Projects
Automating sales research with AI agents
A team of AI agents finds relevant events and companies, analyzes them and prepares the conversations. Preparing and following up events and trade shows takes over 80% less time.
- AI agents
- LLM workflows
- B2B sales
- Automation
- Period
- 08/2026 – present
- Role
- Support for a Berlin-based stealth startup
- Access
- Public
At a glance
- less time preparing and following up events and trade shows
- 80%+
- languages for every report: German and English
- 2
- the region the event radar covers
- DACH
The problem
Finding the right events by hand takes long, and preparing a single trade show takes longer: who exhibits, which of those companies fit, whom to talk to and about what. Most of that work is research that follows the same steps every time.
How it works
Several agents with different jobs work in a chain, and each one hands its results to the next.
- Event radar A web scan keeps a list of events, trade fairs and business competitions up to date: dates, location, prices, participating companies and whether B2B matchmaking is offered.
- Event preparation For a chosen event, the participating companies are analyzed: business model, how well they fit as a target, the key contacts and what I could offer each company.
- Deep dive Companies that an earlier agent marked as promising get an in-depth analysis of the business opportunity.
- Output One report per company plus a spreadsheet, in German and English, with the top contacts, the reason for each and questions to open the conversation.
Results
- Over 80% less time for the preparation and follow-up of events and trade shows.
- A clearly higher success rate when targeting and qualifying the right potential customers.
- The research is put to the test in practice: we attend trade shows together to win customers.
Tools
Large language models, agent orchestration, prompt engineering, automated web research, spreadsheet reports, and my own Ubuntu server.
More projects
See all projects- 10/2026 – present
Finding the right job openings with AI agents
Instead of mass applications: agents analyze companies, openings and contacts, then rank each opportunity by how well it fits a CV.
- 12/2025 – presentAccess code required
Tactical decision-making with multi-agent reinforcement learning
Agents learn tactics in a simulated environment built on NVIDIA Omniverse. The analysis weighs win probability against the cost of equipment.