Approach

Every business is different, so we always start with a conversation.

How we work
About

We're a small London team of marketers, and we use AI to go faster.

About us
  • Cam BowenCam BowenCo-founder
  • Ben PowerBen PowerCo-founder
  • ElectraElectra
  • MahirMahir
  • NatashaNatasha
  • RahabaRahaba
  • ZobaerZobaer
Tool · AI

Grok

  • Always-on bots
  • On their own cloud computer
  • Grok models in Cursor's agents

Runs our bots on their own cloud computer, so routine checks happen without anyone's laptop being open.

Why we use it

Why we run a bot fleet

  • Works while the laptop's shut

    The bots run on a cloud computer in Cursor's cloud, not on our laptops, with a browser, files and a terminal, so the checks carry on overnight.

  • Meetings filed for us

    Meeting Intel turns each Fireflies transcript into a client log entry and Twenty tasks, and never files the same meeting twice.

  • Sites watched

    Site Watch records each problem it finds in Twenty, and serious ones go to our uptime alerts first.

  • Silent pushes caught

    A change pushed straight to main, or a build that fails, gets an issue raised across 12 repositories, and pull requests get a check for bugs and secrets.

  • Filed, not just said

    Chat is where we talk. The repository, Twenty or a pull request is where the work is kept.

  • Grok in Cursor's agents

    It's one of the models Cursor's agents can run on for bulk jobs, and we pick the model on every launch.

Grok Bot is our night shift. A small fleet of bots runs on one cloud computer in Cursor's cloud, not on our laptops, so the routine checks happen whether or not anyone's laptop is open. They talk to us in one shared Google Chat space, and anything that matters gets filed in our repository or in Twenty, not left in chat.

The bots

  • Brain takes our questions and routes them. When we post from the editor, it usually picks the question up within a minute.
  • Meeting Intel reads each new Fireflies transcript and files it in the client's activity log and as tasks in Twenty. It checks the Fireflies ID first, so a meeting is never filed twice.
  • Site Watch checks our sites for problems and records each finding in Twenty. Anything serious goes to our uptime alerts first.
  • A push-and-build watcher catches changes pushed straight to the main branch and failed builds across 12 of our repositories, and files an issue. It also reviews pull requests for bugs and secrets. It doesn't merge anything.
  • A second bot posts notes on keeping our git repositories tidy.

Where the work lands

Chat is the live board. The record is the repository, Twenty or a pull request: if it isn't filed there, it didn't happen. Meeting Intel's notes-only pull requests merge themselves, and everything else waits for a person.

Rules the bots work to

  • No emails, client messages, publishing or live-site changes unless someone on the team asks for it in that moment.
  • No launching Cursor cloud agents on their own, and no Claude models through Cursor.
  • No secrets in git, pull requests or chat.

Grok in Cursor's agents

Separately, Grok is one of the models our Cursor agents can run on for bulk coding jobs, next to Cursor's own models. We set the model on every launch, and the thinking and reviewing stays with Claude Code.

As of 10 October 2026.

About the tool

What Grok is

Grok is the AI model family from xAI. We meet it in two places. Grok Bot is Cursor's product for always-on agents, called Bots, each with a name, a job and a memory, working on a cloud computer with a browser, files and a terminal. And Grok models are available in Cursor's agents for bulk coding jobs.

Where next

Start with your numbers.

We'll tell you which tools you need and which connections are missing.

Next step

Bring your numbers.

We'll tell you if the spend makes sense.

  1. A 15-minute callYour goals, your challenges, and whether we're the right fit.
  2. A 1-hour discovery callWe dig into your numbers. Worth a hundred emails.
  3. A concrete planAds, a custom build or both, as a one-off project or ongoing.
15 minutes · No pitch deck · A straight answer