One sales operating system

A simpler morning routine built around one Kanban card, reliable history, and deliberate sending

Master summary — the gist in 30 seconds

TL;DRKeep the strong pipeline foundation, remove competing daily workflows, and make every normal lead action happen once on the same card.

Input Email, call, booking, Calendar, transcript, or owner decision Output One event, one complete history, and matching Notion plus dashboard cards.

Why this mattersMatt starts every morning with a complete view of who needs a reply, follow-up, callback, or prepared call. He gets leverage from preparation and synchronization without giving away control of sales email.

flowchart LR
 IN["Email · call · booking<br/>Calendar · Fireflies"] --> EVENT["Record the event once"]
 EVENT --> HISTORY["One complete history"]
 HISTORY --> DASH["Dashboard Kanban"]
 HISTORY --> NOTION["Notion Kanban"]
 HISTORY --> REPORT["03:00 report"]

1 · Your morning stays on the Kanban

TL;DRThe existing Pipeline becomes the morning view. A separate Today, Focus, or Calls workspace is no longer part of the routine.

You see Clearly distinct replies, follow-ups, callbacks, calls today, and issues while keeping every lead in stage context. You do Use the phone link, review or send a draft, archive without reply, log a call, move stage, or open history on the card.

Why it mattersThere is no need to rebuild the same work across Pipeline, Morning Review, Missive, and Notion.

flowchart TD
 OPEN["Open Pipeline<br/>Morning mode"] --> SCAN["Scan distinct action states"]
 SCAN --> CARD["Open the next lead card"]
 CARD --> CALL["Use phone link"]
 CARD --> SEND["Review and send"]
 CARD --> ARCH["Archive without reply"]
 CARD --> LOG["Log call"]
 CARD --> HIST["History"]

2 · One call flow and one booking engine

TL;DRLog Call opens notes first, then Lost, Book appointment, or Call back later. No answer reveals the ready email fallback.

Call result Notes or transcript plus one outcome System result Stage, callback, Calendar, Meet, reminders, history, and both cards update from one command.

Why it mattersA callback date replaces overlapping wait and follow-up controls. Dashboard booking and the public booking page cannot create competing appointment truths.

flowchart LR
 CALL["Log call"] --> LOST["Lost"]
 CALL --> BOOK["Book appointment"]
 CALL --> BACK["Call back later"]
 CALL --> NOANS["No answer"]
 BOOK --> GCAL["Calendar + Meet"]
 BACK --> SUPPRESS["Suppress drafts<br/>until due"]
 NOANS --> DRAFT["Prepared manual email"]

3 · Automation prepares, Matt decides

TL;DROnly booking confirmations, reschedule confirmations, and call reminders can send automatically.

Automation Reconcile, draft, prepare the sales call, match Fireflies, save Figma, and explain issues Matt Sends every ordinary sales and proposal email.

Why it mattersThe system creates leverage without repeating the past risk of an unwanted automatic sales reply.

flowchart TD
 BOOKED["Booked stage"] --> PREP["Codex sales prep"]
 PREP --> FIGMA["Figma on both cards"]
 CALLDONE["Sales call done"] --> FF["Fireflies evidence"]
 FF --> SKILL["Matt runs proposal-contract"]
 SKILL --> REVIEW["Review package and draft"]
 REVIEW --> MANUAL["Matt sends manually"]

4 · The plan is split into safe execution waves

TL;DRThe interrupted cloud branch is preserved as evidence, then the build proceeds through seven gated plans.

First Rescue, cleanup, and protect production fixes Then Truth and history, card UX, calls and booking, Fireflies and prep, the 03:00 brain, and final real-account QA.

Why it mattersThis prevents a large stale branch from silently reintroducing defects and makes each slice independently provable.

flowchart LR
 P0["00 Rescue"] --> P1["01 Truth"] --> P2["02 Card UX"] --> P3["03 Booking"]
 P3 --> P4["04 Fireflies + prep"] --> P5["05 03:00 brain"] --> P6["06 First morning"]

⏭️ Next steps

TL;DRVersion B is selected. A fresh planning task turns the locked intent bundle into a technical checklist before implementation starts.

The next planner reads the handoff in order, resolves the event model, Notion action shape, Calendar permissions, and Codex worker dispatch, then writes small test-first implementation commits.

Why it mattersNo owner decision is currently blocking technical planning. The next instance should resolve technical choices autonomously without reopening the selected operator routine.

💡 Fun facts and practical stuff

TL;DRThe history audit found the product intent was consistent. The implementation became complicated because several experiments survived at once.

The private extractor scanned 1,962 Claude JSONL files and found 10,023 unique visible messages. The current app already has nearly every required primitive. The cloud run is stopped and preserved at efbd2424. The final 03:00 acceptance will be manually triggered after QA instead of waiting for the clock.

Why it mattersThis is a consolidation and reliability program, not a blank-slate rebuild.

Full gigaplan →Operator SOP →Technical handoff →Prototype →