[bmdpat]

Architect

A fixed-price Claude architecture review for Azure teams.

Claude now runs inside Azure through Microsoft Foundry. I hold Anthropic's Claude Certified Architect credential and Microsoft's Azure Solutions Architect Expert certification, and this company ships through a Claude agent fleet with CI gates and a public provenance ledger. Send your plan through the intake. The architecture comes back in writing.

$1,500 flat · async · delivered in writing · no calls at any step

§ 001 / CREDENTIALS // verify them

Certified on both sides of the stack.

Claude Certified Architect - Professional

Anthropic · Issued 2026-08-20 · expires 2027-08-20

[ Verify on Credly ]

Azure Solutions Architect Expert

Microsoft · AZ-305 · Expert level

[ Verify on Microsoft Learn ]

Azure AI Engineer Associate

Microsoft · AI-102 · Associate level

[ Verify on Microsoft Learn ]

§ 002 / THE REVIEW // what you get

Six answers, in writing, five business days.

  • Model and deployment choice: Microsoft Foundry, the Anthropic API, or both, matched to your data boundary.
  • Agent boundaries: what each agent may touch, and what it may never touch.
  • Runtime ceilings: budget, loop, timeout, and rate limits, with the defaults I run in production.
  • A cost model per workload, with the numbers behind it.
  • Failure modes and what recovery looks like for each.
  • A build order for the first 90 days.

01

Send the intake

Ten minutes of your time. Your Azure footprint, what Claude should do, your timeline.

02

Fit answer in 2 business days

A written yes or no by email. A no costs you nothing and comes with a pointer at what to read instead.

03

Pay first, then the clock starts

$1,500 flat, fixed and public. The payment link comes with the yes. No work starts before it.

04

The review lands in 5 business days

A written architecture review covering the six items above. Questions handled by email.

§ 003 / RECEIPTS // audit me first

I run what I recommend.

This site, including this page, ships through a Claude agent fleet: agents write the code, tests gate the merge, and a provenance ledger records what each session claimed and proved.

The agent fleet ledger

Every agent session on this company records falsifiable claims with checks that can fail.

[ github.com/bmdhodl/showwork ]

AgentGuard

Runtime ceilings for agents: budget, loops, timeouts, rate. pip install agentguard47.

[ Read the docs ]

The 5090 lab

Measured local model runs on owned hardware, published with raw artifacts.

[ Read the reports ]

§ 004 / INTAKE // ten minutes

Start the fit check.

No payment here. You pay only after a written yes, and before the work starts. No calls at any step.