One CPA's Bottleneck Becomes a Company Brain

He ran a live Hot Seat that turned one CPA's problem list into a 7-day plan

Austin Distel

The problemA five-person CPA firm routes every client and team question through one person's brain, and there's no way to see which fix to build first.

Austin put AZ Moyer of AZ Does Taxes in the hot seat: a 30-minute problem dump, then a pain-frequency matrix plotting each bottleneck by how often it hits and how much it hurts. AZ named three: daily client and team FAQs, monthly tax-code research, and the general drag of being the only person clients trust to answer a quick question. Austin's question cut through it — are there dependencies that make everything else easier or unnecessary? AZ picked team FAQs first, even though client FAQs hurt more, because unblocking the team frees his time for everything else. The room then solution-brainstormed onto signed Post-it notes, sorted the ideas on an effort-impact matrix, and Austin's Hot Seat skill turned the winners into a Notion Kanban board — a seven-day roadmap AZ could hand straight to his team.

You are my AI chief of staff running a Hot Seat session from the Agentic Society Boardroom (May 2026). Interview me in rounds: (1) a problem dump — ask what's blocked, broken, or endlessly repeated in my business; (2) score each problem by frequency (daily to yearly) and pain, and tell me which is most frequent AND most painful; (3) ask whether fixing one problem would make the others easier or unnecessary, and recommend that one even if it isn't the most painful; (4) brainstorm every fix for it across hire/buy/build without judging; (5) score the fixes on an effort-impact matrix and recommend the quick win. Output a 7-day action plan as a markdown Kanban board with owners and due dates. Push back if I try to tackle more than one problem at once.

Takeaway Fix the domino that frees your team first, even when a louder problem is more painful — it's the one that makes every other fix easier or unnecessary.

One rough job-site photo becomes a full content set — and one client call becomes a podcast clip

Nick Blancett

One photo → 9 shots + hero image + promo video per job

The problemA field crew's rough before/after phone photos and long client calls never turn into usable marketing content — there isn't time to produce either by hand.

Nick showed two pipelines built for his construction-industry clients. The first takes a contractor's rough iPhone 'after' photo — usually shot with construction mess still in frame — and produces a clean after image that keeps every piece of real hardware, since the client needs the shower they actually installed, not a generic render. From that one photo his pipeline generates nine additional shots, a hero image, and a promo video, so one field photo becomes a full content set pushed straight to a CDN. The second pipeline turns real client relationships into content: he fed Fathom call recordings into ElevenLabs to simulate a podcast-style conversation with the client, turning long-form calls that already happened into short-form content without a second recording session.

You are my content-multiplication assistant, modeled on the pipeline shared at the Agentic Society Boardroom (May 2026). I have raw material I already produce — job-site photos and recorded client calls — that never becomes marketing content. Interview me for: (1) a sample rough photo and what real details must stay accurate (hardware, materials, branding), (2) a sample client-call transcript, (3) my target platforms and formats. Then: generate a plan to turn one clean photo into a set of derivative shots plus a hero image and promo video, and a plan to turn a call transcript into short-form podcast-style audio via a voice tool. Flag anything that would misrepresent the actual work done — accuracy to the real job matters more than polish.

Takeaway The content is already being created — a photo, a call. The build is turning what already exists into more of it, not producing something new.

He laid out the redaction-first blueprint his own company-brain platform runs on

Isaiah Zimmerman

The problemA company brain that ingests every client call is a liability the moment it surfaces something private in a client-facing answer.

Isaiah, whose own platform Orectic.ai builds agentic runtimes and company-brain data pipelines, laid out the architecture before he had to leave early — and aimed each half at a specific member's business. The redaction rule spoke directly to Erin Kenny's health practice: decide exactly what must be redacted before anything reaches a system that answers client-facing questions, the same PII discipline healthcare already runs on. The client-data architecture spoke to AZ Moyer's firm: a tenant-isolated customer portal where each client's own financial history is walled off, and a managed QuickBooks OAuth connector so the system doesn't refight admin permissions every tax season. His closing habit applied to everyone: over-explain everything on every recorded call, assuming the listener knows nothing, so the transcripts a brain is trained on are usable without extra cleanup later.

You are my company-brain architect, modeled on the redaction-first approach shared at the Agentic Society Boardroom (May 2026). Before ingesting anything, interview me for a redaction list: what counts as private in my business (client PII, financial specifics, health, family, anything else). Then design: (1) a tenant-isolated structure so each client's data stays walled off from every other client, (2) a managed-OAuth connection plan for my source systems instead of one-off logins, (3) a habit checklist for recording calls so transcripts are clean training data (stated purpose, over-explained actions). Do not let any client-facing answer reference another client's information, ever.

Takeaway Redaction is not a cleanup step — it's the first thing you build, before the brain ever answers a client-facing question.