Arian Askarizad · Product Manager
I'm a product manager. I find the problem, talk to the people who have it, decide what to build, and then build it myself. In 2026 I took four AI products from problem to launch with no engineering team, working daily in Claude Code. Each case study below shows the product calls and the code they turned into.
Instant quotes for floor-coating contractors. A homeowner uploads a photo, sees it rendered in the contractor's finishes, and gets a line-item price from the contractor's own rates.
A 24/7 AI receptionist for trade contractors. It answers customer texts and phone calls, quotes jobs, books appointments and collects payments while the owner works.
Interview prep generated from your real résumé: live voice mock interviews, coding interviews where the code actually runs, job-fit scoring, and a Chrome extension that fills applications.
An outreach CRM for agencies running large Instagram pages: discovery, fit-scoring, AI-drafted DMs, meetings and deals, without tripping Meta's automation detection.
Pictimate went from first commit to a paying contractor in under eight weeks. Every product launched with billing wired in from day one.
ReachGram drafts every DM, but a human presses send. BossPal's customer tier never receives the pricing tool. DojoPrep fills applications but never submits. Those lines are product decisions, and they are why the products survive contact with Meta, TCPA and ATS terms.
A BossPal demo died mid-booking when the agent ran out of tool turns and sent nothing. The customer got dead air and assumed the business ghosted them. Since then every failure maps to a customer-visible message, and every background job reports into the channel its output was missing from.
A 301 sq ft garage once quoted $300 less than a 300 sq ft one. The fix made price monotonic by construction, and the test walks every square foot from 100 to 3,000. BossPal's eval harness runs golden and red-team suites against the real pipeline and blocks merge on regression.