📰 AI News: OpenAI and Thrive test a “built in” model for enterprise AI

📝 TL;DR

OpenAI is taking an ownership stake in a holding company called Thrive and embedding its own engineers directly inside accounting and IT service firms. This is a live experiment in what happens when AI is not just bolted on, but wired into how a business actually runs day to day.

🧠 Overview

Thrive buys traditional service businesses like accounting firms and IT providers, then rebuilds them around better data and AI. OpenAI is going beyond a normal vendor relationship. it is taking equity in the platform, sending in its own teams, and co-designing tools with frontline staff.

For everyone watching from the outside, this is a test case for what “serious” enterprise AI transformation might actually look like in the real world.

📜 The Announcement

It was reported that OpenAI will take an ownership stake in Thrive Holdings, a platform company set up to modernise day to day service businesses. Thrive’s first two big bets are an accounting group and an IT services provider, backed by commitments in the hundreds of millions of dollars.

OpenAI will embed research, product, and engineering teams inside these companies to build custom tools for high volume, rules based workflows, with the goal of creating a repeatable model that can be extended to other industries.

⚙️ How It Works

👉 Buy the workflows, not just the licenses

Thrive acquires firms that already sit in the middle of essential processes like tax, bookkeeping, and IT support. Instead of selling them generic AI, it rebuilds how they operate from the inside.

👉 Embed OpenAI teams inside the business

OpenAI is assigning researchers, product managers, and engineers directly into Thrive’s portfolio companies, working alongside accountants and IT staff. The people building the models see the messy reality of tickets, deadlines, and client requests in real time.

👉 Target high volume, rules driven tasks first

The initial focus is processes like data entry, reconciliations, first pass tax workflows, and standard IT operations. These are repetitive but governed by clear rules and service levels, which makes them ideal for automation.

👉 Share upside on both sides

OpenAI does not just collect usage fees. it also gets equity exposure to any performance gains and growth that its tools help unlock inside Thrive’s companies. In return, Thrive gets privileged access to frontier models and bespoke tooling.

👉 Turn it into a template for other sectors

If this “buy, embed, rebuild” model works in accounting and IT, the plan is to replicate it in other industries that run on heavy process and data. Think claims handling, back office operations, and complex support services.

💡 Why This Matters

This is enterprise AI as infrastructure, not just an app - Instead of plugging a chatbot into existing workflows, this approach reshapes the workflows themselves around AI. That is a much bigger shift than simply rolling out another tool across a company.
It finally puts AI builders in the same room as domain experts - Too many AI projects are built in isolation, far from the accountants, agents, or engineers who will use them. Embedded teams mean models are tuned to real edge cases, regulatory constraints, and “this is how it actually works on a Monday morning” reality.
It is a signal of where big labs want to make their money - Consumer subscriptions are useful, but the long-term upside is in transforming operations and productivity inside established businesses. Deep partnerships like this show that serious revenue is expected from reengineering how traditional firms work.
It blurs the line between vendor and owner - When your AI supplier also owns a chunk of the business, incentives align in some ways and get more complicated in others. There will be growing questions about data access, dependence on a single lab, and whether a few AI providers are becoming “shadow operators” of many service firms.
It shows that “AI transformation” is as much about org design as tech - The main learning here is not only which model performs best, but how you organise teams, incentives, and governance so AI projects actually get finished, adopted, and trusted by staff.
If it works, copycats will follow quickly - Investors, holding companies, and large corporates will be watching this experiment very closely. Success could trigger a wave of similar “buy a platform, embed AI, scale operational gains” plays in unglamorous but critical industries.

🏢 What This Means for Businesses

Borrow the “embedded AI team” mindset, even if you cannot buy companies - You do not need a billion dollar fund to act like this. pull AI people closer to the frontline, sit them with ops and customer teams, and let them shadow real work rather than just reading process docs.
Start with one process that is boring, rule heavy, and measurable - Think invoice capture, onboarding emails, or level 1 support tickets. Those are exactly the kinds of high volume, rules based flows being targeted here, and they are the easiest to pilot, measure, and improve.
Negotiate for partnership, not just software access - When you talk to AI vendors, ask what implementation help, training, and co design support they offer. The real value is often in people, templates, and playbooks, not just logins and API keys.
Treat your data like a strategic asset, not a byproduct - This model works because it gets control of clean, rich operational data inside the businesses it touches. Even as a smaller company, you can start documenting processes, cleaning core datasets, and setting sensible data permissions so future AI projects do not stall.
Think “inside out” transformation instead of tool shopping - Rather than asking “Which AI tools should we use,” ask “Which parts of our workflow could be redesigned if AI were a native teammate here.” That simple shift in question moves you closer to the kind of change being tested here.
Position yourself as the human layer on top - For solopreneurs, coaches, and small firms, the opportunity is to ride these infrastructure shifts while staying fully human in how you sell, decide, and support. Let the giants industrialise the back office while you double down on trust, insight, and relationships.

🔚 The Bottom Line

This partnership is a real world experiment in deep, inside out AI transformation that mixes ownership, embedded teams, and unglamorous but vital workflows. If it succeeds, it could become a template for how AI seeps into the backbone of the economy.

You may not control these big deals, but you can absolutely borrow the underlying playbook in your own, much smaller world.

💬 Your Take

If you could embed a tiny “AI team” inside one part of your business this quarter, which workflow would you choose first, and what result would convince you it was worth scaling up?

4 comments