Executive summary
Climate governance · Partner Use Case80 climate action plans. The challenge isn’t more data — it’s a sparring partner that helps navigate between them without removing the specialist from the centre.
AI generates options
CO₂ impact · Cost · Breakeven
Specialist decides
Accept · Edit · Discard · Combine
Portfolio visible
Effort-effect matrix · Full overview
80+ climate actions spanning afforestation, solar, cycling, and waste — each with different timelines, budgets, and tradeoffs. Prioritising across them requires analytical depth no single specialist can maintain alone.
A dialogue-based climate assistant built from scratch. The specialist frames the context; AI returns structured proposals with CO₂ savings, costs, and breakeven timelines. The employee edits or discards each suggestion — and sees the full portfolio update on an effort-effect matrix.
Specialists can explore a far larger solution space in a fraction of the time, while retaining full ownership of every decision. An early proof of concept that structured human-AI dialogue can make public-sector climate governance more rigorous and transparent.
Introduction
How can climate impact become as natural a part of municipal decision-making as economics? That is the question driving a collaboration between Ringkøbing-Skjern Kommune and CHI, led by Frederik Brosbøl Kjeldsen and built technically by Paul Huguet. Together with Henning Donslund, specialist consultant at the municipality, the team developed a Hybrid Intelligence tool that supports greener, more qualified decisions — not by replacing municipal expertise, but by putting it in dialogue with data.
The Context
Ringkøbing-Skjern Kommune is one of Denmark’s most ambitious municipalities on climate. It already works with more than 80 concrete climate action plans spanning afforestation, biodiversity corridors, solar installations, cycling infrastructure, and waste reduction. The challenge is not a lack of ideas — it is the complexity of navigating between them.
When a specialist needs to prioritize across climate actions, three problems arise simultaneously. First, generating new ideas that fit within local context, available budget, and CO₂ targets requires substantial analytical work. Second, refining and deepening existing plans — estimating breakeven points, calculating energy savings, comparing cost per tonne of CO₂ — is time-consuming and requires data that is rarely at hand. Third, prioritizing across a portfolio of 80 actions, each with different timelines, stakeholders, and tradeoffs, is a judgment problem that no spreadsheet can solve alone.
What the municipality needed was not more information. It was a sparring partner that could bring data-driven options to the table while keeping the specialist’s contextual knowledge, governance requirements, and local judgment at the center of every decision.
The HI Element
The tool CHI and Ringkøbing-Skjern Kommune built together is a dialogue-based climate action assistant — built from scratch, not off the shelf. The architecture reflects a deliberate HI design: AI generates options and calculates impacts, humans decide what is relevant, feasible, and worth pursuing.
In practice, the workflow looks like this. A municipal employee opens the tool and either describes their own idea or asks the system to generate proposals based on the 80 existing action plans. The AI returns structured suggestions — each with a title, description, estimated CO₂ impact in tonnes per year, cost in DKK, energy figures, and a breakeven timeline. The employee can accept, reject, edit, or combine these suggestions. They can upload their own documents to enrich the context. They can inspect the underlying prompts driving the AI’s reasoning. And they can visualize the full portfolio on an effort-effect matrix — seeing at a glance where each action sits relative to investment and impact.
This is not a black box recommending decisions. It is a transparent, multi-step interaction in which the human remains the author of every choice. The context window — developed in direct response to user feedback — allows the employee to describe the specific background they are working within before the AI generates anything. This single feature embodies the core HI principle: the machine responds to human framing, not the other way around.
The prototype illustrates this logic concretely. A list of strategy points sits alongside a chat interface. The employee selects “Fremme af cykelinfrastruktur” and the system returns an estimated cost of 6,000,000 DKK and a CO₂ saving of 250 tonnes per year, both positioned on an effort-effect scale. The employee can then edit, refine, or discard the suggestion. The graph updates. The portfolio shifts. The judgment remains human.
The Envisioned Outcomes
The project demonstrates that Hybrid Intelligence is not only relevant to private sector efficiency. Public institutions face some of the most complex tradeoff problems imaginable — balancing CO₂ targets, local politics, budgets, and long time horizons — and these are precisely the contexts where the combination of machine analysis and human judgment creates the most value.
At the tool level, the climate decision support system gives municipal specialists a genuinely new capability: the ability to explore a much larger solution space in a fraction of the time, while maintaining full ownership of every decision. At the organizational level, the co-creation process with Henning Donslund and his colleagues has itself been an exercise in Hybrid Intelligence — features added, scoped, and revised in direct response to what emerged from real use, ensuring the tool reflects the realities of municipal work rather than assumptions about it.
For other Danish municipalities navigating similar complexity, Ringkøbing-Skjern offers an early proof of concept: that structured human-AI dialogue can make climate decision-making more rigorous, more transparent, and more grounded in local expertise — without removing the human from the center of the process.
