The problem

At my AI studio, our CEO and sales team needed a better way to follow up after discovery calls. They wanted to use the buyer's exact words from the call, understand what the prospect's company was actually working on, and tailor our pitch to fit their needs.

After every discovery or demo call, reps were spending 30 to 45 minutes writing long follow-up emails and attaching static PDFs. When our main contact forwarded that long email thread to their boss, half the context got lost and deals stalled.


What we built

We built Meeple during the early days of AI, before modern agent skills or MCP tools existed. That meant I had to build the whole harness from scratch: scraping company data, picking out the most important signals, running the LLM prompts, and mapping the output into a custom follow-up page for each buyer.

Step What it does
Pulling the buyer's exact words Reads the discovery call transcript to find the buyer's role, whether they are the sole decision-maker, their main pain points in their own words, and the next steps we agreed on.
Scraping & filtering company data Scrapes company info and web signals, then filters down to what the prospect's business is focused on right now so the sales pitch matches their priorities.
Mapping LLM output to a custom page Runs the call transcript and company context through a custom prompt harness that fills in a clean buyer page: a personal intro, their challenges in their own words, a tailored business case, and a shared action plan.
One-click case studies & deal tracking Lets reps attach relevant case studies, pricing, and timelines in one click, and shows when the buyer's team views the page as the deal moves from Discovery to Evaluation.

The result

  • Closed 5% more deals: Tailoring every follow-up around the buyer's own words and company priorities helped move more prospects from discovery into evaluation and lifted our deal close rate by 5%.
  • 45 minutes down to 30 seconds: Reps went from spending 45 minutes writing recap emails to reviewing and sending a custom buyer page right after hanging up.
  • Built before agent skills existed: Building our own scraping, filtering, and UI harness early on shaped how I build agent skills and GTM tools today.