IBM and Microsoft unveil AI agents platform as startups target safety.
IBM and Microsoft launch enterprise agent platform.
On 10 November 2026, IBM and Microsoft announced a collaboration to accelerate enterprise use of AI agents ([1]). Their new platform, Imagine Enterprise, comes with 6,000 pre-built AI agents drawn from IBM’s industry consulting projects ([2]). It allows organisations to integrate these agents into existing systems and processes with controls, monitoring and compliance features baked in. The platform is first rolling out for financial services, insurance and retail, with built-in support for connecting to company data and applications. IBM says the goal is to help businesses get beyond pilots and scale automation securely, without having to start from scratch.
This joint push by two tech giants underlines the momentum behind agent-based automation in business. The ready-made agents are tailored for common tasks from claims routing to supply-chain updates, reflecting how far agent capabilities have expanded in specific, rule-based processes. Early adopters like financial services firm SEI are exploring the platform to speed up internal workflows while keeping human oversight in place ([3]) ([4]). For enterprise leaders, the takeaway is that major vendors are now packaging AI agents in a more enterprise-friendly way: with industry-specific expertise, security frameworks and human “forward-deployed” support from day one. This could reduce the time and risk of introducing autonomous agents into business operations.
OpenAI acquires a travel-planning AI agent.
OpenAI’s move into specialised autonomous services accelerated with its acquisition of Odessia, a year-old AI travel agent startup, on 10 November 2026 ([1]). Odessia’s founder, former Sonder CEO Francis Davidson, announced that he would join OpenAI to work on the “future of travel,” after his tool helped tens of thousands of users plan “millions of dollars’ worth of travel” in its first few months ([2]). The service, which arranged entire trips via teams of AI agents, has been shut down for now, but OpenAI has longstanding ties with major travel sites and noted that travel planning is a popular use for ChatGPT ([3]). The deal looks like an acqui-hire aimed at bringing experienced talent in-house; an OpenAI spokesperson confirmed the hire and highlighted that 1.2 billion people use ChatGPT weekly for tasks like health advice, budgeting and trip planning ([4]).
The Odessia purchase shows how demand for AI agents is emerging even in complex consumer-facing services such as travel. For other businesses, it’s a signal that leading AI providers are moving beyond general-purpose chatbots towards domain-specific agent capabilities. Business leaders in sectors like travel and hospitality should expect more off-the-shelf AI assistants from the big players that can plan, transact and deliver outcomes for customers without step-by-step human input. As these tools develop, companies will need to closely monitor their performance and ensure they are integrated with trusted data and clear policies.
Raising guardrails for autonomous agents.
New investments this week underscore the push to make AI agents safer and more reliable in live use. Tel Aviv-based startup Bay emerged from stealth on 10 November with $50 million in funding led by Battery Ventures ([1]). Bay’s platform acts as a control plane for AI agents in the enterprise, discovering all autonomous assistants running on employee devices or in cloud services and enforcing policies on their actions in real time ([2]). The software monitors each agent’s entire session and can pause or block any action that violates company rules ([3]). Bay already counts large firms like Micron, Peloton and Wix among its customers ([4]), and the company claims that some Fortune 100 clients have tripled their use of AI agents once they were confident those agents wouldn’t “run amok.”
On the same theme, New York-based Plurai announced a $13 million seed funding round on 11 November to launch its simulation platform for AI agents. Plurai’s system creates digital twin environments of enterprise tools, data and policies, then runs thousands of test scenarios to find failures or policy violations before agents are deployed ([5]). The platform also generates fixes and fine-tunes smaller, custom AI models to serve as always-on watchdogs for the agent in production ([6]). In design partner trials, the company says this simulation-driven approach dramatically reduced the cost and errors of running AI agents ([7]). These solutions address a core concern for executives: how to embrace valuable autonomous assistants in sensitive workflows without losing control. They suggest that with the right guardrails and pre-deployment testing, businesses can safely scale up what their AI agents are trusted to do.