On February 17, 2026, Infosys announced a strategic collaboration with Anthropic to build agentic AI systems for regulated industries, combining Infosys's Topaz AI platform with Anthropic's Claude models — including Claude Code — and the Claude Agent SDK, the toolkit developers use to build autonomous AI agents that can carry out multi-step tasks rather than just answer questions.

The rollout starts in telecommunications, through a dedicated "Anthropic Center of Excellence" that Infosys is setting up, before expanding into financial services, manufacturing, and software development. The target use cases are specific rather than exploratory: network operations, customer lifecycle management, claims and compliance workflows, and modernizing legacy systems — areas where regulated industries have been historically cautious about handing processes to AI without clear audit trails.

Not a pilot project

Separately, at its Q1 FY26 earnings briefing, Infosys's CEO said the company had already deployed more than 300 agentic AI solutions across client operations and internal platforms, describing the work as "well beyond pilot phase." That scale matters for how to read this partnership: it's not a proof-of-concept announcement from a company dabbling in AI, but an expansion of infrastructure a major Indian IT services firm is already running at production scale.

What this means for IT job skills

For a company the size of Infosys, betting its platform strategy on agent orchestration rather than purely traditional application development is a signal about where internal hiring and reskilling priorities are headed. Skills likely to see rising demand inside IT services firms making similar bets include:

• Agentic workflow design — breaking a business process into steps an AI agent can reliably execute and hand off, rather than writing code for each step manually.
• Agent evaluation and guardrails — testing whether an agent's output is reliable enough for a regulated workflow, and building the checks that catch it when it isn't.
• Working with agent SDKs and frameworks directly, rather than only consuming AI through a chat interface.
• Domain expertise in the target industries (telecom network operations, claims processing, compliance) paired with the ability to translate that expertise into agent specifications.

A note of caution worth keeping in mind

Enterprise "agentic AI" announcements have moved fast over the past year, and independent surveys of the broader market have repeatedly found a gap between how companies describe their AI agent deployments and how autonomously those systems actually operate in practice. Infosys's claim of 300-plus deployments and a dedicated center of excellence with Anthropic is a concrete, verifiable commitment rather than a vague roadmap — but as with any large-scale AI rollout, how much of the described capability is running unsupervised in production, versus still requiring significant human review, is worth watching as results come in over the next few quarters.

(Infosys Newsroom)