Goldman Sachs is one of the world’s leading investment banks.
The firm advises corporations, governments, and institutions on mergers and acquisitions, raises capital across public and private markets, manages $3.6 trillion in assets, and runs one of the most active trading operations on Wall Street.
With over 46,000 employees worldwide, it sits at the centre of global finance.
Right now, it's in the middle of one of the most aggressive enterprise AI rollouts for a company of its scale.
In today’s AI at the Top, we will learn how Goldman Sachs operates with AI and what we can learn as enterprise leaders.
The Strategy Behind Goldman Sachs’ AI Approach?
In Q3 2025, Goldman launched what CEO David Solomon calls “One Goldman Sachs 3.0”.
It’s a new phase of AI adoption within the company.
Under OneGS 3.0, Goldman is targeting a set of highly repeatable workflows for AI and automation, including sales, client onboarding, lending processes, regulatory reporting, and vendor management.
The firm rolled out internal AI tools to reshape day‑to‑day banking and trading workflows and lift employee productivity.
What I loved most about GS’s strategy is the company didn’t roll out AI entirely, just as a fad.
In fact, the CEO mentioned, “We are not going to transform the whole firm with AI.”
Goldman identified 6 specific business processes (not disclosed by the CEO) they believe are crucial for AI-driven re-engineering. The enterprise plans to expand after they automate concrete workflows and measures the results.
Enjoying this read? Share the smartest part with your followers. Click to Share.
In terms of governance, GS applied a similar approach to J&J and Mastercard.
The company formed centralised AI protocols via which all the AI activities run across teams.
So when individual business units run experiments, basics like compliance checks, security protocols, and audit trains are built into the foundation, not as an afterthought.
The company believes the centralised governance helped the deployment move fast.
Unlike most Fortune 500 companies we have studied so far, GS’s key architectural difference is its multi-model nature. Kind of how you can choose different models in Perplexity.
The enterprise works with multiple vendors like OpenAI, Google, Anthropic, and Meta. By doing so, the system routes tasks to the model best suited for the task.
This also puts GS at an advantage. They aren’t exposed to the pricing, performance, or strategic decisions of any single AI provider. If a better model launches tomorrow (highly likely with the speed AI is moving), they can integrate it into their internal platform super fast with minimal costs.
YOU DECIDE
🌟 Want to be featured in the next issue? Reach out with your best AI use case and we’ll spotlight it.
🏢 What company do you want us to cover next?
What has Goldman Sachs’ Internal Platform GS AI unlocked for its Business?
GS AI Platform is the internal gateway to all GenAI activities within.
Every employee who uses AI at GS, be it a dev writing code, an analyst summarising a report, or a banker building a pitchbook, uses the GS AI platform.
Some usecases and what AI unlocked for GS’s employees:
More than 50% of GS’s employees actively use GS AI. The firm wants to hit 100% adoption by the end of 2026.
“In my 40 years in technology, 2025 saw the biggest changes I have seen in my career. And what’s crazy is we haven’t seen anything yet.”
What's Next for Goldman Sachs
There’s no surprise agents will be the next important thing business units will look to adopt.
We have studied 15+ companies so far on AI at the Top.
If I have to find a common thread between large enterprises using GenAI, I’d say most companies crossed Level 1 and are now moving into Level 2.
Level 1: Use AI to save. It could be time, human resources, costs, or increasing speed.
Level 2: Understanding what value the savings have unlocked, and optimising by doubling down on the value.
Of course, Fortune 500 companies have large proprietary data sets that will continue to compound in value and bring company-specific, accurate use cases with the new models and workflows.
What Enterprise Leaders Can Learn from Goldman Sachs’ AI Strategy
Build the platform before the tools. Goldman’s centralised platform made rapid, secure deployment possible. Teams that skip this step end up with fragmented tools, inconsistent governance, and slow scaling.
Start where the data is richest. Goldman went to developers first because the output is measurable. Plus the feedback loop is fast as a quarter of the company’s workforce are engineers. Find your equivalent function where the results are easiest to verify.
Don’t deploy AI broadly. Deploy it specifically. Role-specific copilots drove higher adoption and bigger impact than generic assistants. A banker using a tool built for pitchbooks will use it more than a banker using a general chatbot.
Governance is an accelerator if done right. By embedding compliance into the platform from day one, Goldman avoided the back-and-forth that slows most enterprise AI rollouts. Security reviews, policy checks, and audit trails were already there when it was time to scale.
Sources:
YOUR TURN