New from MSOFlow — early access

AI project management, run from the files you already have.

Your project management tool is a second job. Delivery & Ops Automation hands that job to the AI your team already uses — BYOAi: Claude, Copilot, or ChatGPT. The board stays current, due items surface themselves, and the status update writes itself.

Built and run daily on our own multi-entity operation — every feature is self-tested before it ships.

The reporting tax is real

Daily huddles and status meetings that relay what's already written down. Task boards that are only as current as their last manual update. Project updates reassembled from meeting notes, AI summaries, and task lists that already say the same thing. For anyone running multiple projects, entities, or engagements, the administrative overhead compounds.

How it works

Infographic: three different AI assistants used by different people in different places — Claude used by a PM at HQ, Copilot used by an engineer working remote, and ChatGPT used by a consultant at a client site — all connect into one shared pipe that fans out to the same system of record: the TASKS.md board of record, an append-only decisions.md log, and a dated status history. The files render into a kanban board with To do, Doing, and Done columns, and from the board come live custom dashboards, a daily brief listing what is overdue, due, or blocked, and a drafted status update.

BYOAi — any assistant One system of record C Claude PM — HQ Co Copilot Engineer — remote G ChatGPT Consultant — client site TASKS.md the board of record decisions.md append-only log status/ dated done history Board — renders from files To do Doing Done Live custom dashboards rendered from the board — always current Daily brief overdue · due · blocked — only what needs you Status update drafted from done history — not from memory

Files are the system of record

A simple, open structure — board, decision log, status history — that both people and AI agents read and write. Boards render from the files, never the reverse.

Bring your own AI (BYOAi)

Claude, Copilot, ChatGPT — open conventions mean each person’s assistant follows the same rules. No new tool for your team to learn.

Updates generate themselves

A daily brief flags only what’s overdue, due, or blocked. Status updates draft themselves from the done history. Monthly review decks render straight from source.

Built for

Executives

Run a company top-down: one board of record, every workstream visible, a morning brief that only speaks up when something needs you.

Consulting firms

Bill delivery, not status administration. Per-engagement project records with client-ready status packs generated from the work itself.

PE operating teams & multi-entity operators

One system across the portfolio. Decision logs and dated completion history that hold up in diligence — from day one.

Why trust it

Built and run daily by a CPA with nearly two decades consulting to and leading multi-entity companies — on a live multi-entity operation. The approach isn't ours alone: it extends the MIT-licensed Interpretable Context Methodology published by Van Clief & McDermott, so you can read the research behind it, not just our claims. Records are structured for auditability from day one: append-only decisions, dated completion history, everything tied to source. Read the full write-up — what actually runs, what it's built on, and where it breaks.

  • Append-only decision log — supersede, never rewrite history
  • Dated done-history feeds every generated status update
  • Extends the published Interpretable Context Methodology (ICM) by Van Clief & McDermott — read the research at arXiv:2603.16021, or watch Van Clief build multiplayer AI on the same foundations
  • Open file conventions — no lock-in, your data stays yours

Working systems that tie to source

1 board of record — everything else renders from it
0 retyped status updates — generated from done history
Any AI assistant — open conventions, no new tool
CPA-built and run daily on our own operation

Early access is open

Get the build-in-public write-up and be first in line as the early-access cohort forms.