Skip to content

Implementation Plan & Milestones

Workstreams

  • WS1 — Front-end Experience: Creator studio, curator console, discovery hub, market detail, proof submission UI.
  • WS2 — Agent & Curation Backend: Prompt orchestration, draft storage, curator workflow APIs, analytics.
  • WS3 — Solana Programs: Market factory, market state accounts, settlement logic, proof ingestion.
  • WS4 — zkTLS Proof Service: Integration with TLSNotary/zkTLS provider, proof lifecycle management, submission tooling.
  • WS5 — Observability & Ops: Logging, metrics, alerting, and incident response for AI/curation/proof pipelines.

Milestone Timeline (Indicative)

Milestone Target Key Deliverables
M1 — Foundation Week 0-3 Architecture docs finalized, contract schema draft, proof provider selected, scaffolding for front-end and backend repos.
M2 — Creator MVP Week 4-7 AI-assisted studio with template enforcement, curator queue with approval flow, mocked Solana deployment (record metadata off-chain).
M3 — On-Chain Launch Week 8-11 MarketFactory program deployed, basic market cards on-chain, discovery feed powered by indexer, manual proof hash submission.
M4 — Proof Integration Week 12-15 zkTLS proof generation pipeline live, settlement program verifies hashes or proofs, UI shows proof timeline/status, first end-to-end resolution.
M5 — Quality & Scale Week 16-20 Multi-source proofs support, automated dispute alerts, ranking algorithms for discovery, post-launch instrumentation.

Detailed Tasks

-### WS1 — Front-end Experience - Set up Vue 3 + Vite project, shared UI kit, and Solana wallet integration. - Build Creator Studio: chat-like prompt panel, structured preview, validation errors. - Build Curator Console: queue list, diff view, approve/request changes/reject actions. - Build Discovery Hub: leaderboard, filters, search, market cards with proof status badges. - Build Market Detail: resolution criteria, AI rationale, liquidity, timeline, subscribe notifications. - Build Proof Submission Panel: upload/enter proof details, validation feedback, submission history.

-### WS2 — Agent & Curation Backend - Stand up Postgres (market drafts, curation trail) and Redis (session state, rate limits). - Implement DSPy-based agent orchestrator: prompt templates, response parsing, safety checks, retry logic. - Build FastAPI service exposing REST endpoints for drafts, curator actions, and market deployment triggers. - Integrate human feedback loop: capture curator edits to refine DSPy prompt heuristics. - Implement analytics dashboards for acceptance rate, turnaround time, dispute frequency.

WS3 — Solana Programs

  • Define account structures for markets (metadata, liquidity pools, resolution config).
  • Implement MarketFactory: create markets, enforce staking, emit events for indexer.
  • Implement MarketSettlement: accept proof submissions, verify or store references, execute payout distribution.
  • Provide client SDK for front-end/back-end interactions (anchor or custom).

WS4 — zkTLS Proof Service

  • Choose initial provider stack (e.g., TLSNotary) and confirm proof format compatibility with Solana verifier.
  • Implement proof generation pipeline triggered by resolution window events.
  • Build submission tool/CLI that posts proof hash + metadata to Solana program.
  • Plan fallback path (manual verification or INVALID resolution) if proof fails.
  • Roadmap multi-source aggregation and dispute handling for later phases.

WS5 — Observability & Ops

  • Centralized logging (e.g., ELK/ClickHouse) with structured logs from agents, backend, and proof service.
  • Metrics and alerts (Prometheus/Grafana) for key KPIs: market creation latency, curator throughput, proof latency.
  • Incident playbooks for AI failures, proof errors, and contract anomalies.
  • Security review: secret management for OpenAI/Anthropic keys, signer key custody for Solana.

Cross-Cutting Concerns

  • Security: Penetration test contracts before M3; implement role-based access for curator tools; monitor AI prompt injection.
  • Documentation: Update README, API docs, and runbooks per milestone.
  • Testing: Unit + integration tests for agent prompts, contract simulations, proof verification; nightly end-to-end run.
  • Feedback Loops: Weekly curator retro to adjust guardrails; trader surveys post-launch.

Risks & Mitigations

  • Proof verification complexity: Start with hash recording (M3) before full on-chain verification (M4); engage zkTLS experts early.
  • AI quality drift: Deploy evaluation suite and threshold alerts for acceptance rate; maintain versioned prompts.
  • Solana congestion: Implement retry/backoff and monitor compute budget usage; consider priority fees for proof submissions.