Hiring certified engineers

Audit vs Build vs Embedded: which Claude engagement to buy

A two-week audit, a six-week production build, and a quarterly embed solve different problems. Here is how we choose so you do not buy a rebuild when you needed a diagnosis.

May 11, 2026/4 min read/Claude Certified Engineers

Three offers sit on our pricing page: a two-week Claude Readiness Audit, a six-week Production Agent Build, and a quarterly Embedded Certified Engineer. Teams collapse those into "can you help with Claude" and then wonder why the work does not match the pain.

The choice is not sophistication. It is whether you need a diagnosis, a first production system, or a person in the repo after the project ends.

Audit: you already have something, and you cannot score it

Buy the Claude Readiness Audit when a Claude workflow exists (API, Bedrock, Vertex, Claude Code, a copilot, an internal MCP server) and leadership cannot answer:

  • Is quality going up or down?
  • What does a call cost at p50 and p95?
  • What happens on prompt injection in a document field?
  • Who is the agent authenticating as in Salesforce?

Two weeks inside the codebase. We review prompts, tool schemas, retrieval, cost, and safety, then hand back a prioritized remediation plan with measured baselines. Price is $9k to $16k.

Audit is the wrong buy if there is no system, only a slide. Measuring a prototype that cannot reach production wastes the two weeks. In that case you want a build.

Audit is also the wrong buy if you already know the agent is the wrong shape (catch-all REST tool, no eval, admin credentials) and you are ready to replace it. You can still start with two weeks if InfoSec needs a written baseline. Otherwise skip to build and do not spend a month arguing about a demo you intend to delete.

Build: one production outcome, six weeks

Buy the Production Agent Build when you need one agent that does a real job, in production, with tools, retrieval, budget ceilings, guardrails, and a review queue. Prototype by week two, live traffic by week six. $30k to $65k, fixed fee.

This is the path in six weeks to production. It is the right buy when:

  • The nine-month demo is not salvageable as-is.
  • You have a named outcome (invoice exception handling, contract intake, CRM research with a human queue).
  • You do not yet have an in-house owner who can ship that outcome this quarter.

It is the wrong buy when you want "AI strategy" or five agents at once. We will refuse a six-week scope that is actually a platform rewrite. Certification exists because Architect-level review is supposed to say no.

Embedded: the owner stays after week six

Buy Embedded when the problem is not a single launch. You need a dedicated Claude Certified Engineer in standups, on architecture review, shipping continuously, with a monthly roadmap session. Quarterly commitment, $12k to $20k per month.

Embedded is how you avoid the pattern where a build ships, nobody inherits the eval suite, and six months later you are back on vibes. It includes a certification track for your engineers. If you have people to enable, we enable them on your repo. If you do not, Embedded is the hire-shaped option without a 58-day recruiting cycle.

Do not buy Embedded as a cheaper build. The first month still has to name an outcome. An open-ended "help the AI team" retainer is how you get billable hours. We do not sell those below a quarterly commitment, and we still staff Architect vs Developer on purpose.

A simple rule

You haveYou needBuy
A live or almost-live Claude system you cannot measureA baseline and a ranked fix listAudit
No production agent, or a demo you will not hardenOne live job in six weeksBuild
A system (or a build about to land) and no dedicated ownerA certified engineer in the repoEmbedded
Engineers who will stay and sit the examsA four-week track on your codebaseEnablement

IP is yours on all three. Code lives in your repository from day one. See IP ownership. If you are stuck between audit and build, start with the outcome and the metric. The engagement is just the shortest path that makes that metric honest.

Questions

Should we start with a Claude audit or a production build?
Start with a two-week Claude Readiness Audit if something already exists and you cannot measure it. Start with a six-week Production Agent Build if you need one live outcome and the current demo is not a system you can harden.
When is an embedded certified engineer the right buy?
When you need continuous delivery, architecture review, and enablement inside your team, and a six-week project would leave a gap. Embedded is quarterly, from $12k per month.
What is the minimum engagement?
$9,000 for a two-week audit, or $30,000 for a six-week build. We do not bill hourly, and we do not take open-ended retainers below a quarterly commitment.

Keep reading

Hiring certified engineers

Six weeks to a production Claude agent

Prototype by week two, live traffic by week six. Here is the path we run on a Production Agent Build, and what has to be true on day zero or the calendar is fiction.

August 24, 2026/3 min read

Hiring certified engineers

Architect vs Developer staffing on a Claude engagement

Developer-certified engineers build the tools and agents. Architect-certified engineers decide what must not be built. Here is how we staff both so you do not get a clever system nobody can govern.

June 8, 2026/3 min read

Hiring certified engineers

IP ownership when a vendor builds your Claude agent

If the prompts, evals, and MCP servers do not live in your git from day one, you are renting a demo. Here is the IP rule we sign before week one, and what to reject in a vendor contract.

July 6, 2026/3 min read