McKinsey, Cognition launch AI coding partnership for enterprise software projects
Companies are reworking software teams as AI coding tools take on larger and more autonomous engineering tasks. McKinsey & Company and Cognition say their new partnership is aimed at helping clients target projects, redesign workflows, and prove that the technology delivers returns.
Highlights
- McKinsey and Cognition announced a partnership combining McKinsey's enterprise advisory services with Cognition's AI coding agent Devin to support scalable software development.
- McKinsey's survey of over 300 leaders shows AI doubled productivity for the top quartile but cut productivity for nearly 30% of organizations, making ROI at scale uncertain.
- Clients using Cognition's Devin have accelerated major software upgrades—for example, condensing a five-year ERP project into 18 months—by leveraging tiered AI models for cost efficiency.
Enterprise rollout strategy and tool economics
As reported by Business Insider, the partnership announced Monday pairs McKinsey's advisory work on engineering priorities and operating models with Cognition's AI coding agent Devin, which the company says can understand, plan, write, and test code.The tie-up comes as developers' work across industries shifts from writing code toward supervising AI output. Lari Hämäläinen, a senior partner in McKinsey's technology, media, and telecommunications practice, says many AI coding tools were still mainly useful for narrow tasks such as looking up technical guidance or syntax details until the final months of 2025, but more capable models now make it possible to hand off longer and more autonomous coding work.
Hämäläinen says that pace of improvement can outstrip a company's ability to redesign processes around the technology. He describes the shift as a step change that requires a broad rethink of software and product development processes, especially when companies move from small pilot teams to deployments across thousands of developers.
Productivity pressure and modernization demand
McKinsey's recent survey of more than 300 product and engineering leaders highlights the uneven results companies are seeing. The top quarter of organizations report productivity more than doubling due to AI, while nearly 30% say productivity has dropped, underscoring the challenge of proving a return on investment at scale.That pressure is pushing large companies including Goldman Sachs, Cisco, and JPMorgan to be more selective about who gets access to AI tools and how those tools are used. Art Levy, Cognition's vice president of global partnerships, says Devin can draw on multiple AI models, allowing companies to use more expensive frontier systems for harder engineering work and cheaper models for less demanding tasks.
Levy says that approach is particularly useful for large software modernization programs that companies have delayed. He points to one client where an enterprise resource planning software upgrade expected to take five years is being compressed into 18 months, reflecting a broader market focus not just on whether AI can complete engineering work, but whether it can do so at lower cost.
We have also analyzed Noeri funding round as enterprises seek more cost-efficient AI tools alongside AI coding agents.