Kermit GeoAI: From Tönning to Lake Constance
What Tönning Revealed
At the 33rd UIS Workshop in Tönning, the data landscape was not the problem. Environmental informatics has a mature infrastructure: standardised GIS services, well-maintained protected area databases, open environmental data. What was still missing was the direct path from a technical question to its analysis — without a detour through the GIS specialist, without manual intermediate steps.
Kermit GeoAI demonstrated how that path shortens: enter the question, receive a complete analysis plan, get the result with a standards reference — reproducible, local, auditable. What Tönning confirmed: when the data foundation is structured and accessible, automation becomes reliable. Structure is the prerequisite, not the technology.
Same Project — Four Languages
On Lake Constance, the same fundamental question arises — at a different scale. In the BIM world, the problem isn't a missing analysis step. It's a structural problem: the same infrastructure project is being worked on by four stakeholders, in four completely separate data worlds. The planner works in the IFC model. The GIS analyst works with geodata and layers. The network operator opens their CAFM system. The authority checks code distances and permit documents.
Four perspectives, four data models, four tool worlds. At every interface: media breaks, manual handovers, information loss. This is not a communication problem. It is a structural problem — and it has a structural solution.
Foundation Before Tool
The solution doesn't start with a new tool. It starts with a shared data model: what is an object? Which attributes are mandatory? Which relationships apply across all four domains? On this foundation, uniform data formats and layers emerge — a basis on which planners, GIS analysts, network operators, and authorities are genuinely working from the same state.
QGIS provides an access point without licence barriers: no exclusion of small planning offices, no hurdle for authorities, no excuse for missing participation. The shared data model only works if all stakeholders can actually use it.
On this foundation, the next step becomes possible: systematic automation. First quality checks — data completeness, topology, attribute conformance. The expert defines the rules once; the system applies them to every new data delivery. Expertise becomes replicable. This is the proof of principle — and the bridge to full orchestration.
Scaling Expertise Instead of Assuming It
Kermit GeoAI is the endpoint of this path. MovingLayers built it for its own work — developing data models, standardising structures, automating QA processes — and realised along the way that the next step needed its own tool. One that executes cross-domain analyses, answers technical questions directly, and scales expertise without requiring the user to know every step. Kermit is now a product.
The operations engineer asks the question in plain language. Kermit translates it into a complete execution plan — all analysis steps, all data sources, all processing rules, planned before the first step runs. Ahead of the jump: the agent knows the landing before it leaps. This frees up experts and creates access for all four domains.
A frog doesn't choose between water and land. It needs both — and moves freely between them. Not as a compromise. As nature. Kermit lives between the four data worlds of infrastructure, asset management, and operations — and knows all four.
Whether environmental data or BIM models: Kermit GeoAI is built for both worlds. The white paper shows what getting started looks like in your own infrastructure.
In the white paper, we describe the architecture, use cases, and prerequisites of Kermit GeoAI — as a basis for an initial conversation or your own assessment.
→ Download White Paper→ Request Initial Consultation
Reading Tips
