Changelog

Everything we
changed.

Dated, described in plain language, and honest about the fixes. One person builds this, so the list is short enough to read end to end.

33 changes9 of them fixessince 10 Jul 2026The longer story →

Standard pricing is live.

The founding-access offer has run its course. Pricing is now Starter, Growth, and Agency, the same three tiers on the pricing page for good.

  • ChangedFounding access (the one-time €199 offer) is retired. New accounts choose from Starter (€199/mo), Growth (€499/mo), or Agency (€999/mo), billed monthly or annually.
  • ChangedAll three tiers include the same seven agents. Only the seat count, lead cap, and how much autonomy you hand the agents change from tier to tier.

Claire has to show her work.

The learning loop got an audit. One defect had been feeding Claire her own handwriting as evidence, so we cut that path and put a human-set boundary around everything she is allowed to conclude.

  • FixedClaire counted her own logged notes as customer replies and trained her scoring on them. A reply now has to arrive from outside your team before it counts as one.
  • FixedSends and replies were measured over different windows, which quietly inflated reply rates. Both now use the same window.
  • NewEvery value Claire learns sits inside a range a human approved. She can move within it and cannot step past it.
  • NewEach action and proposal carries the version of Claire that produced it, so a change in her judgement is traceable to a release.
  • NewUndo and skip rates are recorded outside Claire's reach. She cannot mark her own homework.
  • FixedFree-text skip reasons could travel past the workspace that wrote them. They stay put now.
  • FixedCalls, emails, and meetings logged in HubSpot never reached your activity history. The sync reads them properly, and reports honest counts instead of rounding up.

Correct her once, not every week.

Until now the only way to disagree with the agent was to undo what it did. You can tell her instead, and the instruction sticks.

  • NewThe agent has a name. She is Claire, and she is one colleague with one memory rather than seven services with seven opinions.
  • NewStanding instructions. Tell Claire to leave a client alone until October and she will, without being told again.
  • NewClaire tells a question from an order. Asking what she thinks about a deal no longer books work you did not want.
  • NewThe address bar answers back. Put a question in the URL, get the answer on the page.
  • NewA daily ceiling on model spend, and a plain sentence when the model is unreachable rather than a spinner that never resolves.
  • NewApproved drafts can be lifted out and sent from your own mailbox.

Eleven destinations became four.

The app had grown a page per feature, which meant your morning started with a navigation decision. It starts with the work now.

  • ChangedEleven pages collapsed into four: Workspace, Pipeline, Analytics, Settings. The tabs the workspace replaced are gone rather than hidden.
  • NewThe morning ledger. Today's work in the order it should happen, with the reason attached to each line.
  • NewA durable work queue behind it, so an overnight run survives a restart instead of starting over.
  • NewClaire studies how you actually write and drafts closer to your voice with each approval.
  • NewRetained search bills in thirds. The forecast models it that way, keyed to the stages your pipeline really has.
  • NewThe autonomy flip. Claire can act alone on the specific classes of work you have signed off, and nothing else.
  • NewSetup introduces a colleague rather than a scoring tool.

We took the claims we could not back off the site.

A marketing pass that mostly deleted things. What is left is what we can show you in a demo.

  • FixedRemoved the invented statistics and the meta description that promised more than the product does.
  • Fixedrobots.txt, the sitemap, and the unsubscribe page were unreachable to search crawlers and to anyone clicking unsubscribe in an email.
  • FixedThe hero video redirected logged-out visitors instead of playing.
  • ChangedThe app's design system now speaks the same language as this site.

Built for executive search, specifically.

Trying to serve every agency produced advice too general to act on. We picked the vertical we understand best and rebuilt the intelligence around its economics.

  • ChangedLeak detection understands retainers, shortlist stalls, and the gap between a placement and an invoice.
  • ChangedThe pipeline page reads and acts rather than reproducing the CRM you already pay for.
  • ChangedFounding access replaced the free trial: one payment of €199, which grants Growth-tier access with no seat counting on day one.

Live writes prove themselves.

Approving a draft used to end in silence. Now the approval tells you where the work landed and gives you the link.

  • NewReal sending through your connected Gmail or Outlook mailbox.
  • NewLive mode mirrors approved writes into HubSpot and deep-links to the record it touched.
  • FixedLive HubSpot execution could not be reached from the interface at all.
  • FixedDuplicate proposals stacked up when a run overlapped the previous one.

Older work predates this page. It lives in the blog, where the reasoning gets more room than a line item allows.

Ready to delegate
RevOps to AI?

Plans from €199/month · 7-day trial · cancel anytime