HARTFORD, Conn., Jan 15, 2026, 11:32 (EST)
The Travelers Companies, Inc. said on Thursday it is working with AI startup Anthropic to equip nearly 10,000 engineers, data scientists, analysts and product owners with personalized Claude assistants, including Claude Code for programming.
The property-casualty insurer said the tools can tap Travelers’ data and systems to speed software and analytics work, and will sit alongside TravAI, an internal platform used by more than 30,000 employees.
Travelers’ chief technology and operations officer Mojgan Lefebvre said early deployments have delivered “meaningful improvements in productivity,” while Anthropic’s Head of Americas Kate Jensen called it “weaving Claude into relevant workflows.” (Business Wire)
The push matters because big companies have spent the past year testing generative AI — software that can produce text or code from a prompt — without always moving beyond pilots.
A deployment measured in thousands forces harder questions about oversight, data access and whether the time savings are real, not just anecdotal.
Travelers’ rollout also reflects a shift in what buyers want from these tools. “Agentic AI” is the industry shorthand for systems that can take actions across business software, not just answer questions, and that’s where a lot of the spending is headed.
Anthropic is also reshuffling its product leadership as it tries to keep pace. On Tuesday, it moved chief product officer Mike Krieger to co-lead an internal incubator called Labs with product engineering lead Ben Mann and handed product leadership to Ami Vora alongside CTO Rahul Patil.
Krieger told The Verge the company had hit a “watershed moment” and he wanted to get back into “builder mode,” while president Daniela Amodei said the pace of AI progress “demands a different approach.” (The Verge)
On Wednesday, Thomson Reuters said it had launched a Trust in AI Alliance that brings together Anthropic, AWS, Google Cloud and OpenAI, with a goal of aligning on principles for responsible “agentic” systems.
Thomson Reuters CTO Joel Hron said building trust in how agents “reason, act, and deliver outcomes is essential.” Anthropic’s Scott White said “Trust in AI systems is essential” as the tools take on more autonomous work in high-stakes settings. (Thomson Reuters)
Anthropic on Thursday also released a new Economic Index report based on anonymized Claude.ai chats and first-party API transcripts from November 2025, tracking what users ask the model to do and how often it succeeds.
Claude usage remains concentrated among a small set of tasks, most of them coding-related, the report said; the 10 most common tasks accounted for 24% of sampled conversations. It also found that Claude’s success rate drops as tasks become more complex, which it said can change estimates of which jobs are most exposed to AI help. (Anthropic)
The Travelers deal targets exactly where the spending is easiest to justify: code, data work, and internal workflows that already have performance metrics. It also tests a practical question most vendors dodge — can AI stay useful when it has to play by enterprise rules.
But scaling assistants inside a company has a downside: bad outputs and downtime scale too.
On Jan. 14, Anthropic’s Claude service suffered an outage, with the company warning users of “elevated error rates” on its Opus 4.5 and Sonnet 4.5 models before saying it had implemented a fix. (IT Pro)
Buyers are still feeling out where human review fits when an AI system writes code, drafts internal notes, or touches files. One weak link — a sloppy prompt, a mis-set permission — can turn a helper into a headache.
Travelers did not disclose financial terms or how it will measure success beyond broad productivity claims.
For Anthropic, landing big rollouts is one thing; keeping performance steady and governance tight, across clients, is the harder part.
