Hourly clean-energy claims
Full year100% clean over the year.69.7% hour by hour.
The Danish telecommunications and IT services sector, as one buyer, buys exactly its yearly consumption in Danish wind (2025). Hour by hour, only 69.7% of its consumption is met by wind produced in that same hour. HourProof makes each hour's certificate count once, proves the hourly match without revealing consumption, and lets an AI agent close the gap.
Over a year the wind purchase equals the buyer's use. With , wind that blew in the wrong hour cannot cover use, so a gap appears.
In plain terms
The problem
Clean-energy purchases are usually counted over a whole year. Buying a year's worth of clean electricity does not mean using clean electricity every hour.
What HourProof does
It counts clean energy hour by hour, makes sure each hour's is claimed only once, and lets the buyer prove its result without revealing its usage. An AI agent helps cover more hours.
How you can check
The whole flow ran on . Every step links to its , which anyone can look up.
Key results
Each figure shows the period it covers. The weekly figures come from run 20261017T141639Z on Base Sepolia.
Wind certificates bought, divided by electricity used, both totalled over the whole year.
Data and methodThe share of use met by wind produced in the same hour. This is the gap reveals.
How the agent narrows itFor the run's week, the buyer proved its hourly is at least this, with a . Its usage stays private.
See the proofThe exact figure, known only to the buyer. Shown here because this demo plays the buyer; the chain only learns the proven bound.
See the proofThe week's , proven without revealing usage. The note's target was ≤ 428.5 t.
Carbon-aware loadNo proof showed emissions within the 428.5 t target by the deadline, so the coupon stepped up: 60 tUSD paid on 1,000 notes.
The noteOne week, hour by hour
Each square is one hour of the run's week. The green part is how much of that hour's use was covered by certificates from the same hour.
70.63% of the week's use covered, hour by hour
The buyer's yearly wind purchase, spread over the hours as the wind actually blew. The figure is the run's, from its on-chain certificates.
Hover over or focus an hour to see its numbers.
Week of 2025-11-10 (UTC). Hourly coverage this week: 70.8%. Source: demo run 20261017T141639Z, carbon-aware load. This heatmap figure uses the backtest's fractional MWh; the run issued certificates in whole MWh, so the run's figure above differs slightly.
What the chain sees
Certificates cancelled per hour, and a proof that coverage is at least the target. Nothing about consumption.
Proven on-chain, by the run (with the agent)
coverage ≥ 75.00%
What only the buyer sees
The metered load. It stays with the buyer and the data hub; only a commitment to it is published.
True coverage, private (the run, with the agent)
77.47%
How it works
Five steps, each with its own page.
- 01
Hourly certificates
Metered hourly production becomes certificates. Each daily batch carries a post-quantum signature that a contract checks.
Proof run - 02
Claimed only once
The buyer cancels certificates into its claim. Claiming an hour twice, or more than was produced, is refused.
Attack tests - 03
Proven privately
The buyer proves its hourly coverage and residual emissions in zero knowledge. Its usage never goes on-chain.
Proof run - 04
Improved by an agent
An AI agent shifts flexible load and buys extra hours. Its daily plans are committed on-chain and revealed afterwards for checking.
AI agent - 05
Settled on the proof
A sustainability-linked note sets its coupon on-chain from the proven emissions.
Sustainability-linked note