Case Study
AI Marketing Assistant for Tjamigo
AI-powered lead generation assistant for Tjamigo (Stockholm event platform) that automated Reddit prospect sourcing to support founder outreach and customer acquisition.
Visuals
Problem
Tjamigo, an early-stage Stockholm event platform, needed to grow its user base but had no structured outreach pipeline. The founder was manually scrolling Reddit looking for potential users, slow, inconsistent, and unscalable. There was no way to prioritise who to contact or track what had already been tried.
Solution
Built an end-to-end lead generation pipeline: a Reddit scraper collects posts and comments matching event-related intent signals, a LangGraph classification agent scores each prospect by relevance and purchase intent using Groq's fast inference, FAISS deduplicates against previously seen leads, and a React dashboard surfaces ranked prospects with suggested outreach copy for the founder to review and send.
Impact
- 100+ qualified Reddit leads generated in the first run, ready for founder outreach
- Automated classify-and-rank pipeline reduced manual prospecting from hours to minutes per session
- FAISS deduplication prevented re-contacting the same prospects across sessions
What I Learned
- Groq's inference speed made real-time classification feel instant, for pipelines scoring hundreds of posts quickly, latency matters more than marginal quality gains from larger models
- Reddit's API rate limits are the real bottleneck: batching requests and caching raw responses before classification was essential
- Showing ranked prospects with suggested copy rather than raw leads dramatically improved adoption. The tool needed to fit the founder's workflow, not create a new one