What to build
Build products where people can trust the process before they know the result.
Peal lets people and agents submit encrypted information that opens at a scheduled time or a supported block height. Use it to keep decisions independent, publish records after a delay, and coordinate reveals across participants. These are opportunities to build complete products around that capability.
Research with Peal
Independent Discovery Lab
Give agents an unfamiliar simulated world. Each submits a theory, a prediction and a proposed experiment. Reveal the submissions together, run the experiments, then allow a collaborative second round.
Why Peal Nobody, including whoever runs the lab, can read a theory early, and every prediction is provably fixed before the experiment runs.
| The question | Can AI scientists produce new knowledge, and when do agent collectives help? |
|---|---|
| Peal’s natural role | Seal theories and predictions so agents cannot copy another discovery or revise a prediction after seeing it. |
| What you measure | Predictive accuracy, useful experiments, genuinely distinct theories, and whether collaboration improved the result. |
| First version | One simulated world and 10 to 20 agents over two rounds. |
Does seeing another answer change your own?
Compare independent answers with answers given after discussion.
Have people or agents seal an initial judgment, reveal the group’s responses, then take a revision. Measure what changed: accuracy, confidence and disagreement.
Why Peal First answers stay unreadable until the reveal, so their independence is guaranteed rather than requested.
Try building an interactive experiment that shows when discussion improves decisions, and when it only produces agreement.
Build a business with Peal
Performance records for agents and research publishers
Show what your system predicted before the outcome, without publishing its insights immediately.
Build a service that connects to an agent, forecasting tool or research publication and registers its outputs as they are produced. Each submission stays sealed for an agreed period, then opens into a record that readers can inspect.
Give every publisher a performance page showing resolved forecasts, mistakes, revisions and pending results. Add embeddable scorecards so the record travels with the publisher’s website, newsletter or marketplace listing. Publishers can deliver insights privately to subscribers while the same version is sealed for later public release. Peal schedules that release; it cannot stop authorized subscribers from sharing their copies.
Why Peal Each output is hash-stamped before the outcome and opens on schedule, right or wrong, so the record cannot be backdated or trimmed.
| Customers | Research publishers, agent developers, and platforms listing third-party agents. |
|---|---|
| Business model | Subscriptions for automated tracking, analytics, custom domains and API access. |
| Distribution | Every embedded scorecard introduces the service to prospective customers. |
| Peal’s role | Delayed disclosure of registered outputs. A complete track record also needs a submission log and explicit rules for missing or revised entries. |
Hosted agent competitions
Give platforms a repeatable way to engage developers and discover useful agents.
Build a competition service that an ecosystem, developer-tool company or agent platform can launch under its own brand. Organizers set the tasks, deadlines and scoring rules. Developers connect their agents through an API. Submissions stay sealed during each round, then reveal for evaluation. The platform handles onboarding, scheduled rounds, leaderboards and result exports.
Start with tasks whose outcomes can be checked: forecasts, numerical estimates, or answers evaluated against a published rubric.
Why Peal Every entry opens at the deadline, winning or losing, nobody can peek first, and agents join over plain HTTP with no account.
| Customers | Agent platforms, developer ecosystems and competition sponsors. |
|---|---|
| Business model | A fee per event, with recurring plans for organizations running continuous leagues. |
| Distribution | Organizers bring participants; participants share results and reuse their integrations in future competitions. |
| Peal’s role | Independent submissions and automatic reveal of accepted entries, including losing ones. |
Independent expert panels
Sell a structured second opinion before the group starts agreeing.
A customer submits a question: how likely is a launch to slip, what could disrupt a supplier, which assumptions matter most to a new market entry? Invite relevant experts to submit estimates, confidence levels and supporting evidence independently. Seal the first round, reveal it together, then open discussion. Record later revisions separately, so the customer can tell original judgment from consensus.
Build the product around recruiting the right experts, defining answerable questions, and tracking how their estimates perform over time.
Why Peal Nobody, the host included, sees an estimate before the round opens, so no one anchors on the first voice.
| Customers | Strategy teams, consultancies, research firms and specialist advisory networks. |
|---|---|
| Business model | Charge per panel, or sell subscriptions for recurring questions and updates. |
| Distribution | Consultancies offer the service to existing clients; experts take part across many panels. |
| Peal’s role | Keeping first-round submissions sealed until deliberation begins. Customer-confidential material needs recipient encryption as well. |
Independent AI vendor evaluations
Help buyers compare AI products on the same tasks with a documented process.
A business choosing between AI vendors supplies a representative set of tasks and agrees the evaluation criteria. Vendors submit outputs during a fixed window. Independent reviewers then assess anonymized results and seal their scores before the panel discussion.
Deliver a comparison covering quality, failure cases and the evidence behind each score. Record reported cost and latency separately from measurements taken by a controlled test runner. Build it for one buying decision first: document extraction, customer support, translation, or another task with assessable outputs.
Why Peal Vendors cannot see each other’s outputs, and every reviewer score is provably committed before the panel meets.
| Customers | Companies buying AI software, procurement advisers and evaluation consultancies. |
|---|---|
| Business model | Paid evaluation projects, then recurring assessments as models and requirements change. |
| Distribution | Integrate with existing evaluation tools, and let advisers use the product across client engagements. |
| Peal’s role | Coordinated disclosure of submissions and reviewer scores. It does not verify which model ran or whether a vendor’s claims are true. |
Judging software for recurring programs
Give organizers a clear record of how a decision was reached.
Build a judging workspace for awards, grants, accelerators and competitions. Reviewers submit scores and reasoning privately. Initial assessments open together, followed by a separate moderation stage for discussion and final decisions.
The product handles the work around the decision: assigning reviewers, managing conflicts, chasing incomplete assessments, applying rubrics and exporting results. Keep it compatible with the forms and submission platforms organizers already use.
Why Peal Scores open together, and each one carries a proof that it was not changed after the reveal.
| Customers | Program operators, foundations, associations and event agencies. |
|---|---|
| Business model | Annual plans or a fee per program, with branded portals and integrations as paid features. |
| Distribution | Agencies bring many programs; reviewers meet the product in every panel they join. |
| Peal’s role | Scheduled disclosure of independent assessments, with later changes recorded separately. |
Fundraising with a book nobody can read early
Run a startup round where every investor commits blind and everyone clears at one valuation.
The company fixes the allocation and publishes the documents that define the rights. Eligible investors escrow funds and seal their demand: how much, and up to what valuation. Nobody, the company included, sees the book while it is open, so there is no anchoring on the first big name and no allocation by who asked earliest.
At the close the whole book opens at once and a published rule computes one clearing valuation that anyone can recompute. The platform then issues a token that records each investor’s right, and handles transfers and distributions after the round.
This is the idea behind stakeholder tokens, which we are building ourselves; a platform for one market (a region, a sector, a syndicate network) is room for more than one.
Why Peal Nobody, the company included, sees the book or how many have committed until it all opens at once for one price.
| Customers | Startups raising community or strategic rounds, syndicate leads and investment platforms. |
|---|---|
| Business model | A platform fee on capital raised, plus fees for issuing and administering the tokens afterwards. |
| Distribution | Every company brings its investors, and investors come back for the next round they see there. |
| Peal’s role | Sealed commitments until the close, then one simultaneous opening for the clearing rule. Escrow, eligibility, the legal rights and the token live in your product and its contracts. |
Trade with Peal
High-leverage perps on a sealed-batch order book
Fair price discovery for a venue where one bad tick can liquidate a position.
Run a perpetuals venue whose order book clears in short, frequent batches: every few seconds, or every block. Each order for the next batch, with its size, side and limit price, is sealed on the trader’s side. When the batch closes, all of them open together and clear at one price. There is no visible book to trade against while orders are in, no large order to front-run, and no pile of stops to hunt.
That matters most at extreme leverage. At 100x or 1000x a move of a fraction of a percent wipes a position out, and the cheapest way to cause that move is to read other people’s orders first. Peal takes that information away. Leverage itself stays the venue’s risk problem: margin, the oracle, funding and liquidations live in your engine.
Be precise about what traders get. This is a book that ticks in sealed batches, not a continuous hidden one: continuous matching means reading each order as it arrives, which timed decryption cannot do. And once a batch opens, everything in it is public, unfilled limit orders and stops included, so the venue cancels what did not fill or lets traders re-seal it into the next batch.
Why Peal Orders are unreadable to everyone, the venue included, until the batch closes, and every one of them opens in the same instant for one clearing price.
| Who it’s for | Perp venues and appchains that want CEX-grade leverage without CEX-style information asymmetry. |
|---|---|
| Peal’s role | Sealing each batch of orders and opening it on time. Seals open in groups of 64; a busier batch window is split into several groups that open together. |
| What you build | The matching and clearing rule, the policy for unfilled orders, the margin and liquidation engine, oracle and funding, and a UI that shows the batch clock. |
| First version | One market, batches every few seconds, and a public replay of each batch’s orders and clearing price. |
Agent trading leagues
Rank trading agents on calls they could not copy or edit.
Agents submit a sealed position or forecast each round: long or short, size, target. Rounds open on a schedule, get marked against the market, and feed a public leaderboard. Run it on paper for a season, then let the best agents trade real capital or license their signals.
Why Peal Every call is sealed before the round closes and opens whether it won or lost, so a leaderboard cannot be built by copying the leader or deleting the misses.
| Who it’s for | Trading communities, signal marketplaces, funds scouting agents, and exchanges running competitions. |
|---|---|
| Peal’s role | Sealing each round’s calls and opening them all at once, losers included. |
| What you build | Market data, scoring, seasons, payouts, and profiles that show each agent’s full record. |
| First version | Hourly rounds on a few liquid pairs, scored on paper, with a weekly leaderboard. |
Prediction markets that clear in sealed batches
Price events on independent views, and settle them on independent reports.
Traders seal their orders on an outcome for each batch, and the batch clears at one price, so an informed trader’s order does not tip off everyone watching the book. When the event resolves, reporters seal the outcome and their reports open together, so no reporter can wait and copy the others before the market settles.
Why Peal Peal seals both halves of the market: the orders while a batch is open, and the resolution reports until they all open at once.
| Who it’s for | Prediction markets, forecasting platforms, and exchanges listing event contracts. |
|---|---|
| Peal’s role | Sealing orders per batch and reporter votes per resolution, and opening each set together. |
| What you build | Market creation, the clearing rule, collateral and payouts, and reporter staking and disputes. |
| First version | A handful of markets with hourly batches and a seven-reporter resolution round. |
Block trades by sealed request for quote
Get a size quoted without telling the market you are trading.
A trader asks several market makers to quote a block. Each maker seals a price; the quotes open together at the deadline and the best one fills. No maker sees a rival’s quote, and none can leak the request to trade ahead of it while quotes are open.
Why Peal Quotes stay sealed until the deadline, so makers compete on price rather than on who saw whose number, and the requester cannot shop one quote to another.
| Who it’s for | Funds, DAOs and treasuries moving size, and the OTC desks that serve them. |
|---|---|
| Peal’s role | Sealing every maker’s quote and opening them all together at the deadline. |
| What you build | Counterparty onboarding, settlement, and the rule for picking the fill. |
| First version | One asset pair, a five-minute quote window, and settlement through an escrow contract. |
Build DeFi with Peal
Sealed liquidation auctions
Sell liquidated collateral at a fair price instead of to the fastest bot.
When a position goes under, open a short sealed-bid auction for the collateral. Keepers seal their bids; the round opens at a set block height and the best bid wins. No gas war, no priority auction, and no keeper watching the others and outbidding by one wei.
Why Peal Bids are unreadable until the round opens at its block height, so keepers compete on price, and the protocol recovers more for the position being liquidated.
| Who it’s for | Lending protocols, CDP stablecoins and perp venues with on-chain liquidations. |
|---|---|
| Peal’s role | Sealing keeper bids and opening them together at a chosen block height. |
| What you build | The auction contract, collateral custody, and settlement of the winning bid. |
| First version | One collateral type, a few-block auction window, and a fallback to the existing path. |
MEV-free batch swaps
A swap nobody can sandwich, because nobody can read it before it is ordered.
Users seal swap intents for the next batch. At the batch’s block height they all open together and clear at one uniform price, with matching orders netted against each other before anything touches a pool. Searchers see the batch only once its order is already fixed.
Why Peal Intents stay encrypted until the batch is committed, which is the encrypted-mempool property, and the mempool demo already shows it settling real transactions on testnet.
| Who it’s for | DEXs, aggregators, wallets, and appchains that want MEV protection by default. |
|---|---|
| Peal’s role | Sealing intents per batch and opening each batch at its block height. |
| What you build | The solver or clearing rule, routing for the unmatched remainder, and settlement. |
| First version | One pair on a testnet, batches every few blocks, and a report of what each user would have lost to a sandwich. |
Oracle reports that cannot copy each other
Make every oracle node report what it sees, not what the others said.
Each reporter seals its value for the round (a price, a rate, a game result), and all reports open together. The protocol takes the median. A node can no longer wait, read the honest reports and nudge its own to shift the median, or simply copy the majority to save work.
Why Peal Reports open simultaneously, so no reporter can see another’s value first, and every report’s proof shows it was committed before the reveal.
| Who it’s for | Oracle networks, prediction markets settling outcomes, and protocols running their own reporter sets. |
|---|---|
| Peal’s role | Sealing each report and opening the full set at once. |
| What you build | Reporter staking, the aggregation rule, and slashing for reports far from the median. |
| First version | Seven reporters on one feed, a round each minute, with the median published on chain. |
Credit lines set by independent underwriters
Undercollateralized lending where no underwriter can follow the herd.
On-chain credit usually rests on one lead underwriter, and everyone else copies their call. Instead, have several underwriters assess each borrower independently: financials, on-chain history, collateral. Each seals a credit limit, a rate and their reasoning. The assessments open together, the protocol opens a line at the median, and the spread between underwriters becomes a risk signal in its own right.
Because every assessment is on record before anyone saw the others, you can score underwriters over time on which lines were repaid, and pay or weight them by that record. Borrower documents that must stay private are encrypted to the underwriters separately; the sealed round carries only the verdicts.
Why Peal Assessments are sealed until every underwriter has committed, then open at once, so each line reflects independent judgment and each underwriter builds a track record nobody can edit.
| Who it’s for | Undercollateralized and institutional lending pools, credit DAOs, and on-chain trade finance. |
|---|---|
| Peal’s role | Sealing each underwriter’s limit, rate and reasoning, and opening them together. |
| What you build | Underwriter onboarding and staking, the rule for setting the line, repayment tracking, and underwriter scorecards. |
| First version | One pool, five underwriters, and a public scorecard comparing each assessment with how its loan performed. |
Fair launches that clear at one price
A token sale with no visible order book to front-run and no allocation by speed.
Buyers seal how much they want and the most they will pay. At the close the whole book opens at once and a published rule sets one clearing price that anyone can recompute. Everyone above it fills, everyone at it fills pro rata, and bots gain nothing by arriving first. SealBid runs this shape today.
Why Peal Demand is sealed until the close and how many have committed stays hidden, so the book gives away nothing while the sale is open.
| Who it’s for | Token launches, NFT primary sales, and DAO treasury sales. |
|---|---|
| Peal’s role | Sealing every order until the close, then opening the book in one step. |
| What you build | Escrow, the clearing rule, eligibility, and distribution of what was bought. |
| First version | A fixed-supply sale with escrowed bids and a public clearing report. |
Govern with Peal
DAO votes with no running tally
Stop the last-minute whale swing and the bandwagon that follows the early count.
Ballots are sealed for the whole voting period and all open at the close. Nobody can watch the count, wait to see which way it leans, and move a large holding at the last block, and early voters do not steer late ones. Voting power is still checked against the snapshot as usual.
Why Peal There is no tally to watch until the close, and a ballot cannot be withdrawn by choosing not to reveal it.
| Who it’s for | DAOs, protocol governance, grants committees, and any vote where the live count changes behaviour. |
|---|---|
| Peal’s role | Sealing ballots until the close and opening them together. |
| What you build | Snapshot and voting-power checks, ballot format, the tally, and execution of the result. |
| First version | Sealed voting as an option in an existing governance tool, for one DAO. |
Risk parameters set by sealed median
Let a risk council set rates and limits without everyone anchoring on the first number.
Each council member seals a proposed value for a parameter: a borrow rate, a collateral factor, a supply cap. The values open together and the median is applied. A dominant member cannot anchor the room, and nobody can shade their number after seeing the others.
Why Peal Proposals open simultaneously, so each value reflects its author’s own analysis, and each carries a proof that it was committed before the reveal.
| Who it’s for | Lending protocols, stablecoins and treasuries with a risk or monetary committee. |
|---|---|
| Peal’s role | Sealing each member’s proposal and opening all of them at once. |
| What you build | The median rule, bounds on each parameter, and the on-chain update. |
| First version | One parameter set monthly by a five-member council, with every proposal published after the reveal. |
Delegate positions sealed before the debate
Record what each delegate thought before the forum decided for them.
Before a proposal goes to discussion, delegates seal their initial position and reasoning. The positions open together when debate starts, and the final vote is recorded separately. Token holders can then see who reasoned independently and who followed the crowd.
Why Peal First positions are sealed until debate begins and opened all at once, which makes a delegate’s independence something voters can actually check.
| Who it’s for | DAOs with delegates, and the platforms that rank and report on them. |
|---|---|
| Peal’s role | Sealing first-round positions and opening them together. |
| What you build | Delegate profiles, the position format, and scorecards comparing first positions with final votes. |
| First version | Sealed first positions on every proposal for one DAO’s delegate program. |
Build for curiosity with Peal
Games where everyone moves together
Make predicting other players matter more than reacting fastest.
Players seal their moves during a shared window. When the round opens, the game resolves them all against the same starting state. Build a territory game where everyone advances at once, a resource game where nobody sees the competing choices, or a tournament where humans face agents under identical rules.
Accepted moves follow the reveal even if their players disconnect.
Why Peal Every move opens in the same instant, so nobody can wait for the others, and quitting does not stop the reveal.
Try building a five-minute browser game that ends with a replay showing what everyone chose.
A creation nobody sees until the reveal
One prompt. Independent contributions. A shared moment of discovery.
Invite people or agents to contribute a drawing, melody, story fragment or solution. Keep the submissions sealed until the deadline, then reveal them as a gallery or assemble them into one collaborative work. Run it as a daily challenge, a live community event, or a human-versus-agent showcase: the reveal gives people a reason to come back and share what happened.
Why Peal Nothing can be copied before the deadline, and everything appears in the same moment.
Try building a daily page where one prompt becomes a collection of unexpected interpretations.
Bring Peal into an existing workflow
Start with a product people already use: an agent framework, publishing tool, evaluation platform or submission system. Add a sealed step where early access creates a real problem, and build the surrounding experience around the result your customer needs.
Peal handles encrypted collection and coordinated disclosure. Your application handles participation, scoring, moderation, payments and whatever happens next.
See Peal running
Some of this is already built. Try the demos before you design your own.
- Encrypted mempoollive on testnet Transactions travel sealed, builders order what they cannot read, and the batch opens when the block is final. The playground settles real transactions.
- Sealed-bid auctionslive SealBid: bids stay encrypted until the close, open together, and clear at one price.
- Create an auctionlive Open your own sealed-bid auction in a minute, with no wallet and no signup.
- Private agent actionslive An agent signs an intent that stays unreadable until its place in the batch is fixed. Run a real round and check the receipt in your browser.
- Private paymentsSepolia testnet Peal Links: a payment link where the amount and the parties stay off the public chain, on a zero-knowledge ledger.
- Stakeholder tokenscoming soon The fundraising idea above, as a product: a sealed book, one clearing valuation, and a token that records the right.
Get started · API reference · Proofs and trust assumptions · Deployment status · The protocol