#1Open•Verifiable News
Sovereign Journalist
A completely independent agent that collects data, and anyone can verify their identity. Whistleblow without revealing who you are.
Details›
Problem: Currently, it is hard for people to whistleblow because the journalist knows who they are. The only thing stopping them from revealing the source is moral integrity—which often fails.
Solution: Agent running in TEE. Anyone can verify they work somewhere via ZKTLS (email/HR provider) and whistleblow. The journalist agent investigates and publishes immutable, verified news.
Value: Everyone gets completely verified news with no one censoring this journalist because everything is immutable and on the blockchain.
by Gajesh Naik
#2Open•Memecoins
Memecoin Allocation
Allocate memecoins based on social signals with verifiable fairness.
Details›
Problem: Memecoin allocation on social isn't fair or verifiable.
Solution: Use verifiable memecoin allocation through EigenCompute + EigenAI.
Value: We take care of the core verifiable infrastructure.
by Vignesh Muralidharan
#3Open•DeFi
Private Advanced Vaults
DeFi vaults with advanced computational ability that's trusted.
Details›
Problem: Vaults right now have limited computational ability. Offchain isn't trusted.
Solution: Use EigenCompute to run more advanced vaults.
Value: We take care of building trust around the solution.
by Vignesh Muralidharan
#4Open•Gaming
Verifiable Debate
Two or more parties argue while an AI judge scores them against a transparent rubric. Every step runs in attested compute.
Details›
Problem: Internet debates are decided by vibes, moderator bias, and brigading. There's no reliable audit trail or way to settle bets, grants, or decisions.
Solution: Run the judge and rubric inside EigenCompute TEEs, publish the prompt, inputs, and scoring function, produce a signed attestation for the verdict. Tie to onchain payouts.
Value: Fair, replayable outcomes you can actually pay on. Teams get a drop-in primitive for truth-finding, competitions, and procurement decisions.
by Gajesh Naik
#5Open•Gaming
OpenFront w/ Built-In Markets
Verifiable game execution with built-in prediction markets and autonomous tournament payouts.
Details›
Problem: OpenFront's real-time economy creates high-stakes matches, but there's no trust-minimized way to settle wagers. Community reports of scripts/cheats make betting risky.
Solution: Run match simulation and bet-settlement inside EigenCompute; emit per-tick hashes and signed results. Escrow and auto-pay bets on-chain from attestations.
Value: Players get anti-cheat, provably fair payouts without trusting a mod or platform. Unlocks real money tournaments and granular side-markets.
#6Open•Dating / Social
Compatibility Match
Private agent that reads your and others' ChatGPT context to find compatibility as partner, friend or colleague.
Details›
Problem: People have rich context (chat histories, notes, calendars) but it's scattered and private. Icebreakers require reading that context, which means oversharing or trusting black-box models.
Solution: Evaluate compatibility privately inside EigenCompute against trusted data sources; output compatibility scores without exposing raw data.
Value: Better intros, higher reply rates, less awkward meetings. Organizational appetite to deploy at scale.
by Gajesh Naik, Andrew Miller, Xyn
#7Open•Security & IP
Private Code Reviewer Agent
Private agent that reviews code and finds backdoors and ways for people to RUG your money.
Details›
Problem: Teams keep codebases private for security and IP (e.g., Hyperliquid), but this requires trusting the people—huge counterparty risk.
Solution: A private agent that reviews code for backdoors and rug functionality while keeping the code private.
Value: Removes levels of counterparty risk without disclosing business logic.
by Gajesh Naik
#8Open•AI
Verifiable Private Model Evals
Privately upload model weights to the platform, we run evals without leaking your weights.
Details›
Problem: If you write a model and post eval data, it's hard for people to believe you. Even OpenAI and Anthropic are questioned. But they don't want to give out model weights.
Solution: Upload your model into EigenCompute and the program runs evals on the models and signs off on the outputs. Private and verifiable by default.
Value: Privacy, eliminate trust issue with the provider.
by Gajesh Naik
#9Open•Trading
Airline Miles Marketplace
Allow people to give their account access to a confidential environment which will trade their miles with someone else.
Details›
Problem: There is counterparty and trust risk in selling airline miles OTC.
Solution: Reverse engineer APIs to Airlines and how you transfer points. Make a program which can autonomously take login and transfer points. Act as escrow between buyer and seller.
Value: Eliminates counterparty risk.
by Gajesh Naik
#10Open•Trading
Dark Pools
Dark Pool powered by TEEs for private trading.
Details›
Problem: Dark Pools are used for 40% of securities block trades but crypto alternatives only have $1M 24hr volume.
Solution: Run it with TEEs.
Value: Private trading without giving your information ('tell').
by Sreeram Kannan
#11Open•Science
Sovereign AI Scientist
Autonomous discovery agent that executes a committed research mandate.
Details›
Problem: Labs struggle to raise because backers can't verify the work; funds risk being misused (counterparty risk).
Solution: An agent that will do what it's being funded to do. Verifiable compute makes sure the agent runs and isn't doing something else with the money.
Value: Removes counterparty risk and allows you to raise capital.
#12Open•Rewards
Distribution Based on Verifiable Attributes
Attribute-gated drops/credits (e.g., 'India-only,' '1-year Uber user,' 'GitHub power user').
Details›
Problem: Eligibility checks leak PII or get spoofed; brands face fraud/compliance pain and low-signal targeting.
Solution: Evaluate attributes privately inside EigenCompute against trusted data sources; output a signed yes/no without exposing raw data.
Value: Precise, privacy-preserving distribution with lower fraud and higher ROI; compliant targeting that actually works.
by Sreeram Kannan
#13Open•Identity
Identity Incumbering
Granular, revocable delegation so people/agents can act across TikTok/Instagram/X/GitHub—e.g., VA can like/comment/follow, bot can reply to DMs.
Details›
Problem: Today this means password sharing or brittle scripts; no per-action caps, time windows, or audit; hard to revoke and often against terms.
Solution: Issue scoped 'action tokens' via EigenCompute—e.g., '@alice may like ≤50 posts/day on TikTok+IG 9–6 PT'—all executed with per-action receipts and instant revoke.
Value: Safe, scalable delegation: no credential sharing, clear limits, one-click revoke, platform-friendly.
by Sreeram Kannan
#14Open•Private Casino
New Lotto
A 'hot–cold' lottery: verifiable compute enclave generates and seals a target number. Players submit guesses with distance hints. Dynamic pricing as guesses get closer.
Details›
Problem: Fairness & trust: operators could change the target after seeing guesses. Dynamic pricing is exploitable if controlled off-box. No tamper-evident trail.
Solution: TEE-sealed game logic: target number generated by hardware RNG and sealed. Deterministic hints via published metric. Rule-bound dynamic pricing. State & payout in enclave.
Value: Players get provably fair, tamper-resistant guessing game with real suspense and bigger pots without house meddling.
by Sreeram Kannan
#15Open•AI
Private AI Property Verifier
Private eval harness: a tester agent probes your model to verify a stated property (non-political bias, no PII spill, jailbreak resistance).
Details›
Problem: Model claims are unproven; sharing eval sets/prompts leaks IP/PII; vendors can cherry-pick; buyers/compliance can't verify without exposing sensitive data.
Solution: Run service model + adversarial tester inside EigenCompute; auto-generate challenges, score against property spec, emit attested report—weights/data never leave.
Value: Prove properties without revealing models or data; unlock procurement/regulated use; enable pay-on-pass SLAs.
by Sreeram Kannan
#16Open•Verifiable AI Judges
Custom AI Judges
Drop-in judge agents for reality singing/TV talent/debates and food-by-image scoring with transparent rubrics.
Details›
Problem: Human/online judging is subjective, slow, and gamed by brigading; platforms can't prove rubric adherence or avoid bias at scale.
Solution: Run domain-tuned judge ensembles inside EigenCompute—Song: pitch/timbre/lyric clarity; TV/Dance: sync/difficulty/originality; Debate: claim–evidence & fallacies.
Value: Consistent, fair, auditable results in real time; lower cost and higher trust for contestants/brands.
by Sreeram Kannan
#17Completed•Agents
Trading Agent for Prediction Market
Autonomous trader that places/hedges positions on prediction markets using text/news/experts under a declared mandate.
Details›
Problem: Pooled 'commons money' + qualitative strategies = drift, misuse, and over-risk with no visibility.
Solution: Run the agent inside EigenCompute with a frozen spec (model, venues, max stake, risk rules); every order emits a signed receipt.
Value: Raise and operate pooled capital with governance-grade guarantees—no drift, no misuse; audited PnL and auto payouts.
#18Completed•Agents
Verifiable Trading Agents
Framework to launch any automated strategy (crypto/equities/sports lines) with enforceable risk + audit.
Details›
Problem: Bots are black boxes; allocators can't prove what code ran or whether guardrails were respected.
Solution: Deploy on EigenCompute with budget caps, venue allowlists, kill-switch, continuous telemetry, etc.
Value: Allocate to bots you can prove are in-policy; unlock institutional capital and performance-based fees; remove counterparty risk.
#19Open•Gaming & Agents
Agent-To-Agent Combat
'Hunger Games' for agents—plan/act/adapt under shared rules and scarce resources.
Details›
Problem: Competitions get gamed: hidden advantages, covert coordination, unverifiable plays; prize awards lack credibility.
Solution: Run matches in identical, resource-metered sandboxes on EigenCompute with attested scorecards and built-in markets.
Value: Fair, replayable tournaments with trustable leaderboards; prize payouts auto-trigger on attested results.
by Sreeram Kannan
#20Open•Agents & Trading
Verifiable Intent Engines
Natural language → action: 'book NYC under $400,' 'rebalance 60/40,' 'tip top 3 contributors $10.'
Details›
Problem: Server-side NL→action touches money/accounts; users can't verify what actually executes.
Solution: Compile intents into explicit plans inside EigenCompute (APIs/contracts, limits, simulations); get consent, then execute with per-step receipts.
Value: Quick message automation without surprises—compliant, auditable actions safe for consumer and enterprise.
by Sreeram Kannan
#21Completed•Agents & Trading
Copytrading Agent
Mirrors other agents' trades with natural-language policy (e.g., 'follow @alpha 1:1, cap risk 2%, weekdays only') and revenue share.
Details›
Problem: Copytrading is opaque—signals drift, risk caps aren't enforced, followers can't prove what was executed.
Solution: Use A2A to communicate; user initializes agent with fixed parameters for operation and data sharing costs.
Value: Trustable mirroring with guardrails: no shadow trades, no overexposure, auditable PnL for allocators.
#22Open•Agents
English Language Contracts
Create enforceable conditional payouts from plain English ('I'll pay $10 if the US wins the final').
Details›
Problem: Plain-English deals are ambiguous and unenforceable; parsing to code is lossy and oracles get disputed.
Solution: Two friends commit to natural language contract → lock money → disperse when it comes true.
Value: Zero-friction agreements people understand, with clear terms and automatic, dispute-light settlement.
by Sreeram Kannan
#23Open•Data
Incentivised Data Sharing
Let sensitive data owners (e.g., genomics) get paid per query while raw data never leaves the instance.
Details›
Problem: Sharing copies leaks privacy/IP; buyers doubt authenticity; sellers get only one-off payments.
Solution: Run queries inside EigenCompute; data never leaves unless paid for based on rules.
Value: Ongoing revenue without exposure; authenticated results for buyers; no custody risk for anyone.
by Sreeram Kannan
#24Open•Data
Data Attribution Quests
Anybody that helps improve this data on a private benchmark gets additional rewards. Benchmark stays private.
Details›
Problem: Data improvement is hard to incentivize without revealing the benchmark.
Solution: Private benchmark evaluation inside EigenCompute; reward contributors based on improvement without leaking the benchmark.
Value: Incentivize data quality improvement without exposing sensitive benchmarks.
by Sreeram Kannan
#25Open•Agents
Information Curator Agent
Agent that curates information from HackerNews, BBC, Twitter in its private state. Valuable for A2A where other agents want access.
Details›
Problem: There is so much information in this world; so many things happening at every millisecond.
Solution: Agent goes through the web and curates information in private state. This state is valuable—other agents want Twitter data or what's trending.
Value: Good product for A2A where agents pay for curated, timely information access.
by Gajesh Naik
#26Open•Agents
Agent-To-Agent Negotiation
Agents holding private user data (max price, personal details, income, credit score) can negotiate with other agents.
Details›
Problem: Users can't effectively negotiate without revealing sensitive information.
Solution: Agents negotiate on behalf of users while keeping sensitive data private inside TEEs.
Value: Better deals through private negotiation; no data exposure.
by Vignesh Muralidharan
#27Open•AI Judge
Insurance Adjudication
Pre-agreed, neutral AI 'claims judge' that evaluates evidence against the policy and computes payout.
Details›
Problem: Claims are slow, opaque, and perceived as biased; adjuster variance and fraud drive disputes and cost. As seen with UHC.
Solution: Run the judge in attested enclave: lock policy + model version, ingest claim docs/photos/telematics with consent, emit signed decision.
Value: Fast, fair, auditable payouts with privacy; lower loss-adjustment expense and fraud, higher customer trust.
by Sreeram Kannan
#28Open•Climate
Verifiable Carbon Offset Mapper
Satellite takes picture and verifiable AI verifies that carbon has been offset.
Details›
Problem: Carbon offset claims are hard to verify and often fraudulent.
Solution: Use satellite imagery + verifiable AI to confirm actual carbon offset.
Value: Trustworthy carbon offset verification.
by Sreeram Kannan
#29Open•Tooling
MCP Marketplace
Open marketplace where anyone can give their API keys, seeded in private TEE. When someone uses it, disperses to the API key holder.
Details›
Problem: API key sharing is risky and there's no way to monetize unused API capacity.
Solution: TEE-protected API key marketplace with automatic revenue distribution.
Value: Monetize unused API capacity safely; access APIs without direct keys.
#30Open•Identity
Private KYC
KYC verification without exposing personal data.
Details›
Problem: KYC requires sharing sensitive personal information with every service.
Solution: Verify KYC inside TEE; emit proof of verification without exposing raw data.
Value: Compliant KYC without data exposure.
by Sreeram Kannan
#31Open•Privacy
Private Verifiable Surveillance
Govt cameras that only release footage if there's something problematic. Camera data used for one purpose only.
Details›
Problem: Government surveillance data can be misused for purposes beyond crime detection.
Solution: Run analysis in TEE; only release footage if crime is detected. Data can only be used for one purpose.
Value: Security without mass surveillance abuse.
by Sreeram Kannan
#32Open•Finance
Verifiable Credit Score
AI that analyzes your data and gives out credit score verifiably.
Details›
Problem: Credit scores are opaque and controlled by few companies.
Solution: Transparent, verifiable credit scoring using attested compute.
Value: Fair, auditable credit assessment.
by Sreeram Kannan
#33Open•Governance
AI Governance Proxies
AI asks about your values and votes on your behalf. These are delegates who operate on your behalf for corporate governance.
Details›
Problem: It's hard for people to make decisions everyday but AI has knowledge and time.
Solution: AI learns your values, researches proposals, and votes on your behalf with your consent.
Value: Better governance participation without time investment.
by Sreeram Kannan
#34Completed•Infrastructure
Private Database
Put SQLite or Postgres on EigenCompute. Only truly authorized people can access. Only *I* can access.
Details›
Problem: Database access control is often compromised; admins can see everything.
Solution: Database running inside TEE with cryptographic access control.
Value: True database privacy—even operators can't access.
#35Open•Security
Transaction Pre-Screener
Transactions go to EigenCompute for risk analysis before being sent to the blockchain.
Details›
Problem: Users sign malicious transactions without understanding the risk.
Solution: Analyze transactions in TEE before submission; warn users of risks.
Value: Prevent user losses from malicious transactions.
by Sreeram Kannan
#36Open•Security
Private Security Agent
You don't give your code access to the agent owner. Give it to the agent which runs it and gives audit report.
Details›
Problem: Security auditors might exploit or steal your code IP.
Solution: Agent runs in TEE; audits code without the operator seeing it.
Value: Secure audits without IP exposure.
by Sreeram Kannan
#37Open•Governance
Governance Rights Agent
Auction off your governance/shareholder rights to willing-buyers. Agent facilitates the transaction.
Details›
Problem: High effort to trade governance rights; counterparty risk.
Solution: User gives login to voting portal, agent talks to other agents and sells user's rights. EigenCompute removes counterparty risk.
Value: Liquid governance rights market with no counterparty risk.
by Gajesh Naik
#38Open•Finance
GoFundMe Verifiable
Verifiable crowdfunding where funds are used as promised.
Details›
Problem: Crowdfunding fraud is rampant; funds are misused.
Solution: Funds locked and released based on verifiable milestones.
Value: Trustworthy crowdfunding.
by Gajesh Naik
#39Open•Trading
Trading Vaults on HL
Advanced trading vaults on Hyperliquid.
Details›
Problem: Limited vault capabilities on Hyperliquid.
Solution: Use EigenCompute to run advanced vault strategies.
Value: More sophisticated, trustworthy vaults.
by Gajesh Naik
#40Open•Gaming
Minecraft Servers
Verifiable Minecraft servers with provable game state.
Details›
Problem: Minecraft server admins can cheat or manipulate game state.
Solution: Run Minecraft server in TEE with attested state.
Value: Fair, cheat-proof Minecraft.
by Gajesh Naik
#41Open•Trading
Liquidikey for Gift Cards
Trade gift cards safely using TEE escrow.
Details›
Problem: Gift card trading has high fraud and counterparty risk.
Solution: TEE verifies gift card validity and handles exchange atomically.
Value: Safe gift card marketplace.
by Gajesh Naik
#42Open•Data
Access Control of Data
Fine-grained access control for data using TEEs.
Details›
Problem: Once you share data, you lose control of it.
Solution: Data accessed only through TEE with enforced access policies.
Value: Maintain control over your data even after sharing.
by Gajesh Naik
#43Open•Knowledge
Verifiable Pedia
Community-funded wikipedia where AI agent decides what content should be based on verified sources.
Details›
Problem: Wikipedia has bias and edit wars; hard to verify neutrality.
Solution: AI agent curates content with transparent, verifiable methodology.
Value: Trustworthy, neutral knowledge base.
by Gajesh Naik
#44Open•Verifiable News
Verifiable News Synthesizer
Like Grok Twitter but aggregates from different models verifiably.
Details›
Problem: News aggregation is biased; no way to verify methodology.
Solution: Multiple models synthesize news with verifiable, transparent process.
Value: Trustworthy news synthesis.
by Gajesh Naik
#45Open•Data
Pay To Get Info From Chat History
Monetize your chat history by letting others query it privately.
Details›
Problem: Valuable insights locked in private conversations.
Solution: TEE-protected queries on chat history with payment per query.
Value: Monetize personal data without exposure.
by Gajesh Naik
#46Open•Trading
Robinhood Trading Pattern Agent
Agent that trades based on Robinhood retail trading patterns.
Details›
Problem: Retail flow data is valuable but hard to trade on trustlessly.
Solution: Verifiable agent that transparently trades on retail patterns.
Value: Transparent, trustworthy retail-flow trading.
by Gajesh Naik
#47Open•Tooling
Protocol Copilot
AI copilot for protocol development and interaction.
Details›
Problem: Protocols are complex; users make costly mistakes.
Solution: AI copilot that understands protocols and guides users safely.
Value: Safer, easier protocol interaction.
by Gajesh Naik