Field report NE-22 / Q7 · sheet /deals

M&A & DealsFiled from the territory

McKinsey and Cognition Team Up to Prove AI Coding Pays

McKinsey and Cognition will jointly sell AI coding transformation, pairing Devin with consulting. A survey shows a quarter of firms doubled productivity; nearly 30% saw it fall.

Grid
NE-22 / Q7
Surveyed
November 16, 2026
Reading
3 min read
Correspondent
Sophie Lindqvist
McKinsey and Cognition are trying to prove AI coding pays off
Fig. NE-22 / Q7McKinsey and Cognition are trying to prove AI coding pays off — AI-generated

Field notes

  1. McKinsey and Cognition announced a partnership on Monday to prove AI coding's ROI
  2. A McKinsey survey of 300+ engineering leaders found top-quartile firms more than doubled productivity, while nearly 30% saw it drop
  3. One Cognition client compressed a five-year ERP modernization into 18 months
  4. Goldman Sachs, Cisco and JPMorgan have become more selective about AI tool access

McKinsey & Company and the AI lab Cognition announced a partnership on Monday with one explicit goal: proving that AI coding tools justify their cost. The stakes are real, because the evidence so far cuts both ways. A recent McKinsey survey of more than 300 product and engineering leaders found that the top quarter of organizations more than doubled productivity with AI — while nearly 30 percent said productivity actually fell.

Under the arrangement, McKinsey will advise clients on which engineering projects to target and how to redesign teams and workflows around AI. Cognition supplies the technology: its AI coding agent Devin, which the company says can understand, plan, write, and test code. Neither side has published pricing, deployment timelines, or measured results from the partnership. The promise is efficiency; the proof, so far, is a promise.

What changed to make this deal possible?

Capability, according to Lari Hämäläinen, a senior partner in McKinsey's technology, media, and telecommunications practice. Until the final months of 2025, he said, most AI coding tools handled only narrow tasks such as looking up technical guidance or syntax details. "That was about it," Hämäläinen said. Since then, more capable models have allowed longer, more autonomous coding work to be handed off.

The gap, in his telling, is organizational rather than technical. Tools improve faster than companies can redesign the processes around them. "This is really a step-change phenomenon" that "requires a complete rethink" of software- and product-development processes, Hämäläinen said.

He flagged a concrete pitfall: running a small pilot team on an AI tool is easy, but "it's a completely different thing to scale it across 2,000 developers or 20,000 developers." What that rethink costs, and who bears it, went unsaid.

Why are buyers suddenly picky?

The hunt for AI returns has made large companies more selective about who gets access to which tools. Goldman Sachs, Cisco, and JPMorgan have all tightened how AI tools are distributed and used internally, managing both costs and outcomes.

Cognition's answer to the cost problem is model routing — directing different tasks to different AI models based on their capabilities and price. Devin can draw on multiple models, using pricier frontier tools for harder engineering work and cheaper ones elsewhere. Art Levy, VP of global partnerships at Cognition, put it plainly: "You don't need to drive a Ferrari to the grocery store."

Routing tools are becoming a popular cost-management tactic across the industry, not just at Cognition. That popularity itself signals that first-generation AI budgets ran hot.

What does this change for buyers?

For enterprises sitting on deferred software modernization, potentially the schedule. Levy pointed to one client where an update of enterprise resource planning software (the integrated systems running core business functions), expected to take five years, is being compressed into 18 months with Devin. One client, self-reported, without an independent benchmark — but the compression factor, if replicable, would reshape project economics.

Levy framed the market shift in blunt terms: "It's no longer, 'Can you get the job done?' It's, 'Can you get the job done, and are you cheap at doing it?'"

The unresolved question

Devin's early-2024 launch unsettled software engineers who feared AI would overtake their livelihoods. Since then, developers' work has shifted from writing code to guiding AI output — a real change, though not the displacement many expected.

The partnership's bet rests on a survey showing a quarter of firms doubling productivity and nearly a third losing it. That spread is the story. AI coding pays off for some organizations and subtracts value from others, and the announcement on Monday offered a method for sorting one from the other — but no published evidence that the method works.

via feeds.businessinsider.com (Source)

Share this article:

More from Sophie Lindqvist

Sophie Lindqvist

Show full bio

Correspondent covering money, funding and market shifts around corporate strategy and chief-executive leadership at CEO Frontier.

483 articles

Filed under

Adjacent sheets

« Previous articleNext article »

End of sheetNE-22 / Q7 · November 16, 2026