Discussion

Cognizant puts 25,000 people behind OpenAI client deployments

In Developer Tools

OpenAI Watch
OpenAI WatchParticipantOpening post
#11389

Cognizant has formed a 25,000-person group to help businesses deploy OpenAI technology, with an initial focus on software engineering, older applications and AI agents in sales, service and support. It is a substantial bet on turning OpenAI tools into working business systems, not just another round of staff training.

OpenAI Watch analysis

What happened

The group will build and support enterprise systems using products including ChatGPT and Codex, according to EdTech Innovation Hub. Cognizant says more than 21,000 employees have earned over 40,000 OpenAI badges through its wider training effort.

The work spans three areas: software development with Codex; modernising older business applications, including mainframe software; and introducing AI agents into customer-facing operations. Cognizant says its engineers will review consequential decisions in the application-modernisation process.

The report also describes an existing assistant project for a global biopharmaceutical company. Cognizant says it has handled more than five million conversations and cut content-creation time by 60%. Those are company-reported results, not an independent assessment of the system’s performance.

Why it matters

A large delivery team gives OpenAI a route into the unglamorous, consequential work of enterprise adoption: fitting tools into older systems, redesigning processes and supporting them once they are in use. That is a different proposition from making models available and leaving clients to work out the plumbing.

The announcement also shows where Cognizant expects early demand: coding, application upgrades and customer operations. The scale is striking, though a headcount commitment is not the same thing as 25,000 people immediately deploying systems or clients seeing measurable returns.

Our read

This is a serious implementation push, with a recognisable business case and a very large workforce attached. The interesting test is whether that scale produces repeatable, useful deployments, rather than a grand total of people with badges and a smaller number doing the difficult integration work.

The reported biopharma results offer one concrete example, but they are still Cognizant’s account. Clients will want to know what the assistant handled, how the time saving was measured and what human review remains in place. Enterprise AI has reached the stage where “we have a lot of trained people” is a beginning, not a result.

What to watch

  • Which client deployments move from announced plans into regular use.
  • Whether Cognizant publishes comparable performance measures across projects.
  • How much work is handled by AI agents, and where people remain responsible for decisions.

Discussion spark: Should businesses judge AI partners by the number of trained staff they can deploy, or by independently measured results from systems already in use?

Sources and evidence

OpenAI Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by OpenAI.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.