Schema-driven LLM extraction
GPT-4-based extraction of structured ecosystem information — stakeholders, relationships, pain points — from workspace documents, validated against a defined schema.
Mentis · Case studies · B2B software
LLM-based automated ecosystem mapping (ECOMAPPER)
A document-intelligence pipeline that reads large, heterogeneous document collections and extracts stakeholders, relationships, needs and pain points as structured, schema-driven data — feeding the client’s ecosystem-mapping platform.
GPT-4Neo4jStreamlitSchema-driven extraction
In short
- Client
- Swiss B2B software SME — ecosystem-mapping platform
- Sector
- B2B software
- Timeline
- 2025 · 8–12 weeks
- Status
- Prototype and integration delivered (2025).
- Runs on
- Cloud-deployed prototype
The challenge
Analysts extracted stakeholders, relationships and needs by hand from large collections of heterogeneous documents before any mapping could start — slow, inconsistent and impossible to scale.
What we built
Mentis delivered the pipeline and its integration in a defined 48 person-day engagement:
Graph representation
Extracted entities and relations stored in Neo4j, ready for ecosystem mapping and queries.
Analyst application
A Streamlit application to run, review and correct extractions.
Platform integration
Results fed into the ECOMAPPER platform; designed to process document collections scaling into the thousands.
Results
- Automated extraction of structured ecosystem information from unstructured documents
- Designed for document collections scaling into the thousands
- Replaced a significant share of manual document reading and structuring
Qualitative outcomes; hours saved were not measured, so no figure is claimed.
Other case studies
Media & advertising
AI & analytics platform for advertising effectiveness
A multi-source analytics and AI platform that scores campaigns, runs pre-tests, integrates eye-tracking data and recommends optimisations — later extended with an LLM/RAG engine that explains its recommendations from historical campaign knowledge.
Read the case →Healthcare & medical research
AI-assisted digital pathology for severe asthma (SputumAI / Cytomine)
A Cytomine-based medical-imaging and AI environment for sputum cytology in severe asthma: expert annotation at scale, AI-assisted labelling, cell detection and classification, morphology and inflammatory biomarkers — on dedicated GPU infrastructure.
Read the case →Next step
Want a similar outcome?
Tell us the problem in a 30-minute readiness debrief. You get an honest read on feasibility, data, compliance exposure and a first step.