The service company of the future will build some of its own software

A services business usually grows by adding people. More clients mean more proposals, more handoffs, more reviews, more questions, and eventually more staff to keep the work moving.

AI offers another way to grow: turn the parts of delivery that repeat into software the firm owns.

Think about everything that happens around the expertise a client actually pays for. A request comes in. Someone gathers context, checks whether it is a fit, assembles a proposal, finds the right people, tracks approvals, and makes sure delivery knows what was promised. At ten clients a month, a good team can hold much of that together through experience. At a hundred, the same habits become delays and mistakes.

A firm can now build a focused system around that process. It might collect client information once, prepare a first draft of the proposal, flag missing approvals, and carry decisions into the delivery team’s workspace. People still make the consequential calls. They spend less time finding information and repairing handoffs.

That is where the economics change. The firm can deliver faster and more consistently without increasing coordination costs at the same rate as revenue. It can bring new employees up to speed more easily because the process is visible. It can catch some risks before they reach the client. Its best way of working becomes something the whole team can use.

Custom software can also solve a problem that individual AI tools can leave unresolved: teams working from different versions of the truth. Sales knows what the client requested. Delivery knows what it can provide. Finance knows what was priced. When that context lives in separate inboxes, documents, and AI chats, each person may work faster while the handoffs remain fragile.

A shared system can give those teams one place to see the current scope, decisions, approvals, and changes. AI becomes a multiplayer experience within the work: it can help draft a proposal, surface a conflict, or prepare a handoff using context the next team can also see and correct. The value comes from connecting people’s work, not just accelerating each person’s tasks.

This is also why “productized services” needs a little precision. Some firms will package their expertise into clearly defined offers with a repeatable scope and price. Others will continue to do highly tailored work. Both can build a repeatable system behind the service, leaving more time for the judgment and attention that make the work valuable.

The shift is underway, though its business impact is still being worked out. In Thomson Reuters Institute’s 2026 report, based on a late-2025 survey of 1,514 professionals in legal, tax, accounting, risk, and government roles, 40% said their organizations use generative AI, while only 18% said their organizations collect metrics around AI’s return on investment. Using AI in individual tasks is becoming common. Redesigning delivery around it (and measuring whether clients receive a better service) is the harder opportunity. ^[1]

AI can reduce the initial coding effort involved in building a focused tool, making that opportunity more accessible to smaller firms. A narrow custom system becomes another option alongside buying and configuring existing software. They can start with one costly workflow, build a narrow tool around it, and expand it only if it proves useful. That tool still needs testing, security, maintenance, and an owner. The savings have to survive those costs. ^[2]

Start with the handoff that gets harder every time the business grows. If sales promises work that delivery cannot staff, connect scope and capacity. If changes disappear between account management and finance, connect approvals and billing. Build around a specific failure, and measure whether the system reduces delivery time, staff hours per engagement, and rework.

Then make each engagement improve the next one. A useful qualification rule, an approved proposal structure, or a recurring delivery issue should become part of the shared process. With the right permissions, teams and their AI tools can use that knowledge without repeatedly reconstructing it. The company starts accumulating an operating capability that survives beyond any one employee’s memory or chat history.

That capability creates commercial choices. A firm can offer a faster turnaround, serve smaller clients profitably, or sell a clearly scoped service at a fixed price while retaining some of the efficiency gain. Growth becomes less dependent on adding people in direct proportion to revenue. The return on building software is a service the company can deliver repeatedly, across more clients and teams, with fewer hours and fewer mistakes.