For marketing agencies and in-house performance teams

A performance marketing agent that does the boring bits

Gene plugs into your ad accounts and lives where your team already works. Ask it questions. Let it run the Monday report while you do the thinking.

The problem

Reporting eats Mondays

Every Monday, someone is rebuilding the same deck. Pulling numbers from Google Ads, Meta, GA4, TikTok. Joining them in a spreadsheet. Pasting them into a template that hasn't changed in a year. Hours of it, every week.

Every question means a login

"What did we spend yesterday?" "Which ads dropped this week?" Each one means logging into a platform, finding the right view, copying the answer back into Slack. The data exists. The friction is in getting to it.

The same work, every client

Set up reporting for one client. Then do it for the next one. And the next ten. The conventions, templates and filter rules live in shared docs and the head of whoever's been there longest. There's nowhere they can actually live.

Everyone's tooling is different

One person uses ChatGPT, another uses Claude, a third has a clever spreadsheet nobody else understands. The team has no shared way of working. Conventions one person figures out never reach the rest.

Works where you're already working

Conversational queries

Anything you'd otherwise click around a platform UI for. "Top 5 ads by CTR in the UK campaign last week." "Spend by channel yesterday." "Which audiences are fatiguing on Meta." Gene picks the right data source and answers.

Reports arrive without being asked

The point of a Monday report isn't that someone asks for it. It's that it's there on Monday morning. Hand Gene a Slides template and a schedule once. From then on it clones the deck, fills it with last week's data, and drops it into the channel before anyone's finished their coffee. No reminders. No missed weeks.

Not trying to reinvent your team

Gene is a data-access layer. It makes your team — and the AI tools they already use — more effective, not redundant.

Your edge is your people and their judgement. Gene takes the boring data-wrangling off their plate so they spend their time on the parts that actually matter.

Same for the tools they've already adopted. The ChatGPTs, Claudes and custom agents someone on the team built last month — Gene doesn't replace them, it feeds them. Anything that speaks MCP can pull clean, brand-aware numbers from Gene, so the agent your team built last month suddenly has reliable performance data behind it.

Knows your clients

Gene is configured around the brands you actually work on.

You set up each brand once. The ad accounts that belong to it. The campaigns that count. The timezone, the currency, the way your team names things. From then on, every query is brand-aware. Ask "how is Acme doing this week?" and Gene already knows which accounts to hit, which campaigns to include, and what "this week" means in your timezone.

One ad account often serves several brands. A single Meta account holding three clients' campaigns is normal in agency life. Gene splits the data using filter rules you define (say, campaigns whose names start with ACME_), so one client's numbers never bleed into another's report.

Built for teams, not individuals

AI tooling is fragmenting fast. Gene is the centralised performance marketing brain your whole team shares.

Every account manager has their own ChatGPT, their own Claude, their own MCPs, their own prompts in a doc somewhere. The conventions one person figures out don't reach the rest of the team. New hires start from zero.

Gene is the opposite. One agent for the whole team. Same brand context, same conventions, same definitions of what a good report looks like. Getting better every time you teach it something.

Why not just plug a few MCPs into Claude?

You could install a Google Ads MCP, a Meta MCP and a GA4 MCP into Claude Desktop and reach your data that way. Here's why Gene is different.

Generic MCPs don't know your clients. Every query starts from zero. You specify accounts, campaigns and date ranges by hand, every time. Gene already knows.
They don't share context with each other. There's no normalised view of "campaign" between Google and Meta, so blended cross-platform answers are hard. Gene maintains the schema that lets data join cleanly.
They don't run on a schedule. Reports happen when someone remembers to ask. Gene runs them on Monday at 8am, every week, whether anyone remembers or not.
They don't enforce team-wide rules. If your agency always excludes branded search, or filters out test campaigns, every teammate has to remember to apply that. Gene applies it once at config time.
They don't handle the platform approval work. Google Ads, Meta and TikTok each have their own developer access process. Each MCP makes you do that yourself, per team.

MCPs are tools. Gene is the layer on top of them. The part that knows who you are, who your clients are, and how your team works.

Where Gene is today

Gene is in early development. The architecture is designed, the first phase is being built, no users yet.

Gene is a project of Trampoline.