See what each fix is worth. Then ship the ones that pay.
A one-week audit puts a dollar value on every fix to your LLM feature. Hands-on implementation ships them inside your repo, highest saving first, so the savings reach your bill. Each step is quoted separately at a fixed price, scoped from your findings, and none is required.
- 01
Audit
1 week, read-only, $5,000 fixed. Every fix priced in $/month. You keep the report either way.
- 02
Hands-on implementation
4–8 weeks in your repo, highest-saving fixes first. Quoted separately from the audit as fixed-price packages, measured in dollars against the baseline.
- 03
Ongoing
A monthly retainer to keep costs down as usage grows, only after two or more packages.
The flagship: AI Cost & Reliability Audit and Hands-on Implementation
- 01Booking now
AI Cost & Reliability Audit
Founders, CTOs and engineering leads with LLM features in production.
AI spend is growing faster than revenue, nobody can say what a feature costs, and customers find quality problems first.
Audit · About 1 week · $5,000 fixed · pilot free
- Every fix priced: estimated savings per month, ranked
- What the feature really costs per successful outcome
- Where quality and reliability slip, and what it costs you
- A cost model for pricing and margin as usage grows
Optional next step · Hands-on Implementation · quoted separately
- Highest-saving fixes shipped first
- Fewer failures: retrieval, prompts, fallbacks
- Quality gate, so savings never cost quality
- Cost and quality dashboards per feature
- 02Follows the audit
Hands-on Implementation
Teams that have had the audit, or already have a measured baseline, and want the fixes shipped in production.
The fix list is priced, but nobody on the team has the weeks to ship it safely, so the savings stay on paper.
Implementation · 4–8 weeks · fixed price per package
- Highest-saving fixes shipped first
- Lower spend: routing, caching and prompt trimming
- Fewer failures: retrieval, prompts, fallbacks
- Quality gate in CI, plus cost dashboards per feature
Then · Ongoing · retainer after two or more packages
- Monthly eval and cost review
- Model and prompt upgrades through the eval gate
- Pairing until your team owns it
Compare the two
| AI Cost & Reliability | Implementation | |
|---|---|---|
| For | Founders, CTOs and engineering leads with LLM features in production | Teams that have had the audit, or already have a measured baseline |
| Trigger | AI spend growing faster than revenue, cost per feature unknown, customers find regressions | The fix list is priced, but nobody has the weeks to ship it safely |
| Audit length | About 1 week | 4–8 weeks |
| Price | $5,000 fixed, audit only · pilot free | Fixed, quoted per package |
| You get | Every fix priced in $/month and ranked, real cost per outcome, cost model, quality test set | Savings and reliability fixes live in production, measured in $ against the baseline |
| Next step | Hands-on Implementation · 4–8 weeks · quoted separately | Ongoing retainer, after two or more packages |
Supporting audits
Read-only audits for when the cloud bill or margin per customer is the bigger problem. Same rules: fixed scope, fixed price, every saving in dollars, and the report is yours.
- 03Booking now
Cloud Cost Audit
Find the cloud waste worth cutting, ranked by $/month with a risk rating, and give every dollar an owner. 1 week, fixed price.
- 04Booking now
SaaS Unit Economics & Margin Audit
Find out which customers and plans make money, which lose it, and what to change in pricing and limits. 2 weeks, fixed price.
How every engagement works
The same rules apply to every audit and implementation.
You keep the report
Either way. It's never tied to buying the follow-on, and you can hand the fix list to anyone.
The implementation is scoped from your findings
Your report ends with a priced fix list built from your own data, not a price list.
Fixed scope, fixed price
Audits are fixed. Each implementation package is fixed too, moving to a retainer only after two or more packages.
Changes are priced separately
Anything outside the signed package gets its own price, agreed before the work starts.
Anonymized case-study clause
Every audit agreement includes one: industry, scale and relative metrics, never your name without sign-off.
One implementation at a time
I work inside one team's codebase at a time, so yours gets full attention.
Where to start?
Start from the problem. If it's more than one, or you're not sure, bring it to the free 30-minute cost check and I'll tell you where to start.
Get a free cost check- Your LLM bill is growing, and nobody can say what a feature costsAI Cost & Reliability Audit
- You know which fixes would save money, but nobody has the time to ship them safelyHands-on Implementation
- Your cloud bill is growing, resources are untagged, and nobody owns costCloud Cost Audit
- You don't know which customers or plans lose moneySaaS Unit Economics & Margin Audit
Hiring for an AI engineering role instead, contract or full-time?
For hiring managersGet a free 30-minute cost check
Not sure where to start? Bring the question. You leave with a clear answer, and no pitch if nothing fits.
Choose a timeOr email [email protected]