Seven ways to bring AI into a $10M company. And when each one is right.

How should a $10M company decide how to bring in AI?

Start with two questions. Do you have a technical leader with time to own AI? And who will keep the work running after launch? Most non-tech companies between $10 million and $50 million in revenue answer no to the first and nobody to the second. That is why AI spreads across a company without changing how the business runs. The Federal Reserve Banks' 2026 Small Business Credit Survey report found that just 7% of firms using AI had it fully integrated into the business. Another 49% were still experimenting.¹

How do the options compare?

When is each option the right choice?

When is doing it yourself with ChatGPT, Claude or Copilot enough?

For almost every company it is the right first step, and for a small team it may be all you need. These are good tools. They help people draft, research and summarize, and a seat costs little. Most owners already know this firsthand: Vistage's Q4 2025 CEO Confidence Index found that 76% of small and midsize business CEOs personally use generative AI.² The limit shows up at the level of the business. The tools are not connected to your systems, and results depend on who is typing. Read the full comparison with an AI partner.

When is the AI inside your existing software enough?

It is enough when most of the work you want to improve happens inside one platform. CRM, accounting and field service vendors keep adding AI features to products you already pay for, and most can be switched on without a project. Use them. They tend to fall short when a workflow crosses systems, such as a quote that starts in email, pulls prices from a spreadsheet and ends in your accounting software. They also assume your process matches the vendor's default.

When should you hire an AI engineer?

Hire when AI is core to what you sell, you have a steady backlog of things to build, and a technical leader can manage the role. Budget for it honestly. Kore1's 2026 AI engineer salary guide puts mid-level base pay at $140,000 to $210,000, before benefits and recruiting.³ In a non-tech company, that engineer often works alone, with no one to set priorities, train the team or take over if they leave. Compare hiring with an outside team.

When is a fractional CTO the right choice?

A fractional CTO is right when you need part-time senior leadership across all of your technology, with AI as one part of it. Think security, software vendors, systems that need replacing, or a development team without a leader. Fractionus's 2026 cost guide puts US fractional CTO retainers at $9,000 to $22,000 a month.⁴ Many fractional CTOs advise and oversee rather than build, so plan for who will do the building and keep it running. Compare a fractional CTO with an AI partner.

When is an AI automation agency the right choice?

An agency is often the fastest and least expensive choice for one clearly defined automation, such as routing inbound leads or pulling data from invoices. Taskip's 2026 pricing guide puts a basic automation build at $1,500 to $12,000.⁵ If you already know what to build and someone on staff can own it after handoff, an agency may be all you need. It fits less well when you first need to decide what to build, or when people have to change how they work. Compare an agency with an AI partner.

When is a large consulting firm the right choice?

A large firm is the better fit when the work needs an enterprise brand behind it: programs across several countries, heavy regulation, close lender or board scrutiny, or AI work tied to an advisory relationship you already have. These firms bring deep benches and tested methods. For an owner-led company, the usual tradeoffs are cost, a scope weighted toward strategy, and less time with the senior people who sold the work.

When is an AI partner the right choice?

An AI partner fits a non-tech company without a CTO that wants AI working across several workflows, used by its people and maintained after launch. One team handles strategy, the build, change management and upkeep, so nothing falls between vendors. It is the wrong choice if you need one small automation, want advice only, or already have engineers who can own AI. MIT NANDA's 2025 study found that work done with outside partners reached deployment about 67% of the time, against about 33% for internal builds. The authors caution that the gap may reflect the companies as much as the approach.⁶

Can you combine these options?

Yes, and most companies should. Keep the licenses your team already uses, and turn on the AI features in your core software. Add outside help for the work that crosses systems or asks people to change how they work. Some companies start with an outside team and hire an engineer later, once there is a running system and a backlog to hand over. Whatever the mix, write down who owns each system after launch.

How Modern Thesis handles this

Modern Thesis is an AI strategy, implementation and stewardship firm for non-tech companies, typically with $10 million or more in annual revenue. Every engagement starts with a strategic assessment in the first week of the project. We map how work really moves, measure time, cost and error rates on the workflows that matter, and rank the opportunities by return. We use the tools you already pay for where they fit and build custom where they don't. Jim Coleman leads strategy. Our AI specialists and Forward Deployed Engineers lead the build, and stewardship has no end date. You own the code, accounts and data. If AI won't pay for itself somewhere, we'll tell you.