All pathsFlow 4.0 path / Every client gets it

Share a Playbook that can keep getting better

We brought a public AI Consulting starter into Flow, read the license before adding it, and opened its files locally. The copy can keep following the source while we work in it.

Written by
Manav Sehgal
Published
Reading time
2 min

Try this path ↓

New from GitHub preview for the AI Consulting Starter showing six files, the version, and CC BY 4.0

A Flow Playbook lets someone use your method from a folder they can work in. Share the brief, checklists and examples together. Readers can bring the folder into Flow and choose which upstream changes to apply on later updates.

“I’ve improved the kickoff checklist. My team should be able to take that update and keep working in their own copy.” Illustrative scenario.

Open a published Playbook

  1. In Flow, choose File ▸ New from GitHub…. Paste the public folder address and press Read Folder.
  2. Read the file count, version and license in the preview. Choose where to put the copy, then press Add. Flow opens the README.
  3. Use Check for Updates… later to inspect upstream changes. Select the files you want and press Update Selected.

Try the AI Consulting Starter on GitHub. To see how a full Playbook is made from an empty folder, read how AI Consulting was built in Flow.

Read the method where you’ll use it

The starter includes an explanation of working backwards, a mechanisms primer, and material for taking an AI pilot to scale. Once added, the files are available in the sidebar. The Mechanisms Primer opens as a document you can read in place.

The imported starter expanded in the Folds sidebar with Mechanisms Primer open in the Reader

To share your own method, choose Publish Folder… ▸ GitHub Repository. Confirm the destination and the files before publishing. Keep personal preferences in your personal Fold and put the reusable method in the Playbook folder you intend to share.

Evidence

What we checkedResult
Import previewVerified. Six files, about 75 KB, and a CC BY 4.0 license appeared before Add.
Imported filesVerified. All six Markdown files matched the published files byte for byte.
Model costVerified. The public import called no model.
Publisher and selected updatesVerified in a separate release review. The shots show the public reader import.

Walked on: Flow 4.0.0 (4383), isolated arm64 Release bundle from 0a584a48, macOS 26.6, 1512 × 949 pt, walked by Codex at 21:30–21:35 PST.

Data: Orionfold’s published AI Consulting Starter at orionfold/flow

Written by the Flow product team from the walk's own records. Numbers are labelled verified, derived, assumed or unknown.

MORE PATHS

Another job,
walked the same way.

All paths →
One brief, three ways, captured during the walkthrough

Flow · Canvas

One brief, three ways

Consultants and analysts

Configure Jobs, ask Flow Chat, then trace the result through Connections, Evidence and History. Two routes to one brief, with sources, time and cost recorded.

  1. Configure the brief’s Jobs
  2. Ask Chat to refresh the same section
  3. Trace the result on the canvas

71 sJobs run to review ready

Diagrams that draw anywhere, captured during the walkthrough

Flow · Diagrams

Diagrams that draw anywhere

Writers and technical teams

Write a diagram as text. Read it in either appearance, copy it as a picture, and carry it into the PDF preview.

  1. Copy a starter from Charts Gallery
  2. Read it or copy it as a picture
  3. Preview the drawings in a PDF

20 / 20Guide diagram forms drawn

A checklist you can sign, captured during the walkthrough

Flow · Sign-off

A checklist you can sign, inside the document

Consultants and small agencies

Tick the deliverables, confirm a name and keep an acceptance record beside the document. Draft another checklist from selected text.

  1. Tick the checklist and type a name
  2. Confirm what Flow will record
  3. Find the acceptance in History

1 recordwritten by one sign-off