The AI Question Every Service Firm Should Ask Before It Signs
Think about the core systems your firm runs on: your CRM, your document management, your accounting platform, your email. You license the software, but the data and the work you do in it are yours. You can export it, and at renewal you can walk away if something better comes along.
AI is increasingly being sold differently. You pay per seat or per interaction, and everything your firm teaches the tool (your prompts, your workflows, the way it has learned how you work) lives inside someone else’s platform. If you leave, most of what your firm built stays behind.
For accounting firms, advisory practices, law firms, consultancies, wealth managers and other service businesses, that’s a bigger deal than it looks. AI is starting to shape how work gets done, how clients are served and how client information is handled. So before you ask “What can this AI do?”, ask a more important question: “Who keeps what it learns about our firm?”
Lock-in measures dependence, not loyalty
For decades, software vendors protected recurring revenue by making it hard to leave: proprietary formats, integrations that are painful to unwind, contracts with steep exit costs. It’s a rational strategy, but it quietly shifts the vendor’s focus from “How do we keep earning this business?” to “How do we make leaving harder?”
A client who can’t leave isn’t the same as a client who chooses to stay. Lock-in shows up as retention on a dashboard, but underneath it is hedging, resentment and a procurement team already scoping the exit. When a firm keeps what it has built and can leave at any time, every renewal is a real decision, and the provider has to keep earning it.
With AI, the stakes are higher. Being locked into a system that holds your firm’s know-how and your clients’ information is very different from being locked into a project management app.
It’s a compliance question
Most professional service firms already carry a legal or ethical duty to know how outside providers handle client information. None of these rules were written specifically for AI, but they apply to AI vendors just as they apply to any other vendor:
- Investment advisers and broker-dealers. The SEC’s 2024 amendments to Regulation S-P require covered firms to oversee their service providers and make sure those providers notify the firm of a breach affecting customer information. Smaller firms had to comply by June 3, 2026. (SEC compliance guide)
- Accountants, tax preparers and bookkeepers. The FTC’s Safeguards Rule treats many of these firms as financial institutions. It requires them to protect customer information and to take steps to ensure their service providers do the same. (FTC Safeguards Rule)
- Lawyers. The American Bar Association’s Formal Opinion 512 (July 2024) applies the duties of competence, confidentiality and client communication to generative AI tools. It expects lawyers to understand how a tool uses the information they put into it. (ABA announcement)
Each of these comes down to the same thing: you’re accountable for where your clients’ information goes and who controls it. That’s much easier to answer when your data stays in your accounts and every action the AI takes is logged than when it all sits inside a black box.
It’s a value question
There’s a second reason this matters, and most firm owners haven’t thought about it yet.
The value of a service firm has traditionally lived in two places: client relationships and people. Both are valuable and both are fragile. Relationships can walk out the door. Key people retire or leave. Anyone who has bought or sold a practice knows that risk gets priced in.
AI can capture part of what makes a firm work: refined workflows, institutional knowledge, the way your best people handle specific situations, and documented processes that don’t depend on any one person. If that knowledge is yours, it stays with the business.
If it lives only inside a vendor’s platform, it doesn’t. When the subscription ends, the workflows, rules and context your firm built go with it.
To be clear, this isn’t a promise of a specific valuation bump. It’s a simple principle: the more of your firm’s operating know-how lives in something you keep and can transfer, the stronger your position when it’s time to bring in a partner, plan succession or sell.
What your firm should keep
The AI model itself is an engine. It’s fine to rent it, and you should, because models improve and change price constantly. What matters is what your firm keeps. Four practical tests:
- Your workflows, rules and institutional knowledge are yours. What your firm teaches the system belongs to your firm, not the vendor.
- Your data stays your data. It isn’t used to train someone else’s product, and you can get it out in a usable form.
- The model is replaceable. You can switch the underlying AI model without losing everything your firm has built on top of it.
- It’s documented and auditable. You can see what the system did and why.
If any of these is missing, the vendor holds what your firm has learned, and the leverage that comes with it.
How we built Sofia
Sofia is the AI platform we build at FINdustries for service firms. The platform is ours; what it learns and builds for your firm is yours. We designed it around these four tests, because we believe trust is architecture rather than a promise.
- What Sofia learns about your firm belongs to your firm. Your workflows, institutional knowledge and data are built in your accounts, on your data, for your way of working.
- The model is swappable. The AI model underneath is a supplier. If a better model arrives or pricing changes, your workflows stay put.
- It’s built on open standards. Sofia connects to your systems through standards such as the Model Context Protocol, an open standard for connecting AI to business tools and data.
- Every action leaves a receipt. You can see what was done, when and why, so control comes before delegation.
Our bet is simple: if clients keep what they’ve built and can leave, the right ones will stay because the work is worth it.
Questions to ask any AI vendor before you sign
- Do we keep the workflows, rules and data our firm builds, or do they stay with you if we leave?
- Can we export our data in a usable format?
- Can the underlying AI model be swapped without rebuilding our workflows?
- How does your setup help us meet our obligations to oversee service providers that handle client information?
- If we sold the firm or changed leadership tomorrow, what would transfer?
A vendor who answers those clearly is telling you they plan to keep your business by being worth it.
The bottom line
AI is becoming the operating layer of how service firms think, work and serve clients. Rent the engine. Make sure your firm keeps what it learns.
Don’t just ask what the AI can do. Ask who keeps what it learns.
Want to see how this works in practice? Talk to FINdustries about Sofia.
