Changelog

What shipped.
What it cost. What it earned.

Orbit’s own brief, written in the same voice it uses on Mondays. Newest at the top.

Outbound engine and voice corrections.

The autonomous CMO can’t punt the #1 growth lever to Apollo and Instantly. So we shipped Outbound — cold email and LinkedIn sequencing in the founder’s voice, replies drafted in the founder’s inbox, every step tagged for the experiment loop.

  • Outbound module — sequence builder, voice-scored steps, signal-driven lists.
  • Reply drafter — three-line responses in voice, paired with CRM update.
  • Voice corrections loop — founder edits feed the voice model, not the prompt.
  • ReadyPermit pilot live. moat-hook-v3 · +22% reply.
Pipeline motion. ReadyPermit reply rate 18% · two meetings booked from sequence step 2 · zero manual sends this week.

Opportunity Radar with drafted assets.

Watching the right window isn’t enough — the asset has to be drafted by the time the founder reads the signal. Radar now ships with a per-signal draft attached.

  • Sources · Product Hunt, HN, X, LinkedIn, news cycles, conference calendars.
  • Per-signal draft · response post, ad variant, comment hook generated with the signal, not after.
  • Decay · time-to-ship stamped on every signal. Three-hour-old reads as cold.
  • ICP overlay · per-company filter so noise doesn’t outrank a real moment.

Ad engine writes back to Stripe and Meta.

Recommendation isn’t the bottleneck — execution is. Ad engine now runs the kill criteria and the budget rebalance on its own, within the founder’s ceiling.

  • Closed-loop budgets — daily rebalance across Meta, Google, LinkedIn, X.
  • Kill criteria — pre-stated, triggered automatically, logged in the brief.
  • Pricing test → Stripe — winner on the upgrade page is written, not staged.
  • Funnel attribution — ROAS tied to activation, not clicks.
Northstar · moat-hook-v3 scaled +35% · ROAS 3.2 at cohort day 30 · CAC -12% MoM.

Direction console and content factory.

The first slice. One screen for the founder. Long-form in, channel-tuned out. Voice fidelity is the bar — below 70 won’t ship.

  • Direction console — objective, brief, decision drawer, heartbeat.
  • Content factory — LinkedIn, X, blog, shorts. Voice score per draft.
  • Voice model — trained from Fireflies and writing samples per founder.
  • Brief generator — plan vs. actual, top win, top loss, what I changed, what I’m testing, what I need from you.

The bet. Written down.

Spec committed. Seven modules drawn. Mars-red picked. First company connected. The system has a name.

Read the next one on Monday.

Onboard a company today. The first brief lands at 06:30 PT.