Technology & company
Not another mushroom app: an agentic system for nature and the outdoors
Behind Waldschatzfinder there is no stitched-together AI gimmick but an in-house data platform that brings together weather, soil, satellite and geodata from more than 15 sources for all of Europe – and, with Aragorn, an AI agent that accompanies people live all the way to the right trees in the forest. We are convinced: technically, that puts us light years away from what people know from nature apps.
Updated: 13.10.2026 · Figures live from our systems
The platform in numbers
Agentic AI, not a chatbot
Aragorn is not a language model with mushroom knowledge but an agent that works with our systems. As far as we know, it is the first AI agent of its kind for forest and outdoors.
Plans and acts on its own
Chooses from 16 tools, combines forecasts, weather, daylight, geodata and the user’s own places, and starts missing analyses by itself.
Situation-aware
Knows the location, forest, tree species, weather, daylight, season, current tour and the user’s preferences with every request.
Acts in the real world
Sets targets, records tours, guides by compass, warns of wrong turns and dusk – even without signal.
Learns from outcomes
Per-user memory and a learning loop from feedback and real find outcomes – with human approval.
Hard guardrails
Safety, links and claimed actions are checked deterministically – not left to the AI.
Privacy built in
Personal details and coordinates only leave our systems as placeholders.
An in-house data platform
Weather, soil, radar
Observations, forecasts, soil moisture and precipitation radar for all of Europe – continuously updated and held in our own database.
Satellite and geodata
Forest, tree species, elevation and soil – down to 10 metres, combined into site profiles for every grid cell.
Own models
Growth models per species, computed nightly for thousands of locations and checked against real finds.
Works offline
Installable as an app, with offline operation for the forest – no app store required.
Frequently asked questions
Is Waldschatzfinder just another mushroom app?
No. Many apps recognise mushrooms in photos or collect find locations. Waldschatzfinder is an in-house data platform that calculates growth chances for all of Europe every day – and, with Aragorn, an AI agent that guides people all the way to the right trees in the forest and accompanies them live.
What does “agentic AI” mean at Waldschatzfinder?
Aragorn doesn’t just answer questions. It decides itself which of its 16 tools it needs, combines forecasts, weather, daylight, geodata and the user’s own places, starts missing analyses, sets targets, records tours, guides with a compass – and learns from success and failure.
Which data sources does the platform use?
More than 15 public and scientific sources – from weather services to satellite, soil and geodata. The full list with licences is on our sources page. How we combine them is our own development.
How is quality ensured?
With an automated self-test suite of more than 100 checks – server, the real widget in a browser and real AI conversations – which also verifies that no personal data reaches external AI services. Safety rules such as “never say anything about edibility” are implemented as fixed filters, not just as instructions to the AI.
Can I work with Waldschatzfinder?
Yes – for partnerships, licensing, integrations and strategic conversations. Write to us via the contact form; we reply personally and confidentially.
Who’s behind it
Waldschatzfinder did not come out of an AI template. Its architecture draws on experience from quite different fields – local AI on its own hardware, real-time image, sound and speech analysis, orchestration of multiple GPUs, geodata processing, offline operation and its own server infrastructure. These building blocks interlock and are not easy to replicate: it takes deep knowledge in several disciplines at once.
Axel Jerabek · Founder and developer, Munich
- Since 1997: Development of Flow Center, media asset management software from the broadcast world for production, archiving, distribution and monetisation of media – since 2005 as co-founder, shareholder, CTO and head of research & development at Flow Works in Munich, today with AI modules that run on in-house infrastructure instead of the cloud. Around 30 years of software architecture for large media collections. flowworks.de ↗
- Systems architect and consultant: IT consulting for infrastructure, local AI and information security (NIS2, TISAX; TISAX-certified for Flow Works) – including security architecture for the media landscape of a major car manufacturer and projects for broadcasters and research institutions. consulting.jerabek.fi ↗
- Patents: DE10101075A1 and DE10101075C2 – “Method and media distribution system” for storing and transmitting video data.
- Since 2025: Founder, architect and developer of Waldschatzfinder – data platform, forecast models, geodata and the AI agent Aragorn.
- Physical AI – vælen: Develops vælen, a self-hosted perception system for cameras and microphones (“ambient perception”): it detects objects, people, movement and six behaviour signals such as falls, hears freely described sounds, transcribes speech, reads visible text, recognises faces and vehicles, flags anomalies without hand-written rules – and fuses all signals into one picture you can question in natural language, with cited sources and targeted re-analysis. An AI agent can control the system through a permissioned interface. Everything runs locally on its own hardware, from a compact mini-PC to a workstation handling several real-time camera streams, backed by an automated test suite. More about vælen ↗
- Focus areas: System architecture for media and large data volumes, agentic AI, edge and physical AI, computer vision, media on ARM and PC platforms and self-run infrastructure.
- More: www.jerabek.fi · wetransco.de · GitHub
Partnerships, licensing, strategic conversations
Interested in the technology behind Waldschatzfinder – for a collaboration, a licence or more? We look forward to a confidential conversation.