Practical notes on designing, building, and shipping small software.
A practical look at the prompts behind our pest-control software: building complete workflows, protecting service history, and simplifying an MVP after using it.
A disputed phishing report disrupted my work at Bear Trail Works. Days after Doppel said it was working with Cloudflare toward reinstatement, I still have no substantive update, clear case owner, or restoration timeline.
Compare the workflow, the ongoing work, and the cost of changing tools before you commit to a custom build.
You can ship a product and still have no business. Selling, marketing, and networking deserve a place beside building.
A practical look at the prompts behind our pest-control software: building complete workflows, protecting service history, and simplifying an MVP after using it.
Astra’s biggest gains are in carrying difficult work through to a usable result. Here is what the benchmarks, developer docs, and early customer reports actually show.
AI can turn a prompt into a prototype. Turning it into software people can depend on is still engineering.
Check your real usage, use reset forecasts as clues, and turn fresh Codex capacity into something worth shipping.
A practical way to estimate an MVP by identifying the real work, exposing cost drivers, and turning a budget into better product decisions.
A practical one-page brief for turning an MVP idea into observable behavior, firm boundaries, and evidence a builder can verify before release.
Choose the smallest artifact that answers your riskiest product question before you spend time building the wrong thing.
AI can accelerate the path from a clear brief to working software. It cannot choose the right problem, verify demand, or own what you ship.
A field report on using specifications and coding agents to move a Flutter POS MVP toward a real beta user—while keeping judgment where it matters.
Describe the outcome, set clear approval boundaries, and require evidence without prescribing every implementation step.
/plan, then /goalUse /plan to agree on a change before editing, then /goal to keep Claude Code working until a clear, testable outcome is met.
Create a clear brief, delegate bounded work, and verify the integrated result when an AI agent coordinates multiple assignments.
A practical way to turn an early idea into a clear, buildable first release—without confusing scope for progress.
Use constraints to protect the core experience, make faster decisions, and give a first version a real chance to teach you something.
Look for a partner who can clarify the work, make tradeoffs visible, and help a project move from a promising idea to a useful product.
I help founders and teams design and build thoughtful software. If you are looking for a partner for your next project, let’s start a conversation.