BlastRocket
See every product change before you launch.
The problem
AI coding agents let a team ship changes far faster than it can review them. The bottleneck moves from writing code to understanding what the code changed from a user’s point of view.
Code diffs, test results and preview links leave the accountable person guessing: what did users actually see change, does it hold up at phone, laptop and ultra-wide sizes, in the right state and role, and who needs to decide before launch?
BlastRocket is the evidence and approval layer between AI-assisted implementation and production.
Key decisions
One product change can touch dozens of screenshots across routes, states and screen sizes.
Review every comparison one by one, or group them.
Group related diffs into change bundles, so people review the product change rather than the pixels. Every bundle opens to its exact before, after and diff.
Grouping can be wrong, which is why the raw evidence is always one click away.
AI makes changes fast, and could approve them just as fast.
Let AI auto-approve likely noise, or keep a person accountable.
Humans keep launch authority. AI groups, summarizes and drafts a fix brief for a coding agent, showing its rationale, confidence and model, but it never approves, merges or deploys, and likely noise is never approved automatically.
More human clicks than a fully automatic pipeline, and a clear record of who said yes.
A missing screenshot can look the same as a passing one.
Treat absent evidence as fine, or as a blocker.
Fail closed. A strict launch gate blocks until every required capture, decision and policy is satisfied, and evidence is immutable: new captures supersede old ones without rewriting them.
A strict gate can feel heavy, so teams can choose advisory or approval-required modes instead.
The work
How it was built
- 01DiscoverCalls and walk-throughs with the people who would use it
- 02CompileTranscripts, files and constraints in one requirements brief
- 03SpecifyA detailed, sourced spec from Perplexity, reviewed and locked
- 04BuildDesigned and built with the latest AI models, in the production stack
- 05TestUser testing on every preview, with real users on real tasks
- 06IterateFindings turned into the next round of changes, worst problems first
- 07ShipTests, security review and release, then back to testing and iterating↻ Back to Test
Where AI was used
AI groups changes into bundles, writes the launch briefing and drafts fix briefs for coding agents, with its rationale, confidence and model shown. It never approves, launches, merges or deploys, and a deterministic metrics-only mode keeps images away from AI providers. Built with coding agents under the same rules as the other products.
What’s next
Running the capture runner on my other AI-built products first, then opening it to early teams.