AI helps you produce more changes. Launch Trace helps you decide which are ready for customers. In ten business days, I trace the two customer journeys your business can't afford to have break, like signing up and paying, or the main thing your product does for a customer. I examine the evidence behind them and identify what must hold before your next launch or commitment. Deep on the two journeys that matter most. A security pass across everything else.
Fixed price · Two customer journeys · Ten business days
Book a free 20-minute fit callShow me what you're shipping and the commitment it needs to support. I'll tell you which two journeys I'd trace and why, and whether a Launch Trace fits.What I look for
- Do the specifications describe acceptable customer behavior, and do the tests prove that outcome?
- What evidence demonstrates success, who evaluates it, and who owns the decision to proceed?
- What can be checked automatically, and where does the workflow need human judgment?
- Can a rejected result, timeout, or partial failure resume without repeating completed actions? The same question applies when a provider you depend on pauses your account, leaks your stored keys, or shuts down.
- Are permissions, data boundaries, costs, and capacity appropriate for the business action?
- Which unresolved failures block the next launch or customer commitment, and which can wait?
What you get
A system map
A diagram of how each of the two customer journeys moves through your code, data, outside services, and permissions, and how it behaves once it is running.
Critical Workflow Qualifications
For each journey: acceptance criteria, evidence of completion, automated checks, human reviewer, rejection handling, and safe resumption. Findings identify launch blockers and evidence still missing.
Priority changes
Solution design for up to three priority decisions, with the smallest useful changes, alternatives, and tradeoffs tied to the customer outcome. Each design is written so your team or your coding agent can execute it without me in the room.
A roadmap and readout
An implementation and verification sequence sized to your team, and the readout: a recorded 60-minute walkthrough of the findings with Q&A.
A plain-English verdict
A straight answer you can read without an engineer in the room: ready to launch, or not yet, and what stands between.
A review summary you can forward
A one-page independent review summary you can send to a customer or investor. It says which two journeys I reviewed, what I found, and the date of the review.
Every Launch Trace includes
- One product, one milestone or consequential problem, and the two customer journeys your business can't afford to have break, traced by hand from start to finish. The Launch Scan bonus adds a security pass across the whole app.
- A system map and evidence-backed Critical Workflow Qualifications for both journeys
- An evidence path with acceptance criteria, reviewer ownership, rejection handling, and safe resumption
- Solution design for up to three priority architecture decisions
- An implementation and verification roadmap sized to your capacity
- A plain-English verdict: ready to launch, or not yet, and what stands between
- A one-page independent review summary you can forward to a customer or investor, with findings as of the review date
- A recorded 60-minute readout with the founder and implementation owner, followed by Q&A
- Thirty days of email questions about the findings after the readout
- Fixed price for the signed scope, delivered over ten business days. Unexpected complexity within that scope is my responsibility.
I personally perform every Launch Trace. No associates and no handoffs.
Bonuses
Launch Scan
An automated security pass over your code and your live app, run at kickoff and again after you fix what it finds. Exposed keys, open database policies, vulnerable dependencies, and the deployed surface. The findings go into the qualifications next to everything else, so security and reliability are judged on the same evidence.
Launch Kit
The templates and guides I use during the trace, packaged so your team can qualify the next workflow without me: the qualification format, the brief format for handing work to an engineer or a coding agent, and a guide to the signals worth watching in launch week.
From finding to launch decision
Sample workflow qualification
Verifying the lead-to-CRM workflow
For the timeout-and-retry finding above, this illustrative plan shows what the team would need to demonstrate before the next customer cohort.
- Acceptable outcome
- One qualified lead creates one CRM record and one email, even after a timeout or retry.
- Evidence to produce
- A failure-injection test, CRM records, email event logs, and the customer-visible result tied to the same workflow run.
- Automatic checks
- Assert exactly one CRM write and one email across timeout, retry, and resume scenarios.
- Human review
- The implementation owner evaluates the test and event evidence. The founder confirms that the customer outcome meets the agreed commitment.
- Rejection and resumption
- Keep the workflow blocked if the evidence fails. Fix the failure, resume from recorded state without repeating completed actions, and rerun the checks.
- Launch decision
- Proceed only when the acceptance criteria are demonstrated and the decision owner accepts any remaining risks. Missing evidence stays explicit.
After the Launch Trace
The Launch Trace stands on its own. Take the recommendations and roadmap to your team or any qualified development partner. If you want me involved while your team implements the plan, an optional three-month advisory engagement adds weekly implementation guidance, focused design and change reviews, and re-verification of the customer workflows against the outcomes we agreed on.
How it works
- Free 20-minute fit callYou show me the product, how you're shipping changes, and the next customer commitment. I tell you which two journeys I'd trace and why, and whether a Launch Trace fits.
- Define acceptable outcomesAt kickoff, we confirm the product, the milestone or consequential problem, and the two customer journeys I recommended, then agree on what a successful outcome looks like for a customer on each.
- Trace behavior and evidenceUsing read-only access, I follow the code, data, dependencies, and operational behavior. I examine what the tests prove, who reviews the result, and how failure or rejection is handled.
- Qualify the journeysI separate demonstrated behavior from supported risks and untested assumptions, identify launch blockers, and design the smallest useful changes and re-verification steps.
- Make the next decision clearIn a recorded 60-minute readout, I walk the founder and implementation owner through the evidence, readiness recommendation, and roadmap. Your team owns the final launch decision.
Who it's for
Solo founders and lean teams with a working product and a concrete customer commitment. You are building with AI, shipping quickly, and need confidence that the complete workflow delivers the right outcome. Products that use AI and ordinary SaaS built with AI both fit, whether you are approaching launch or supporting paying customers.
That includes founders who didn't write the code themselves. Maybe you built with Lovable or Cursor, or hired a dev shop. You have revenue or funding and a date coming up, and you need to know whether the product will hold.
- You didn't write the code yourself and need an independent read on it before a launch, a customer, or an investor depends on it
- Agent-written pull requests or rapid deployments are outpacing your capacity to validate them
- A tiny team is shipping customer requests within hours and needs a repeatable evidence and approval path
- Recurring defects and support work are eroding customer confidence
- Infrastructure or model costs are growing faster than the value of usage
- Features take longer to ship, or another engineer struggles in the codebase
- A larger customer needs security, isolation, auditability, or recovery controls
- Your security scan is clean and you still do not know what happens when a real customer, or the agent, takes the wrong path
- You are weighing a rewrite, infrastructure change, or engineering hire
- A launch, integration, or customer cohort has meaningful consequences if it fails
You need someone who can implement the plan: you, your employees, a development partner, or the coding agent you built with. I supply the workflow assessment, solution design, and verification plan.
Probably not for you if
- You have an idea, but no working product or customer problem to examine
- You want someone to build features or take on implementation work
- You need a penetration test or a compliance certification
- Your team cannot implement the recommendations
Questions founders ask
What does Launch Trace cost?
A fixed price for one product, one milestone or consequential problem, and two customer journeys over ten business days. The number is stated in the statement of work after the fit call, and it does not change inside the signed scope. Unexpected complexity within that scope is my responsibility, not an overage.
What access do you need?
Read-only access to the source repositories, infrastructure or infrastructure-as-code, deployment configuration, and observability for the agreed scope. I also examine specifications, tests, review evidence, and relevant customer requirements. Cost reports, support patterns, or release history may help. I don't need write access.
How much of my team's time does this take?
Approximately two to four hours across kickoff, up to three engineer interviews, and the readout. Most of the Launch Trace runs asynchronously over ten business days.
What exactly do I get?
A system map, Critical Workflow Qualifications for the two customer journeys, solution design for up to three priority decisions, an implementation and verification roadmap, a plain-English verdict on whether you are ready to launch, a one-page independent review summary you can forward to a customer or investor, and a recorded 60-minute readout with Q&A. Each qualification describes acceptable behavior, evidence, automated checks, human review, rejection handling, and safe resumption. Findings distinguish observed defects, supported risks, and hypotheses that still need validation. The Launch Scan bonus covers the rest of the app with an automated security pass.
Our engineers could do this themselves. Why hire you?
Your team knows the product. I bring an outside view shaped by twenty years across distributed systems, security, API platforms, and cloud infrastructure. The job is to test assumptions against the implementation and turn the evidence into focused decisions your team can execute.
We use AI code review too. What does Launch Trace add?
AI review can help your team handle more code, using the specifications, tests, and repository context available to it. Launch Trace examines whether those specifications describe acceptable customer behavior, whether the tests prove the right outcome, and whether the complete workflow can recover from partial failure. Your tools can inspect code and constrain an agent. I assess whether the complete workflow is ready for the commitment you are about to make.
Is this a security audit?
It includes one, and it keeps going where an audit stops. Every Launch Trace comes with an automated security pass over the code and the deployed app: exposed keys, open database policies, vulnerable dependencies, injection sinks. Those findings go into the qualifications next to everything else. Then I trace the two journeys as a real customer and, where it applies, as the agent: whether one tenant can reach another tenant's records, whether an admin route checks a role on the server, whether a payment webhook is verified and a refund revokes access, whether the agent's tools are scoped to the person who triggered them. A security scan cannot answer those questions because it does not know what your product is supposed to do. The reliability half gets the same treatment: timeouts, retries, partial failure, provider outages, and cost under load. If you need a penetration test or a compliance certification, that is a different engagement and I will say so on the call.
Our platform's security scan came back clean. Do we still need this?
Keep running it. A clean scan means the code is free of the patterns a tool can recognize: a committed key, a table with row-level security turned off, a known-bad dependency. It does not mean the policy on that table is correct for your customers, that the deployed app refuses a request for someone else's record, or that the workflow survives a retry without charging twice. The Lovable row-level security incident is the clearest example: the apps worked and the databases were open behind them. The scan checks what the code contains. Launch Trace checks what the product does.
There are fixed-price code audits for a few hundred dollars. How is this different?
Those are a reasonable first step and I include the same kind of automated pass. The difference is what happens next. A code audit reads your repository and reports what it finds. Launch Trace follows two customer journeys through the running system, including the AI surface and the operational behavior, and tells you which failures block the next commitment and which can wait. I also do not sell the fix work, so nothing in the roadmap is shaped by what I would like to build for you afterward. You get a ranked plan, the evidence behind it, and a readout from the person who did the tracing.
What happens after the Launch Trace?
Take the roadmap and work it with your team or any qualified development partner. Optional advisory is a separate three-month engagement, priced in its own statement of work, with weekly implementation guidance, focused design and change reviews, and re-verification against the agreed customer outcomes.
Do you assess products built with AI, or products that use AI?
Both. I trace the complete customer workflow: behavior, review evidence, permissions, failure handling, costs, and responsibility for the decision to proceed. The same questions apply whether AI wrote the code, takes actions inside the product, or both.
About
AI makes it possible for a small team to build and change software faster. I think that's good. It means more people can build something they own. As implementation speeds up, the work of deciding whether the behavior is correct, testing the customer outcome, and handling failure needs to keep pace.
That's the part I care about.
I've spent 20 years building and securing distributed systems, at Microsoft, Black Duck, runZero, and a handful of smaller companies. For the last 10 of those years, I've been helping founders launch and secure their software. I've watched systems fail under load, under attack, and under growth, and most of my opinions about retries come from watching them go wrong at scale.
I'm a founder too. I build my own products with the same constraints you have, limited time and my own money, and I'm honest in public about what breaks. So I'm not going to tell you to rebuild for a million users. I'm going to tell you the smallest set of things that need to hold at your next stage, and which ones can wait.
I personally trace the two customer journeys your next commitment depends on, examine the evidence, design the smallest useful changes, and hand your team a plan it can implement and verify. Whether I stay to help guide the work is your call.
What evidence would give you confidence to ship?
In twenty minutes, show me the product, how you're validating changes, and the customer commitment ahead. I'll tell you which two journeys I'd trace and why. If a Launch Trace fits, I'll say so. If it doesn't, I'll say that too.
