Home/The project/Roadmap

Challenges that will skyrocket innovation

Today, Karst Firewall 5.0 is a cross-border wildfire twin. This is where we want to take it next: a living, multi-hazard decision platform that learns, sips energy, scales from the Karst to the whole Mediterranean, and gives insurers, infrastructure operators and planners modular tools to build on. A roadmap, openly shared.

Planned directions — not yet live
The direction

From one hazard to many — from a tool to a platform

Four shifts shape everything below: from a single hazard (fire) to many; from static maps to living data that updates itself; from a finished tool to a modular platform others can build on; and from one plateau to a sea. None of this is live yet — it is the direction we are taking the twin, openly and with our partners.

Where we'd start Our recommended first moves — highest impact, least new plumbing

Three things to build first

01 · Now
The Dynamic Fuel State layer

Shovel-ready: the satellite NDVI / vegetation-health pipeline already feeds the platform. Turning it into a live “how dry is the fuel today” layer sharpens every fire forecast for the least new effort — the clearest win.

02 · Now
Flood & extreme-heat layers

The fastest way to become multi-hazard: the cross-border weather network and 3D terrain we already run are most of what flood and heat-wave mapping needs. Two hazards that touch every citizen — not only fire crews.

03 · Now
Scale to all FVG & Slovenia

Mostly data and compute, little new science. Covering the whole region multiplies the platform’s reach and value before the harder leap to the wider Mediterranean.

Everything else below — compound multi-risk, agentic & frugal AI, the modular decision-support platform, the Mediterranean twin — builds on these three foundations.

Pillar 1 · Multi-hazard

From a wildfire twin to a multi-hazard twin

The Karst does not face only wildfire. The same living twin — terrain, weather, exposure — can read flood, heat, slope and seismic risk, and, crucially, where they overlap on the same people and assets.

Flooding & flash floods

Now · 0–12 mo

Pluvial and flash-flood exposure from rainfall, terrain hydrology and the karstic underground drainage — where water gathers fast on the limestone, and which roads, homes and assets sit in its path.

Impact: High · reuses the weather network + 3D terrain

Extreme heat & “heat bombs”

Now · 0–12 mo

Heat-wave and urban-heat exposure from the station network and land cover — a public-health early warning for the most vulnerable people and places, and a fuel-drying signal that feeds the fire model too.

Impact: High · directly serves citizens & health services

Slope instability & landslides

Next · 1–2 yr

Landslide susceptibility from slope, soil and rainfall — sharpened after fires, when burnt ground sheds water and the risk of debris flows climbs. A natural pairing with the post-fire recovery view.

Impact: Medium · compounds with wildfire & flood

Seismic & compound risk

Next · 1–2 yr

Overlay seismic hazard and cross-correlate the layers into one compound-risk index: where do wildfire, flood, heat and earthquake stack on the same zone? Compound exposure, not four maps read in isolation, is what planning and insurance really need.

Impact: High · the multi-hazard payoff
Pillar 2 · Living data

A twin that reads the territory as it is today

Static fuel maps go stale the moment they are drawn. Reading vegetation from space, continuously, makes the twin’s fuel “alive” — and a living fuel layer is the single biggest accuracy gain available to the fire model.

Dynamic NDVI → Dynamic Fuel State

Now · 0–12 mo

Turn the satellite “greenness” (NDVI) signal into a live fuel-moisture and fuel-state layer, so the fuel the simulator burns is today’s fuel — green and damp, or cured and flammable — not last season’s static map.

Impact: High · the NDVI pipeline already exists

Always-current fuels

Next · 1–2 yr

Burn scars, drought stress, harvests and regrowth update the fuel map automatically from each new satellite pass — and flag stressed, fire-prone vegetation weeks ahead, turning the twin into an early-warning instrument for the land itself.

Impact: Medium · keeps every hazard model honest

See how we already watch the Karst from space →

Pillar 3 · Intelligence

Intelligence that learns — and sips energy

The platform should get better on its own, and do so frugally. More capable does not have to mean more power-hungry: the goal is intelligence that learns from every fire, always under human oversight, while using less energy each year.

Agentic assistance

Next · 1–2 yr

Software agents that watch the live data, draft the morning risk briefing, surface anomalies and propose interventions — always presented for a human to approve, never acting unsupervised.

Impact: High · saves operators time daily

Frugal & green AI

Next · 1–2 yr

Smaller, distilled models and inference at the edge — closer to the sensors — for lower energy, lower cost and resilience when connectivity drops. Sustainability built into the AI itself, not just the forest.

Impact: High · cuts running cost & carbon

Self-learning & self-improvement

Later · 2–4 yr

Close the loop: the platform calibrates itself against what really happened — every fire it predicted well or badly becomes training signal — so accuracy improves over time without a manual rebuild.

Impact: High · but needs careful governance
Pillar 4 · A platform to build on

Decision support, assembled from Lego bricks

The real prize is a modular platform — composable building blocks that let a twin or a decision-support tool be assembled for a new territory, a new hazard or a new audience, instead of rebuilt from scratch each time.

A “Lego bricks” architecture

Next · 1–2 yr

Reusable bricks — a hazard model, a data feed, a map layer, a report — that snap together. Stand up a new digital twin by composing parts that are already tested, not by writing it all again.

Impact: High · multiplies everything else

Decision support for new sectors

Next · 1–2 yr

A decision-support tool (DST) tuned for insurers (risk pricing & portfolio exposure), infrastructure operators (network resilience) and spatial & urban planners (risk-aware planning) — new audiences for the same evidence base.

Impact: High · opens a sustainable business model

Plug-in interconnection

Next · 1–2 yr

Open APIs and connectors so the twin talks to GIS, emergency-management, environmental and IoT systems — one holistic cockpit for managing territorial sustainability, rather than another silo.

Impact: Medium · the glue for an ecosystem

Regenerative engineering

Later · 2–4 yr

Aim beyond resisting hazard toward restoring it: nature-based, net-positive interventions that leave ecosystems and communities stronger than before. The twin becomes a design tool for regeneration, not only defence.

Impact: Strategic · the long-term ethos
Pillar 5 · Scale

From the Karst to the Mediterranean

The cross-border Karst is the proving ground. The same engine can grow outward in deliberate steps — each one mostly data and compute, the science already in place.

01 · Now

All Friuli-Venezia Giulia

Extend from the Karst AOI to the whole region — the natural next catchment, with data partners already in place.

02 · Now → Next

All of Slovenia

A truly national, cross-border twin — completing the picture for the whole Slovenian side, not just the border strip.

03 · Later

The Mediterranean

The climate-and-fire frontline. A shared twin for a shared sea — the ambition that gives every brick above its full meaning.

The sequence

Now, next, later

The same initiatives, sequenced by impact and readiness. This is our honest view of the order — not a contract, but where the leverage is.

Horizon 1

Now · 0–12 months

  • Dynamic Fuel State (NDVI) layer
  • Flood & extreme-heat hazard layers
  • Scale to all FVG & Slovenia
Horizon 2

Next · 1–2 years

  • Compound / multi-risk index
  • Agentic assistance + frugal AI
  • Modular DST for insurers & infrastructure
  • Plug-in / API ecosystem
Horizon 3

Later · 2–4 years

  • Seismic correlation in the compound index
  • Self-improving autonomy (governed)
  • Regenerative-engineering design suite
  • A Mediterranean-wide twin
Principles

What holds across every step

Frugal & green by designLess energy, less cost — efficiency is a feature, not an afterthought.
Human-in-the-loopAI proposes; people decide. Automation assists, it never acts alone.
Open & modularComposable bricks and open interfaces, so others can build on the work.
Cross-border & sharedHazards ignore borders; the data and the tools should too.
Regenerative, not just protectiveLeave the territory and its communities stronger than we found them.
Privacy & transparencyAnonymised by default, methods explained, results open to scrutiny.