How it works: your AI team, built and run by us

How it works: your AI team, built and run by usHow the work actually runsThis is the differentiator, and it is specific enough to check:

**The spec is the authority.** Written before anything is built, broken into dependency-ordered tasks, then locked and versioned. An unlocked spec is where scope drift comes from.

**Two human gates, no more and no fewer.** The client approves the plan, then reviews the code. Everything between those points is the engineer's job. Neither gate is delegated to a model.

**Manual-first.** No workflow is automated until it has been run by hand and the client has watched it work.

**Infrastructure setup.** Between the locked spec and the first role going live, we build the foundation. We set up the isolated client environment, configure the data model, and establish the connectors to your existing systems.

**Agents fan out, they do not stack roles.** Research, build and review are separate agents. One agent grading its own homework will pass itself every time.

**A recorded update every working day.** Around five minutes, camera on, same three beats: what shipped shown on screen, what is blocked, what is next.

**The delivery bar.** Nothing reaches the client without evidence it meets a written definition of good for that specific work. Tests pass plus a hand-run smoke test, dependency integrity checked, no committed credentials, row-level security on where the datastore supports it.

Embedded AI engineering

**Definition**: an embedded AI engineer works inside a client's existing team rather than alongside it. They use the client's repository, spec, channel and review process, and the client's own people approve the plan and review the code. The distinction from contracting is where the understanding ends up: because the client's engineers were in the loop, the reasoning stays in the building when the engagement ends.

- [When to embed rather than hire](https://maxpertise.net/blog/engineering-team-at-capacity)

- [Embedded engineer vs agency vs contractor](https://maxpertise.net/blog/embedded-engineer-vs-agency-vs-contractor)

- [What breaks when an agent writes your code](https://maxpertise.net/blog/what-breaks-when-an-agent-writes-your-code)

- [Spec-driven development with AI agents](https://maxpertise.net/blog/spec-driven-development-with-ai-agents)

Engagement terms

**Intake**: Written intake first. A call is an escalation, not the first move.

**Proposal**: Proposal and a profile of the proposed engineer back inside one business day.

**Interview**: The client interviews the engineer before anything is signed.

**Term**: Three-month minimum, with the first month risk-reduced.

**Guarantee**: Proof guarantee in the first 30 days.

**Flexibility**: Fast swap if the engineer is not the right fit.

**Ownership**: Clean exit, no lock-in and no handover fee. You own the code and the data.

**Speed**: Signed to embedded in about 10 days.

**Pricing**: One monthly retainer per AI team. Our pricing is built for predictable operations. We do not bill per token, per credit, or per seat. These models create a tax on growth and make it impossible to forecast your monthly spend. Instead, we bill a single monthly retainer per AI team. This retainer covers the build, the hosting, and the ongoing operation of the roles you need. The focus is on the outcome, not the usage.

Maxpertise is an AI-native engineering company. We embed native AI engineers inside your team, live in about 10 days. Please enable JavaScript to view the site, or email contact@maxpertise.net.