How it works
Working software in week one. Yours on day thirty.
A four-week engagement, structured to ship. No discovery phase that drags into month three. No 40-page requirements doc. Real software, in production, by the end of the month.
The four weeks
One phase per week. No drift.
Week 1
01
Discovery + foundation
A 30-minute call to figure out what's actually broken. Not a workshop. Not a survey. A conversation with whoever runs the workflow, and we leave with a one-page scope. By Friday: a working dashboard you can log into. Real schema, real data model, your branding. Not a Figma file.
You give me
30 minutes for discovery, your existing data, brand colors and logo.
You get
Signed scope, deployed dashboard skeleton at a private URL.
Weeks 2–3
02
Iterate + deepen
Two weeks of building the actual workflow. Claude integration, MCP tools, automations, the parts of the loop that produce the unlock. Weekly demos every Friday. You see the build advance in real time, redirect if something's off, kill features that turned out not to matter.
You give me
One 30-minute review per week. Slack/email turnaround on questions inside 24 hours.
You get
Working features shipping every week. Honest weekly notes on what's done, what's next, what's at risk.
Week 4
03
Ship + handoff
Production deploy on infrastructure you own. Code transferred to your GitHub. Database in your Supabase. Domain pointed at your Vercel. Documentation written for whoever maintains it after me. If we're staying in touch via the retainer, week 4 also locks the next month's roadmap.
You give me
Access to your hosting accounts (or a 20-minute call to set them up).
You get
Production system on your infra, all code, all data, all credentials, and a Loom walking through the architecture.
On day 30
What's actually in your hands.
No ambiguity. If we agreed to it on day one, it ships by day thirty.
- ✓
A production web application running on your Vercel project
- ✓
A Postgres database in your Supabase project, with all your data inside
- ✓
A custom domain (or subdomain) pointed at the app
- ✓
The full source code in your GitHub organization
- ✓
Any MCP servers, cron jobs, or Cloudflare workers we built — deployed to your accounts
- ✓
A written architecture overview + a Loom walkthrough
- ✓
A list of every external service the system depends on, with monthly cost estimates
You are not locked into me. If you fired me on day 31, the system would keep running. That's the test.
Pricing
Scoped per engagement. No tiers. No 'contact sales.'
Internal dashboards, customer-facing portals, and AI agent systems all carry different scope. We figure out the right number on the discovery call and you get a fixed-price proposal before any work starts.
The build
Fixed price.
Scoped on the discovery call against your actual workflow. 50% upfront, 50% on handoff. No hourly billing. No "out of scope" surprises — if we agreed to it on day one, it ships on time.
The retainer (optional)
Month-to-month.
New features, new MCP tools, new tenants, expansions, integrations. The retainer buys my time on a steady cadence — not access to your software. The code, data, and infra are yours from handoff. Cancel and your tool keeps running exactly as it did the day before.
You can also skip the retainer entirely and come back ad-hoc when you want something built. Up to you.
What this is not
For buyers who've been burned.
Not a discovery phase that drags.
Discovery is one call. The scope is one page. We start building in week one or this isn't the right engagement.
Not a slide deck.
No "strategic recommendations." The deliverable is software running in production with users using it.
Not a long-term lock-in.
You own the code from day 30. The retainer is month-to-month and cancel-anytime.
Not an AI demo.
Claude does real work in the loop — calling tools, updating your database, drafting outreach. Not a chat widget glued to your homepage.
Ready to scope a build?
Tell me what's broken about your team's workflow. I'll respond within a day with whether I think this is a fit.