A B2B middleware layer that resolves contested outcomes for prediction markets, parametric DeFi, and onchain insurance — through AI evidence aggregation, reputation-weighted jurors, and onchain settlement. Customers stop building dispute teams. End users get transparency. You capture a take-rate on every resolution.
Every prediction market faces the same problem: ambiguous market wording, edge-case real-world events, and motivated reasoning by large stakeholders. When something goes wrong, resolution gets handled in a Discord thread or an opaque internal review — and users lose confidence.
Polymarket's Ukraine-suit market resolved against community sentiment via UMA whale votes. Kalshi rules internally. Manifold lets the creator decide. None of this scales as markets professionalize.
Every market builds the same ad-hoc dispute machinery. Lawyers, ops staff, on-call engineers reviewing screenshots. It's expensive, slow, doesn't scale across categories, and creates regulatory liability.
Existing decentralized oracles (UMA, Augur, Kleros) have known capture vectors: token-weighted voting favors whales, jurors copy obvious answers, and dark-DAO bribery markets are already a documented attack.
Polymarket lost an estimated $4-8M in user retention after the Ukraine suit dispute went viral. Kalshi has a four-person disputes team that's purely a cost center. Smaller markets (Manifold, Zeitgeist) lose users every time a creator-resolved market is alleged to be unfair. No one has solved this — we will, and sell it back to all of them.
Here's exactly what happens from the moment a market closes to the moment funds settle. Click any step to expand.
Anyone proposes the outcome with a small bond. If unchallenged within the window, it settles automatically.
Most outcomes are unambiguous. Optimistic resolution keeps the cost of routine settlement near-zero and reserves expensive jury machinery for the cases that actually need it.
Anyone can dispute by posting a counter-bond. Both bonds enter escrow; the wrong side gets slashed.
Bonds are large enough to deter frivolous disputes but small enough that legitimate ones aren't priced out. Curve scales with market volume.
An agent gathers and ranks evidence — news, official sources, onchain data, social signal — and produces a structured brief.
Smart contracts and DAO voting are open-source commodities. The proprietary evidence engine — trained on our resolved disputes, with a credibility graph that compounds — is what nobody else can clone.
A reputation-weighted jury is randomly selected from the relevant domain pool (politics, sports, crypto, real-world).
A juror who's resolved 200 NFL disputes correctly is more credible than a generalist. Reputation is non-transferable — earned by voting with eventual consensus, slashed otherwise. Whales can't buy influence.
Jurors submit encrypted votes with mandatory reasoning. Reveal phase exposes everyone's vote simultaneously.
Prevents herding ("just copy what everyone else is voting"). Mandatory reasoning blocks lazy jurors. Recursive evidence ratings reward principled minority dissent on close calls.
Continuous slash curve based on consensus distribution. Bonds resolved, rewards distributed, market unfrozen.
For irresolvable disputes: tiered appeals with increasing bonds, full DAO vote, or market fork as the nuclear option.
Larger jury (45+), 2x bond requirement. ~3% of disputes reach this.
Full token-holder vote. Highest cost, highest bonds. ~0.5% of disputes.
When wording is fundamentally broken: split into two markets, holders self-select.
Each view is a separate sales artifact. Operators see one thing, jurors see another, end users see a third — and the public auditability of the archive is the marketing flywheel.
The B2B sales screen. Live disputes, resolution stats, bond flow, juror leaderboards, SLA compliance. This is what you put in front of Polymarket and Kalshi.
The human-in-the-loop trust layer. AI evidence brief, full source table, vote interface with continuous slash curve. Where your moat compounds with every resolved dispute.
The retention play. End users see exactly what's happening to their money in plain English — timeline, evidence, consensus, insurance. The screenshot for the marketing deck.
Every dispute, every juror vote, every bond transaction — publicly auditable. Becomes the second product (data API) once volume builds. The transparency moat.
Ranked by realism for years 1–3. The protocol is decentralized for credibility. The company is a SaaS business with traditional revenue, traditional accounting, and an obvious acquisition path.
1–3% protocol fee on every bonded resolution. Scales linearly with prediction-market volume. Predictable, defensible, no enterprise sales motion required.
Tiered pricing for customers who need more than the base protocol: priority resolution SLAs, custom jury composition, white-label UI, dedicated support.
Polymarket, Kalshi, sportsbooks, parametric insurance protocols. $250k–$2M/yr contracts with custom terms, on-call support, and integration engineering. These take 6–12 months to close but each one is a significant revenue line.
Anonymized resolution patterns, accuracy benchmarks, juror behavior data. Sold to event-driven hedge funds, research firms, and competing oracles. Becomes meaningful only after volume builds — a strong year 3+ play.
Small spread on dispute insurance pool yield. Real money but raises serious regulatory questions (running an insurance pool ≠ running a protocol). Probably year 3+ once legal posture is settled.
Four phases, 36 months. Honest about what we don't know, what we need, and what we'd want to validate before each step.
Become the default resolution layer for crypto-native event markets. Launch the data product. Sign exclusive deals with category leaders. Acquisition window opens — the buyers are the markets themselves (Polymarket, Kalshi, FTX-successors), oracle protocols (Chainlink), or compliance/data firms (S&P, Bloomberg).
Existing oracles solve part of the problem and get acquired by their largest customer or stagnate. Our differentiation is the AI evidence layer (proprietary), domain-specialized reputation (compounding), and being a B2B middleware company (not a token bet).
| Capability | Us | UMA | Kleros | Augur | Internal teams |
|---|---|---|---|---|---|
| AI evidence aggregation | |||||
| Domain-specialized reputation | |||||
| Whale-resistant voting | |||||
| Public auditability | |||||
| B2B middleware (sells to markets) | N/A | ||||
| Continuous slash curve | |||||
| Insurance pool | |||||
| Cost to customer per resolution | $5–50 | $0 + bond | $10–100 | High | $200–2k |
UMA is locked in with Polymarket and won't be displaced for that customer. Our wedge is the long tail — the other 50+ prediction markets, parametric DeFi protocols, and DAO governance systems UMA doesn't serve well. Win those, then pitch the whales.
UMA or Chainlink build their own AI evidence layer. Mitigation: ship first, capture juror network, build proprietary reputation graph that compounds. By the time they react, switching costs are real.
Honest list. This is what gets us through Phase 1 and into the first revenue.
The demo is real. The market is real. The path to revenue is concrete. What's needed now is a team and the runway to ship Phase 1.