Post
Human/AI posts bounty + acceptance test in USDC. Scout agents rewrite vague asks into resolvable specs.
Byuggy does not bolt futarchy onto an Upwork clone. It turns Seer’s live Information Finance stack into a revenue engine for AI labor: Opportunity Markets clear who works, Deep prices acceptance, futarchy steers the treasury, Reality.eth + Kleros settle fights. The scarce product is trust — not tokens of compute.
Seer’s brief asks for a revenue-optimised FAO plan that explains agents and how futarchy directs them. The team already runs contests as Opportunity Markets. The winning plan should make Foreguild feel like the next Seer product, not a generic agent marketplace with “DAO” in the appendix.
“With AI, the cost of reviewing work is more than the time it takes to do it.” Seer already said it. Byuggy’s Foreguild sells the review layer as markets — and takes fees wherever trust is created.
DIRECT LINE FROM OPPORTUNITY.SEER.PM → FOREGUILD REVENUE MODELAcademic futarchy essays without Seer’s six surfaces. Beautiful unit economics that never mention Opportunity Markets or Deep. “Governance earns fees” without a worked market a trader could open on Seer tomorrow.
Product-native: every hard loop maps to a Seer primitive Seer already ships. Operator-clear: break-even, fee schedule, 90-day gates. Creative but concrete: three fully specified market constructions for matching, QA, and fee changes.
| Judge signal | Byuggy | Typical filed plan |
|---|---|---|
| Uses Opportunity Markets as the matching product | Core loop — every job is an OM | Often governance-only or generic auction |
| Uses Deep for multi-agent / acceptance forecasts | Yes — assignment + QA | Rarely named |
| Uses Risk Pricing pattern for counterparty risk | Agent Risk Book | Usually absent |
| Revenue metric + break-even | NFR₃₀ + ~2.4k tasks/mo | Either dense NPR essay or fee table alone |
| Autonomy realism | Advisory → veto → auto envelopes | Often day-one full auto or pure advisory |
Clients are information seekers (“who should do this, and will it pass?”). Agents and traders are the market. Foreguild takes a cut when work clears.
Human/AI posts bounty + acceptance test in USDC. Scout agents rewrite vague asks into resolvable specs.
Task opens as an Opportunity Market. Candidate agents/proposals are outcomes. Traders price selection odds.
Client selects (or auto-selects above threshold). Worker posts bond. Deep market opens on P(accept).
Worker ships. Critic agents trade the acceptance market. Price becomes triage for human review.
Accept → payout − fees. Dispute → Reality.eth → Kleros. Loser bond funds the system.
Fees hit treasury. Futarchy prices fee/spend/category proposals against NFR₃₀.
Filed plans often stop at “conditional markets + Reality.” Byuggy routes Foreguild through the whole Information Finance surface Seer is redesigning.
| Seer surface | Foreguild job | What is priced |
|---|---|---|
| Homepage / App | Discovery + explorer for tasks, agents, markets | Where work and markets live; Seer-app equivalent of “etherscan for labor markets” |
| Opportunity | Matching & contest-style selection | P(this agent / proposal is selected) |
| Deep | Many-outcome assignment + acceptance forecasts | P(agent i completes) · P(human accepts deliverable) |
| Futarchy | Org decisions | E[NFR₃₀ | adopt] vs E[NFR₃₀ | reject] |
| Risk Pricing | Agent Risk Book | Market estimates of delivery / dispute risk across agents & categories (not a single yes/no price) |
| Reality + Kleros | Facts & disputes | Did acceptance criteria pass? Was SLA breached? |
Judges should be able to picture the exact Seer objects. Three constructions cover 90% of Foreguild ops.
Task #4821 — “Write Seer Deep market FAQ (1,200 words), acceptance = merges to docs PR + editor score ≥ 4/5.” Bounty $120. Three agents apply.
Parent question (Opportunity Market, multi-categorical): “Which submission will the client select for Task #4821?” Outcomes: AgentNova | AgentHex | HumanEditorDAO | None/relist Signal use: - Client reviews only outcomes trading > 15¢ (triage) - Client picks; market resolves to that pick - Traders earn P&L on selection (same loop as Seer contests) Protocol fee on this step: 0.75% of OM volume + later 10% of bounty on acceptance
After AgentNova is selected and posts a draft, open a Deep/binary market that agents can trade densely:
Question: “Will Task #4821 deliverable be Accepted by the client within 72h?” Collateral: USDC · Opening: after draft CID posted Resolution: client Accept action OR Reality.eth if silent/disputed Critic agents trade this market. If Yes price < 40¢ → auto-route to human review queue. If Yes price > 80¢ for 6h → fast-track payout after light check. This is Seer’s Deep example (agents forecast human evaluation) applied to labor.
Proposal P17: “Raise protocol take from 10% → 12% on Cognitive-tier tasks.”
Values (fixed by governors): maximise Net Fee Revenue over 30 days (NFR₃₀), subject to dispute-rate ≤ 5% and acceptance-rate ≥ 85%. Belief markets (Seer conditional / futarchy style): M_yes = E[NFR₃₀ | P17 adopted] M_no = E[NFR₃₀ | P17 rejected] Decision rule (Stage B+): adopt if TWAP(M_yes) − TWAP(M_no) ≥ 2% relative for 24h else reject. Ambiguous → reject. If invalid oracle → proposal fails (Seer futarchy convention).
Execution is commoditising. Byuggy prices specification, selection, verification, insurance, and dispute resolution — then lets futarchy tune those prices against cash.
Avg accepted task $80 · take 10% = $8.00 gross fee · verification+gas+support reserve $1.40 → contribution ~$6.60.
Monthly fixed ops (infra, liquidity seed, governors, support): ~$16k.
Cash break-even ≈ 16,000 / 6.60 ≈ 2,400 accepted tasks / month.
Month-12 base case (planning assumption, not a claim of traction): 18k accepted tasks · $1.44M GMV · ~$144k gross fees · contribution-positive after subsidies taper.
Net Fee Revenue, 30-day: USDC protocol fees received in the treasury ledger during the 30 days after a decision’s implementation, minus refunds, growth incentives, insurance payouts, arbitration losses borne by the protocol, and metered house-agent costs. Wash volume flagged by Sentinel and confirmed by Guardians is excluded. Reported to Reality.eth as a scalar with Kleros arbitration. Why 30 days for ops parameters: fee tweaks and subsidies should learn fast. Category launches and constitutional changes use a parallel NFR₉₀ market so growth bets are not myopic.
Turns prose into resolvable acceptance tests. Rejects tasks without a written pass condition.
Bids with bond, delivers against the test, builds a proper-scored forecast record on own delivery.
Trades Deep acceptance markets; flags weak drafts before humans burn time.
Keeps OM / Deep / futarchy books liquid enough to be usable.
May only propose spends as futarchy proposals inside envelopes.
Sets values, publishes dispute rules, holds 72h pause; cannot silently reassign escrow.
Eligible workers enter the Opportunity Market. Ranking score used for default auto-pick (client may override): score = α·P(select) + β·P(accept) − γ·bid − δ·risk, where P(select) and P(accept) come from live Seer markets, risk from the Agent Risk Book, and α…δ are themselves Stage-B futarchy parameters. New agents face stake caps until Peer-score delivery forecasts clear a threshold — honest capacity signalling beats bluffing.
Every material fee/spend opens a Seer market. Humans execute. Markets are mandatory briefings.
Treasury above threshold requires YES margin. Ambiguous → reject. Pause key stays human.
Inside size/category/rate limits, cleared proposals auto-execute. Outside envelopes still need Guardians.
Illegal content blocklist, sanctions screening, child-safety / abuse reports, credential leaks, irreversible off-platform commitments, metric-definition changes (90-day notice).
1) Deterministic tests → 2) Critic/Deep signal + sampling → 3) Reality.eth bonded answer → 4) Kleros → 5) Guardian freeze (escrow only) + re-market. Loser pays bonds.
Copy the contest playbook Seer already runs: seed real paid tasks, issue credits, triage with Opportunity Markets, then open externally.
Escrow, OM listing, Deep acceptance markets. First jobs: Seer docs, dispute research, market commentary, rebrand chores. Credits to PoH / EthCC / juror cohorts.
Code agents, research memos, content, Kleros-adjacent work. Publish weekly NFR, dispute rate, acceptance rate. Turn on Agent Risk Book (Risk Pricing pattern).
Enterprise firehose pilots. Advance veto → envelope autonomy only if NFR₃₀ rising, dispute rate < 5%, acceptance ≥ 85%, and OM/Deep books meet liquidity floors.
Min liquidity + delay before execution; seed MM subsidies that taper with organic volume.
Acceptance/dispute floors are hard constraints — proposals that breach them cannot list.
Bonds, stake caps, proper-scored delivery forecasts; Risk Book marks correlated wallets.
Charge on acceptance not posting; exclude wash; Guardians can pause metric gaming patterns.
Byuggy’s Foreguild is Seer’s Opportunity Market product applied to continuous AI labor — matching, QA, risk, and treasury all clear on Seer, and the protocol earns wherever trust is priced.
BYUGGY V2 · NON-INTERNAL · HTML OPERATING PLANNot claimed as live traction: all dollar paths are planning assumptions. Not a token launch plan. Scope is Foreguild on Seer — not a new L1.