ImpactCheck

Final Carbon Audit Report Dashboard
Strategy Recomendations
Carbon Benchmark Comparison Markets & Industry
Emissions Factors Estimation AI Agent

Inspiration: AI infrastructure is the new heavy industry. As projects like Stargate move from sketches to multi-billion-dollar build-outs, we noticed a dangerous gap: tools like Watershed or Persefoni are built for CFOs to report on damage after it’s done. We built ImpactCheck for the VCs, builders, and regulators who need to see the climate impact before the concrete is poured.

What it does: ImpactCheck is a preemptive carbon intelligence platform for teams planning AI infrastructure. Instead of manual data entry, users upload real planning artifacts like procurement bills, data center design docs, and construction bids. The platform extracts candidate activities, ranks the top 100 carbon hotspots, and maps them with Climatiq to estimate emissions. It then generates an audit-ready CSV, multi-region footprint comparisons, US and EU compliance readiness checks, and scenario-based recommendations to reduce emissions before the project even starts. We also built it with a SaaS mindset from day one, including Stripe-based subscription tiers.

How we built it: We built ImpactCheck completely from scratch during the HackEurope 2026 Hackathon in Stockholm. Frontend: We used Lovable to generate a clean, terminal-style wizard flow and a dashboard that feels clear and trustworthy. Data Processing & AI: We used Claude and Gemini for the document processing pipeline, chunking PDFs and Excel sheets, extracting free-form activities (like “GPU server rack” or “Concrete mix C30”), and generating scenario recommendations. Carbon Engine: We integrated the Climatiq API to map unstructured activities to vetted emission factors, including regional context where possible. Business Logic: We implemented Stripe to support SaaS billing and upgrades.

Challenges we ran into: Turning messy procurement and construction documents into a structured emissions forecast was not easy, but honestly, getting sample data from real world examples to verify our engines precision was hard. Most of the documents we’re built for, like bids, procurement bills, and internal design plans, are usually kept private inside companies, so there isn’t much high-quality public data to test against. With some determination and creativity we were able to find sufficient material, making sure our extraction pipeline could reliably pull text, units and make inferences. On top of that, getting accurate Climatiq matches from free-form text required smarter ranking and guardrails so the mapping stayed reliable.

Accomplishments that we're proud of: We’re proud that ImpactCheck isn’t just a dashboard, it’s an end-to-end “document-to-decision” pipeline. We connected document parsing, emissions factor mapping, and scenario recommendations into one flow that’s actually usable. The features are not only super useful, but they work. And by adding Stripe and subscription tiers, we showed the shape of a real product with a scalable model, not just a one-off demo.

What we learned: The highest leverage point for any project’s footprint is in the planning phase. If you wait until the servers are wired up to care about carbon, you've already lost. Sustainable AI is a procurement problem. Also optimizing caffeine intake timing for productivity.

What's next for ImpactCheck: We will harden the core platform and build live-tracking integrations to monitor projects as they transition from digital blueprints to active deployments. Simultaneously we will be opening our doors to the investors and customers ready to be compliant with new EU directive and ready to be at the frontier of the sustainable AI revolution. We also plan to integrate a greater amount of data sources to improve the prediction quality of the emission calculator agent, including Paid Tier APIs from Climatiq, live energy grid prices and mix data, and live wind data.