Karsten Wenzlaff, Advisor
August 26th, 2025
Agentic AI | July 14, 2025

Image: Freepik AI
Last Friday several outlets reported that Goldman Sachs is now using an AI software engineer named Devin. Now it may seem like what's the big deal here, everyone is using AI engineers, but this is a major U.S. bank, and Goldman says some developer teams using Devin are already producing 3x to 4x more output.
This is not a trial in a lab. It is a working system inside the bank. According to AInvest, Goldman has about 12,000 developers deployed across hundreds of teams (read 'at risk'). It's one of the first major publicly known cases of a major financial institution putting AI to work in a coding production environment.
Devin can complete full software tasks. It can set up development environments, run scripts, write and test code, and fix bugs on its own. It's agentic AI and can do these things without being told what to do at each step.
Having said that, it doesn't mean it always works (just yet). As reported by Fast Company, when Cognition Labs tested Devin, it only succeeded 3 out of 30 GitHub problems. That’s why Goldman still needs human developers to supervise the output currently. Teams check the code for accuracy and make sure it’s secure, and reviewed before it gets deployed.
Goldman is sending a clear message that AI tools are delivering measurable gains. Some software teams are producing three to four times more than before, and tools like the GS AI Assistant are already being used by 10,000 employees at Goldman for non-coding tasks. Any company that ignores AI now, seemingly will only fell behind on speed, output, and efficiency.
With Devin handling routine tasks, developers are now spending more time reviewing code, planning architecture, and testing fringe use cases, which are all tasks more valuable to the final product experience than basic coding (warning to junior developers).
So, companies must rethink how they hire, train, and structure their technical teams to focus on oversight and system integration.
Devin is currently a closed, proprietary tool unavailable to the public. It's also expensive and under NDA. That leaves smaller firms with0ut access and at a disadvantage. Canadian fintechs should be piloting open or commercial agent-based tools or partner with AI providers that offer more access to remain competitive.
AI code can fail spectacularly if unreviewed by a human currently. This can lead to security holes and/or logic gaps. Canadian companies also need human engineers to oversee similar governance, such as audits, version controls, quality assurance gates, and compliance reviews.
Not too long ago, we said the 'robots were here'! Well now Goldman is using AI to write their banking production code. Productivity is rising but it comes with new risks. Roles are changing, and oversight and governance is critical. Canadian financial technology companies must not fall too far behind. Those that jump in now will likely be in a better position to compete later.
The National Crowdfunding & Fintech Association (NCFA Canada) is a financial innovation ecosystem that provides education, market intelligence, industry stewardship, networking and funding opportunities and services to thousands of community members and works closely with industry, government, partners and affiliates to create a vibrant and innovative fintech and funding industry in Canada. Decentralized and distributed, NCFA is engaged with global stakeholders and helps incubate projects and investment in fintech, alternative finance, crowdfunding, peer-to-peer finance, payments, digital assets and tokens, artificial intelligence, blockchain, cryptocurrency, regtech, and insurtech sectors. Join Canada's Fintech & Funding Community today FREE! Or become a contributing member and get perks. For more information, please visit: www.ncfacanada.org
![]() | ![]() | ![]() |
|---|---|---|
![]() | ![]() | ![]() |
Leave a Reply