Narada AI, a startup emerging from two years in stealth mode, launched its AI assistant at TechCrunch Disrupt 2024. Designed for enterprise use, Narada’s assistant goes beyond typical AI chatbots by directly performing tasks across several workplace applications. The technology is built on the concept of ‘LLM Compilers,’ developed in collaboration with UC Berkeley researchers, allowing it to interact with various apps, even those lacking APIs, by navigating their front-end interfaces.
The assistant can draft emails, create calendar invites, organise meeting notes, and even pull information from tools like SAP and Salesforce, all from a chat interface. Narada’s CEO Dave Park believes the assistant’s ability to work across applications without APIs – through an in-house method called Web Redemption – is a key innovation. The feature allows the assistant to map out and navigate applications as a user would, letting it adapt even if layouts or settings change over time.
Though promising, the assistant’s functionality raises questions around data privacy, as it requires full access to users’ emails, contacts, and calendars. Narada assures users that their data won’t be used for model training. With one Fortune 500 client already testing the tool, Narada hopes its customisation and enterprise focus will differentiate it in the competitive AI assistant market.
Google has delayed the release of its next-generation AI agents, part of a project called Astra, until 2025 at the earliest. CEO Sundar Pichai outlined the timeline during the company’s Q3 earnings call, indicating that significant AI advancements are still under development.
Project Astra, first demonstrated at Google’s I/O conference in May 2024, aims to integrate AI with real-world understanding. Applications include smartphone apps capable of recognising objects through the camera and answering questions based on the environment. The project also envisions advanced AI assistants capable of carrying out tasks such as purchasing items or booking flights on behalf of users.
Reports earlier in October suggested that Google had planned to release a consumer version of an AI agent by December 2024. However, this release now seems unlikely unless the agent operates separately from Astra’s technologies. The decision reflects the challenges involved in developing reliable AI capable of complex interactions and real-time reasoning.
Companies like Anthropic have launched similar generative AI models with some success, but these models have also encountered difficulties in completing basic tasks. Google’s cautious approach may reflect a broader need to ensure functionality before releasing the technology to the public.
LinkedIn has introduced its first AI agent, Hiring Assistant, designed to automate many of the time-intensive tasks recruiters face, such as drafting job descriptions, identifying candidate matches, and handling initial outreach. Initially available to a select group of large enterprises, including AMD, Siemens, and Zurich Insurance, Hiring Assistant is expected to expand to more users in the coming months. By automating repetitive tasks, LinkedIn aims to free up recruiters to focus on higher-impact aspects of their jobs.
Built using LinkedIn’s data from over 1 billion users and backed by Microsoft’s OpenAI partnership, Hiring Assistant can refine job requirements based on existing listings, generate candidate pools, and filter applicants by skills rather than traditional markers like location or education. This AI assistant is part of LinkedIn’s broader push to integrate AI into its platform, following similar tools for resume and profile optimisation, career coaching, and job search support.
In its current iteration, Hiring Assistant is already making strides in streamlining recruiting, with plans for future updates to handle interview scheduling, candidate follow-ups, and more. LinkedIn, which has seen AI-driven growth in its premium subscription base, views Hiring Assistant as a key product in its business offerings for recruitment professionals, aiming to enhance LinkedIn’s impact in the hiring sector.
Nvidia-backed biotech firm Iambic Therapeutics has introduced Enchant, an AI model that aims to reduce the time and cost of drug development. Enchant, trained on extensive pre-clinical data, is designed to predict a drug’s early performance with impressive accuracy. In Iambic’s studies, Enchant achieved a 0.74 accuracy score in predicting drug absorption in the human body, compared to previous models which peaked at 0.58. This predictive power could help pharmaceutical companies identify promising drugs sooner, significantly cutting down on failed late-stage trials.
According to Iambic’s co-founder Fred Manby, Enchant could potentially slash development costs by half, as researchers could more accurately assess a drug’s success at the earliest stages. Nobel laureate and Iambic board member Frances Arnold also highlighted Enchant’s unique capabilities, noting that unlike models like Google DeepMind’s AlphaFold, which focus on molecular structure, Enchant evaluates pharmacokinetic and toxicity properties crucial to drug success.
With Enchant, Iambic is poised to set a new standard in the pharmaceutical industry by addressing some of the biggest hurdles in drug development, including high costs and late-stage failures. The AI technology’s rollout could mark a major shift, making drug discovery both faster and more efficient for a variety of treatments.
CTGT, a startup founded by Cyril Gorlla and Trevor Tuttle, aims to improve the safety and transparency of AI models. Operating in a field known as ‘explainable AI,’ CTGT’s platform identifies biased outputs and hallucinations in AI models, with a particular focus on applications in healthcare, finance, and other high-stakes industries. Rather than training additional models to oversee the AI, CTGT employs mathematically-guaranteed interpretability techniques, allowing companies to identify errors more efficiently and accurately.
CEO Gorlla highlighted the dangers of relying on inaccurate or biased AI decisions, emphasising that models are increasingly deployed in critical areas where errors can have serious consequences. CTGT’s clients include three unnamed Fortune 10 companies, one of which used the platform to correct biases in a facial recognition system. By offering both managed and on-premises solutions, CTGT also addresses data privacy concerns, giving companies control over their information without compromising security.
CTGT has gained support from major investors, including Mark Cuban and the co-founder of Zapier, and is a graduate of the Character Labs accelerator. As the startup expands, it plans to build out its engineering team and enhance its platform to meet the rising demand for AI interpretability. Analytics firm Markets and Markets estimates that the explainable AI sector could reach $16.2 billion by 2028, a promising outlook for companies focused on AI safety and transparency.
Kenya partners with Google to enhance its digital infrastructure and empower its citizens in the evolving digital economy. The collaboration aims to create a robust digital ecosystem that meets current technological needs while anticipating future demands.
Kenya seeks to empower decision-makers with real-time insights by utilising AI and data-driven technologies, enhancing operational efficiency and facilitating effective governance. A key focus of the partnership is revitalising the tourism sector through Google’s technology, attracting more international visitors and showcasing the country’s unique landscapes, wildlife, and cultural heritage.
Additionally, prioritising cybersecurity measures is critical to building trust among citizens and ensuring a secure digital environment. The initiative will also promote skills training to equip Kenyans with essential digital competencies, fostering innovation and creativity while contributing to the overall growth of the nation’s economy.
Through this partnership, Kenya addresses immediate technological needs and lays a foundation for sustainable development in the digital space. By enhancing digital literacy and integrating advanced technologies, the collaboration positions Kenya as a leader in the region’s technological landscape.
Why does it matter?
The comprehensive approach ensures that as the digital economy expands, citizens are well-prepared to navigate the challenges and opportunities that arise, ultimately driving growth and resilience in the face of rapid technological advancements.
ForceField is unveiling its new technology at the 2024 TechCrunch Disrupt, introducing tools aimed at fighting deepfakes and manipulated content. Unlike platforms that flag AI-generated media, ForceField authenticates content directly from devices, ensuring the integrity of digital evidence. Using its HashMarq API, the startup verifies the authenticity of data streams by generating a secure digital signature in real time.
The company uses blockchain technology for smart contracts, safeguarding content without relying on cryptocurrencies or web3 solutions. This system authenticates data collected across various platforms, from mobile apps to surveillance cameras. By tracking metadata like time, location, and surrounding signals, ForceField provides insights that aid journalists, law enforcement, and organisations in verifying the accuracy of submitted media.
ForceField was inspired by CEO MC Spano’s personal experience in 2018, when she struggled to submit video evidence following an assault. Her frustration with the justice system sparked the creation of technology that could simplify evidence submission and ensure its acceptance. Now the startup is working with clients such as Erie Insurance and plans to launch commercially by early 2025, focusing initially on the insurance sector but with applications in media and law enforcement.
The company, which is entirely woman-led, has gained financial backing from several angel investors and strategic partnerships. Spano aims to raise a seed round by year’s end, highlighting the importance of diversity in tech leadership. As AI-generated content continues to flood the internet, ForceField’s tools offer a new way to validate authenticity and restore trust in digital information.
San Francisco-based Advex AI has launched publicly at TechCrunch Disrupt 2024, aiming to address data shortages for training AI systems using synthetic imagery. Co-founded by CEO Pedro Pachuca and CTO Qasim Wani, Advex has already secured funding totalling $3.6 million and boasts seven major enterprise clients. Advex’s synthetic data platform uses a proprietary diffusion model to generate thousands of ‘fake’ images from a small sample, helping clients train machine vision systems with limited original data.
Advex’s solution is particularly valuable in sectors like manufacturing, where recognising subtle defects can be crucial but challenging with limited real data. For example, a car manufacturer needing to train a system to detect seat material flaws could upload just a few images of tears, with Advex generating thousands of variations to expand training data. Such applications span industries, from automotive to oil and gas, reducing costs and time associated with real data collection.
While synthetic data isn’t a new concept, Advex distinguishes itself through its custom diffusion model, which Pachuca says is faster and more realistic than traditional simulation methods. Unlike game-engine techniques, Advex’s model can rapidly create images tailored to the data gaps in a client’s specific AI system, helping it operate more effectively in real-world scenarios.
Microsoft is anticipated to report its slowest revenue growth in a year as investors focus on AI-related earnings and the impact of heavy investments in the technology. While Microsoft has led the way in generative AI, helped by its significant stake in ChatGPT creator OpenAI, adoption of its enterprise AI assistant, Copilot, has lagged. Recent reports suggest a hesitant market for Copilot’s $30-per-month subscription, with many companies still in pilot phases.
Analysts from Morgan Stanley and Visible Alpha expect Microsoft’s capital expenditures in the September quarter to have surged nearly 72% year-on-year, driven by high AI and cloud computing costs. Azure, Microsoft’s cloud unit, likely grew by 33% for the quarter, although that marks a slight dip from prior growth. Despite this, Microsoft hopes for stronger AI-driven revenue in Azure and is targeting faster growth in the second half of the fiscal year.
In the wake of a financial reorganisation in August, Microsoft’s earnings have become harder to predict. With high AI-related costs weighing on margins, Microsoft’s shares have seen minimal growth since July, underperforming the S&P 500. Meanwhile, analysts anticipate a revenue rise of around 14% to $64.5 billion, a modest improvement amid investor concerns over Microsoft’s AI strategies.
Scepticism around Microsoft’s 365 Copilot assistant remains, though some analysts believe recent AI upgrades could drive demand. Microsoft’s productivity unit, including LinkedIn and Office, is expected to maintain steady growth, and the company remains optimistic about AI’s potential to strengthen its productivity suite.
Amazon has announced the international expansion of Rufus, its AI-powered shopping assistant, which will now be available in multiple new markets across Europe and the Americas. Originally launched in the US earlier this year, Rufus assists users with product searches, personalised recommendations, and side-by-side comparisons. This expansion aims to make Amazon’s shopping experience more seamless by answering shoppers’ questions in natural language, whether they’re looking for gift ideas or specific product advice.
Rufus has been trained on Amazon’s extensive data library, including product listings, customer reviews, and other public information. By integrating Rufus into Amazon’s Shopping app, the company is competing more directly in the AI space, a move that underscores its efforts to stay competitive with other tech giants. Users in newly added regions can now access Rufus by updating their Amazon app and selecting the chatbot icon, which activates an intuitive, chat-based interface.
While this initial version of Rufus is still in development, Amazon acknowledges that it may not yet be perfect but promises regular updates. The company is also investing in generative AI to enhance services for sellers, like automated listing descriptions. This broader AI strategy includes Amazon’s recent $230M investment in startups to drive further innovations in the field.
