survey, and Gartner predicts over 40% of agentic AI projects will be cancelled by 2027 due to escalating costs, unclear business value, or weak risk controls.
- The Ministry of Finance's three-wave approach — first digitise and integrate, then re-engineer services around the customer, then automate and infuse AI — demonstrates that sequencing is critical; AI cannot be effectively deployed on scattered or unintegrated data.
- Foundational infrastructure is a prerequisite for AI success: the Ministry of Finance unified 52 federal entities onto shared financial systems and built a sovereign data platform before deploying AI, achieving measurable results including 87% reduction in processing time, 99.8% reconciliation accuracy, and 50,000 customers served up to nine times faster.
- Robust AI governance structures are essential, including dedicated leadership such as a Chief AI Officer, an AI Centre of Excellence, AI champions embedded in every business department, and adherence to international standards such as ISO 27001 and the emerging ISO 42001 for AI management.
- The UAE's General Women's Union is advancing women's empowerment through a Cabinet-approved national strategy, a national policy with five priorities, over 155 cross-sector partnerships, and digital platforms including the UAE Women Portal, a women in sport platform, and VR tools for heritage learning.
- Inclusive and human-centred design is a critical gap in digital government: the 5% of citizens who cannot use digital services — including elderly people, those with language barriers, and those with cognitive or accessibility challenges — are being systematically excluded, and success metrics must move beyond adoption and satisfaction rates to encompass inclusivity.
- The UAE has enforced a federal digital accessibility policy requiring services to be co-designed with users, tested for inclusivity, and audited by TDRA, underpinned by four principles of human government: accessible by default, human-centred by default, trusted by default, and flexible by default.
- Innovation and control must be balanced using a traffic-light governance approach — green when the path is clear, yellow when assessment is needed, and red when risk, policy, or trust is in question — enabling governments to move with speed and direction without proceeding blindly.
“For years, digital governance was mainly about moving services from counters to websites, from websites to apps, and now from apps to unified platforms. But now the question is changing. It is no longer 'can the customer access the service with minimum effort?' It is 'can the service go first to the customer, understand his needs, guide and advise them, and help them to complete what they want or what they need to do?'”
“The demo is easy, the implementation is hard. You can see a demo that will impress you in a matter of clicks, but the real government service has to work with real people, real data, rules and exceptions, integrations, real cost, and real accountability. Only 39% was able to prove real ROI on their AI projects.”
“A confident wrong answer is more dangerous than a simple 'I don't know.' In government, AI cannot just sound right, it must be right. It must be traceable, explainable, and aligned with policy.”
“Innovation without control becomes noise. And control without innovation becomes a delay. The real balance is not to stop innovation, but give it signals. Green when the path is clear. Yellow when we need to stop and assess. Red when risk or policy or trust is questionable.”
“95% of digital services are used. What about the 5%? These are elderly people who cannot navigate websites, people with language barriers, people with cognitive issues with web design. For years, governments measured the success of a service by adoption and satisfaction. But no one cared about how inclusive it is and how human it is.”
“We moved in three waves and that order mattered a lot. First, digitise and integrate. Second, re-engineer services around the customer. Third, automate and infuse AI. You cannot put AI on scattered data.”
“For the first time, we can hear every customer and not just a sample.”
How can governments prove and measure the real return on investment (ROI) of agentic AI projects?
Dina Fares cited a 2025 McKinsey AI Global Survey finding that only 39% of organisations could prove real ROI on their AI projects. This highlights a critical gap between AI adoption and demonstrated value, making it an important area for further research, particularly for governments seeking to justify investment in agentic AI.
What strategies can governments employ to prevent agentic AI projects from being cancelled due to escalating costs, unclear business value, or weak risk controls?
Dina Fares referenced a Gartner prediction that over 40% of agentic AI projects will be cancelled by the end of 2027. Understanding how to mitigate these risks is essential for governments planning large-scale AI deployments and warrants further investigation into governance frameworks and cost management strategies.
How can governments design and implement a 'service brain' architecture that is channel-agnostic and future-proof for interfaces that have not yet been imagined?
Dina Fares raised the concept of a headless, channel-agnostic service brain that can support future interfaces beyond apps, kiosks, and voice. As technology evolves rapidly, further research is needed into how governments can build backend service architectures that remain robust and adaptable to unknown future channels.
How can governments ensure that customer-facing AI is not only confident but also accurate, traceable, and explainable, given that a confident wrong answer is more dangerous than admitting uncertainty?
Dina Fares emphasised that in government contexts, AI must be right, traceable, and aligned with policy — not merely sound convincing. This raises important questions about AI explainability and accountability frameworks that require further research, especially where public trust is at stake.
How can governments strike the right balance between innovation speed and risk control in agentic AI deployments?
Dina Fares introduced a traffic-light model (green, yellow, red) for managing AI risk, noting that innovation without control becomes noise and control without innovation causes delay. Further research is needed into practical governance models that allow governments to move quickly while maintaining appropriate safeguards.
What are the data readiness and integration prerequisites that governments must address before successfully deploying agentic AI in public services?
Meshal BinHussain stressed that AI cannot be placed on scattered data and that the Ministry of Finance first had to unify 52 federal entities onto shared financial systems. This points to a need for further research into data infrastructure requirements and integration strategies as foundational steps for AI adoption in government.
How should governments structure AI governance, including roles such as Chief AI Officer, AI centres of excellence, and AI champions, to ensure effective and accountable AI deployment?
Meshal BinHussain described the Ministry of Finance's governance structure for AI, including a Chief AI Officer and AI champions in every department. Further research into optimal organisational structures for AI governance in government could help other nations replicate or adapt these models effectively.
How can governments address the digital inclusion gap — specifically the 5% of citizens who cannot access or use digital services — as AI and digital transformation advance?
Manal AlAfad highlighted that while 95% of digital services are used, the remaining 5% of citizens — including elderly people, those with language barriers, and those with cognitive or accessibility challenges — are being left behind. This is a critical area for further research into inclusive design, digital accessibility policy, and human-centred government service delivery.
What does a 'human government' look like in practice, and how can governments design services with people rather than for people as AI capabilities mature?
Manal AlAfad introduced the concept of a 'human government' as the next evolution beyond agentic AI government, emphasising design with users rather than for them. Further research is needed to define, operationalise, and measure what a truly human-centred government looks like, particularly as technology risks outpacing the needs of vulnerable populations.
How can digital accessibility policies be effectively implemented, tested, measured, and audited across federal government entities to ensure inclusive service design?
Manal AlAfad described the UAE's digital accessibility policy, which requires services to be designed with users, tested, and audited by TDRA. Further research into best practices for accessibility auditing, compliance measurement, and iterative service redesign would benefit governments seeking to replicate this approach.
How can international collaboration and knowledge-sharing on women's empowerment initiatives be facilitated through official channels between countries?
Ghalya AlMannaee invited countries interested in benefiting from the UAE's international women's empowerment initiatives to engage through official channels. This implies a need for further exploration of frameworks and mechanisms for international partnerships in gender policy and digital empowerment programmes.
How can governments achieve the transition from legacy financial systems to AI-embedded, data-driven operations in a structured, phased manner that delivers measurable results?
Meshal BinHussain outlined a three-wave approach (digitise and integrate, re-engineer services, then automate and infuse AI) that produced significant efficiency gains. Further research into replicable phased transformation models for government financial systems would be valuable for other nations at earlier stages of digital maturity.