Sound familiar?
What usually stops an agent from launching
If one of these is true for your agent, that is where we start.
“We can’t say what it costs per customer, and a few accounts could blow the budget.”
Cost“When it does something wrong, nobody can explain why.”
Evidence“Security wants to know it only sees what each user is allowed to see.”
PermissionsThe outcome
What you’ll have when we’re done
One outcome: an agent you can put in front of customers. You can check it four ways.
A measured failure rate
An evaluation suite built from your real traffic, running on every change, so you know how often the agent gets it wrong and whether a fix helped.
ReliabilityA spending limit per customer
Started in shadow mode, so nothing is blocked while you watch what the limit would do. Then enforced, with a planned rollback.
CostA receipt for every decision
Every model call allowed or stopped, with the customer, workflow and run behind it, and a timeline for every run.
EvidencePermissions enforced and tested
The agent only reads what each user may see. It cites its sources, or says “I don’t know” when the evidence isn’t there.
GovernanceHow it works
One path, in three steps. Prices in USD. Remote. Fixed scope and written deliverables, not open-ended billing.
1 · Diagnose
Production Readiness Diagnostic
$5,000 fixed · 2 weeks
A scorecard against the four outcomes, your top blockers ranked by business risk, and a fix plan with options. Read-only access to start; nothing changes in production. The fee is credited toward the build if you go ahead.
2 · Build
Production rollout
Typically from $18,000
A written proposal with options, scoped from the Diagnostic. Spending limits start in shadow mode before they enforce. Your team keeps the evaluation suite, receipts and runbooks; there is no long-term dependency on me.
3 · Keep it there
Monthly partnership
Optional · after a rollout
One focused track a month on whatever is slipping now: a new workflow, a cost regression, a failure that came back. It continues only while it keeps paying for itself.
Payment: 50% on start / 50% on delivery, or net-30. If the problem call shows I can’t help, I say so before you pay anything.
Want only the KazenAI product, with your team doing the integration? kazenai.com →
Fit check
Best when customers are about to depend on it.
Strongest fit: a customer-facing AI feature or agent in production or close to it, and a technical owner.
Good fit
- A launch, a large customer, or a pricing change depends on the agent behaving.
- Model spend or failures are growing faster than your confidence in them.
- You can share traces, request logs or billing exports under scope.
Not a fit
- Idea-stage demos with no real users yet.
- Open-ended staff augmentation without a defined outcome.
- Guaranteed savings figures, compliance certification, or work outside agreed scope.
Track record
20+
customer workflows where I architected production LLM features at DevRev
MSc
Data Science, University of Zurich
Top 0.5%
Kaggle Competitions Expert
Tooling I bring
Built by me, not client case studies. You don’t choose between them: they are how the rollout is delivered. Agent FinOps is KazenAI’s product; run receipts are part of it.
Also in the toolkit
Retrieval and evaluation, ready to adapt.
Internal tooling I adapt to your stack. Not separate products.
Cite-or-abstain retrieval →
- Answers come only from sources the user is allowed to see, with citations.
- It says “I don’t know” when the evidence is thin.
- 0 cross-source leaks in its permission fuzz tests.
Evaluation and run evidence →
- Evaluation harnesses built from your real failures.
- Every run leaves its plan, diff, tests, cost and failure class.
- Approval gates before anything risky ships.
Evaluating Agent FinOps as a product? Request a demo on kazenai.com →
How I work
Fast, and checked at every step.
I build with AI coding agents under strict gates. That is why one engineer can deliver quickly without guessing.
Speed
- AI coding agents do the volume; I own the design and the review.
- Every change runs a test suite before it lands. KazenAI alone has about 2,500 automated tests.
- Fixed scope and price, agreed before work starts.
Evidence
- Integrations are verified against the live provider APIs, not only mocks.
- Every engagement ends with written evidence your team keeps.
- Limits and claims are stated with their boundaries, never as guarantees.
What you do next
How engagements run
Clear sequence. No obligation until you approve the Diagnostic.
1. Send a problem brief
Short, non-confidential description of the agent and what is blocking it. I reply within one business day.
Required first step2. Problem call
30 minutes on your agent, who depends on it, and what is in the way.
If there is a fit3. Diagnostic
Two weeks, $5,000 fixed. You get the scorecard and a fix plan with options, or I tell you I’m not the right fit.
Scoped delivery4. Choose an option
Pick the rollout option that fits, or stop there. The Diagnostic fee is credited if you go ahead.
Your decisionReady to start?
Prefer email for a short intro, or if your company blocks web forms.
Background
I am an independent AI systems engineer. I take customer-facing LLM features and agents from demo to dependable production: measured reliability, a spending limit per customer, and a record of every decision.
Previously at DevRev, I built LLM features across 20+ live customer workflows. Before that, I did applied AI work with AXA XL and Sulzer in Zurich. I hold an MSc in Data Science from the University of Zurich and am a Kaggle Competitions Expert ranked in the top 0.5%.
KazenAI is the product I built from this work. Clients can use it; it’s never required. KazenAI is a product, not a consulting service, with its own scope, hosted demo and public Python packages. Under the engagement agreement, what I build for you is yours, KazenAI stays mine, and your code, systems and confidential data never go into it.
Send a consulting problem brief
This form is for consulting enquiries. I reply within one business day, and if there’s a fit, I send a private link for a 30-minute problem call.
Looking for the KazenAI product? Request an Agent FinOps demo on kazenai.com. Prefer email? [email protected]