How it works

Brief. Build. Ship.

One founder, one coordinator per business, specialists for the work, and one operator keeping it honest. Every piece of work has a named owner, a success signal, and a Multica issue.

The loop

Five steps. No back-channels.

The founder talks to one coordinator. The coordinator routes the work. Specialists never close their own issues. The operator records everything.

  1. 1

    Brief.

    Jono Horne

    The founder names the commercial outcome — a campaign, a test, a feature, a decision. Plain English. No deck.

  2. 2

    Shape.

    Dan

    The coordinator (Dan for Defendwise, Bruce for Venture) turns intent into a Multica issue with an assignee, a success signal, an approval gate, and a deadline.

  3. 3

    Build.

    Lisa

    Specialist agents pick up their lanes — Lisa finds the signals, Rach drafts the article, Picasso creates the blog image, and Woz wires the page and publish path. Each one attaches the artifact and signals done inside the issue.

  4. 4

    Review.

    Dan

    The coordinator reads against the brief, checks copy, image, and publish readiness, applies the universal test ("would a smart MSP owner respect this?"), and either closes the issue or packages it for founder review.

  5. 5

    Ship.

    Woz

    After approval, Woz publishes the page or post, the coordinator closes the issue, and the operator (Bob) keeps the audit trail honest — every scheduled run, every decision, every artifact.

A real day

Watch one piece of work move.

Here's what shipping a Defendwise blog post actually looks like across the AI team: research, copy, image, approval, publish. Pulled from the same operating pattern the pod uses in Multica.

  1. 06:30Lisa

    Lisa researches the signal first: fresh security news, threat patterns, MSP pain, source links, and the angle worth writing about.

  2. 07:00Rach

    Rach turns Lisa's signal brief into a blog draft: headline, article body, pull quotes, and social snippets attached to the Multica issue.

  3. 07:30Picasso

    Picasso picks up the same issue and creates the blog image: concept, prompt, final visual, and export notes matched to the article.

  4. 08:00Dan

    Dan reviews copy and image against the brief, chooses the strongest version, and packages the approval request for Jono.

  5. 08:05Bob

    Bob sends the approval request to Telegram. Jono reads on his phone and replies "go" or asks for a tweak — in under two minutes.

  6. 08:20Woz

    Woz publishes the approved post: page wired, blog image placed, metadata checked, preview loaded, then pushed live.

  7. All dayBob

    Bob keeps the audit trail current. Anyone reading the issue tomorrow can trace what shipped, who approved it, when it ran, and what came back.

What this isn't

A few things to know.

Not a chatbot

Every agent is a specialist with a defined role, a workflow, tools, and an audit trail. Not a general-purpose chat window.

Not autonomous

Founder approval is required for anything public, paid, or irreversible. The team drafts and packages — the founder decides.

Not invisible

Every scheduled run, every artifact, every decision is logged in Multica or the OpenClaw audit trail. Nothing happens off the record.

See the AI team for yourself.

Meet the teamBrowse the toolsWhat shipped this week
LoadingChecking the team…