Three rundowns.
Short files on the work that's shipped out of this studio. What the problem was, what I touched, what moved. Written end-to-end by the operator. No case-study writer, no embellishment.
100% call answer rate in fourteen days.
The problem
A third-generation roofing company, mid seven figures a year, losing revenue to a simple bottleneck: they weren't answering the phone. Leads from ads, door-hangers, Google, all landing in the same inbound line, routed to a rotating cast of part-timers. On any given week, forty to sixty percent of calls went to voicemail, and voicemail conversion was effectively zero.
The owner knew it was happening. He didn't know how many jobs it was costing him. He'd hired three "AI phone" companies in twelve months, all of them promising "automated receptionists" that turned out to be bad voice bots with no ops layer behind them.
What I did
- Ran the Rundown across inbound flow, ad attribution, booking software, and dispatch. Eight-page memo in week one.
- Built a voice agent that answers every call inside two rings, qualifies the lead, and hands warm jobs to the dispatcher over SMS.
- Wired a real attribution pipeline (UTM + call → job → invoice) into a single operator dashboard.
- Rebuilt the intake script with the owner, not around him. Nothing templated.
- Ran it alongside the existing team for three weeks before switching over.
The outcome
By day fourteen, the call answer rate was one hundred percent, the first time in the company's thirty-year history. Inside ninety days we'd closed a $100K+/month pipeline leak: jobs that previously went to voicemail, now booked — a chunk of them full roof replacements. The owner kept the dispatcher and his lead setter. The agent handled the volume they were drowning in, not the work they were good at.
He calls it "the phone thing." That's fine.
Brand-deal triage, four hours down to twenty.
The problem
A seven-million-follower creator with an inbox of three hundred brand pitches a month. Every one of them looked like a friendly intro and a PDF. Every one of them had terms, clauses, rates, deliverables, and deadlines buried inside. Her team of one manager and one assistant was spending the equivalent of three full workdays a week just reading and sorting, before a single yes or no.
She wasn't taking worse deals. She was missing the good ones underneath the noise.
What I did
- Rundown across the inbound flow: what came in, where it lived, who touched it, how decisions were made. Mapped the real bottleneck (it wasn't the inbox).
- Built a triage pipeline that reads every inbound pitch, extracts the contract terms into a structured record, and scores it against her stated preferences.
- High-fit deals surface in a daily digest, with the numbers she cares about surfaced, and the unacceptable terms already flagged in red.
- Rest are auto-declined with a warm, on-brand template she approved once.
- Manager stays in the loop, but not in the grind.
The outcome
Triage time collapsed from four hours per deal to twenty minutes. Her manager got twelve hours a week back. More quietly: three deals closed in the first month that previously would have timed out in the queue. She's taken fewer deals, and made more money from them.
A studio of one that runs itself.
The problem
I wanted to work alone. Not because I dislike teams. I've run them. I wanted to work alone because the best work I've ever made, I made in a room with nobody else in it. The trouble is that solo doesn't scale. An operator at a studio of one runs out of hours before he runs out of ideas.
I spent a year asking whether a small team of agents could give me the output of a team without the overhead of one. The answer is yes, but not in the way the demos suggest.
What I did
- Built eight agents, one at a time, each for a real role I was already doing. Not for a role I imagined.
- Named them, scoped them narrowly, gave each its own system prompt, tools, and a single owner (me).
- Wrote a coordinator (Yuri) that routes work between them. Wrote a checker (Suki) that stops them from doing anything I didn't ask for.
- Everything lives in-context: my notes, my clients, my calendar, my codebase. Nothing talks to the internet that shouldn't.
- Documented the whole thing publicly, in the Field Notes, as I went.
The outcome
I ship roughly three times what I shipped as a solo operator. More importantly, I ship the right things, because the team handles the work I'd otherwise skip. The studio now delivers at a cadence I used to need two hires for, at a margin that lets me keep the roster narrow. Two clients per quarter, by choice.
The team is not the product. The operator is. The team just makes the operator finishable.
Every engagement starts with one week.
If the work above reads like what you need, or what you've been trying to get out of three vendors at once, the Rundown is the door.