Mapping the Agentic AI Landscape: 6,700+ Startups, One Taxonomy
Published in Vento VC x Innovis VC, 2025
This is a project writeup, not a reproduction of the report’s findings. It describes what the study was, what I worked on, and who I worked with. For the actual analysis and conclusions, see the full report linked below.
What the study was
The Agentic AI Landscape Study was a collaboration between Vento VC, an Italian VC fund, and Innovis VC, Europe’s largest student-led VC association. Work ran March-May 2025, with the report delivered in June 2025.
The scope was a structural analysis of 6,700+ startups in the agentic AI space, segmented by:
- Orchestration architecture
- Memory systems
- Tool-use patterns
The team’s method combined Python/Pandas workflows for ETL, aggregation, and segment comparison, and delivered a sector taxonomy with explicit stated assumptions and limitations.
My role
I worked the technical evaluation and due diligence side of the study: assessing agent reasoning capabilities, tool-use patterns, and the reliability and hallucination challenges that show up in production agentic systems, as distinct from demo-stage systems.
Team
The study was a joint effort, publicly credited by Vento VC to: Adriano Michieli, Julius Scholl, Weronika Gorecka, Andrea Alivernini, Leo Maximilian Block, and myself.
Full report
The complete report, including the taxonomy, segment analysis, and stated assumptions and limitations, is available here:
Recommended citation: Vento VC & Innovis VC (2025). Agentic AI Landscape Study.