0. Executive summary
Oracle Risk Stack chỉ đáng làm nếu nó biến “oracle anomaly” thành “protocol impact + decision path” nhanh hơn, đáng tin hơn, và rẻ hơn một analyst/risk engineer đang trực thủ công.
Category to claim
Risk Ops
Oracle incident intelligence + decision support, không phải oracle provider.
Beachhead
5–20
Qualified lending/perp/vault teams để phỏng vấn/demo trước khi build rộng.
First paid proof
$500+
Setup hoặc monthly pilot. Nếu không thu được gì sau direct asks, kill subscription wedge.
Verdict
- Go, nhưng chỉ ở chế độ validation có kỷ luật. Không build platform trước.
- Wedge đúng: Watchtower + incident timeline + blast-radius stub + policy replay narrative.
- Moat tương lai: incident database, protocol-to-feed dependency graph, policy replay benchmarks, not UI.
- Không nên bán: raw price feed, generic dashboard, “AI autopilot”, automated liquidation guardrails giai đoạn đầu.
Điểm khắt khe: nếu 10–15 cuộc nói chuyện với protocol/risk teams không tạo được 2–3 LOI/pilot trả tiền hoặc ít nhất explicit weekly-use commitment, dự án phải pivot sang research/service memo hoặc dừng UI build.
0.5 Critical stress test
Cold verdict
Hôm nay Oracle Risk là product thesis + PRD + mockup, chưa phải business. Business chỉ bắt đầu khi một buyer có budget nói: “Tôi trả tiền để giảm thời gian triage oracle incidents / audit risk decisions / prove governance response.”
| Assumption | Critical question | Evidence today | Verdict |
|---|---|---|---|
| Protocols đau vì oracle risk | Đau đến mức trả tiền ngoài existing tools? | Incident history và industry writing xác nhận oracle risk tồn tại; chưa có buyer interviews. | Promising, unproven |
| Dashboard + alerts đủ giá trị | Alert có vào workflow quyết định không? | PRD có flow; chưa có runtime alert quality. | Unproven |
| Blast-radius map là wedge | Map có đủ chính xác để risk team tin? | Chưa có database feed→protocol→TVL. | Hardest gap |
| Policy replay là moat | Có đủ incident data để replay có ý nghĩa? | Chưa compile incident corpus. | Buildable |
| Team có lợi thế Chainlink | Lợi thế này tạo sales access hay chỉ là thesis? | Team có Chainlink thesis + treasury context, nhưng chưa có partnership/channel. | Narrative edge only |
Buyer / budget question
Buyer khả thi không phải “DeFi user”. Buyer là Head of Risk, protocol founder, governance risk delegate, vault operator, hoặc market-maker ops lead. Budget đến từ risk tooling, analytics, consulting, infra monitoring, hoặc custom reporting — không phải retail subscription.
Kill / pivot criteria
- Kill UI build: sau 15 qualified demos, dưới 3 người muốn dùng hàng tuần hoặc dưới 2 người đồng ý trả pilot/setup.
- Pivot to service: khách thích incident memo/report hơn dashboard; bán “Oracle Risk Brief + custom blast-radius analysis” trước.
- Pivot to internal treasury/tooling: không có external WTP nhưng nội bộ dùng được cho LINK/Chainlink risk monitoring.
- Do not proceed to auto-action: trước khi có false-positive benchmarks, audit logs, legal review, and customer trust.
1. Sources & confidence
| Source | What it supports | Confidence |
|---|---|---|
| Internal PRD: oracle-risk-stack-prd-roadmap-2026-03-19.md | Product modules, 30/60/90 roadmap, non-goals, MVP scope. | High internal |
| TT-HANDOFF.md | Risk taxonomy, ICP, open tasks, product ladder. | High internal |
| Chainlink Data Streams official/docs | Low-latency pull oracle, sub-second data, market data risk indicators. | High official |
| Pyth Price Feeds official | First-party market data, confidence intervals, plan structure. | High official |
| Chaos Labs Oracles official | Price + risk + proof feeds; competitor with real-time risk positioning. | High official |
| Gauntlet oracle-risk resource | Oracle KPIs, fragmented lending meta, risk perspective. | Medium |
| OpenZeppelin Defender/Sentinel public pages | Monitoring/ops tooling comparison, not direct oracle-risk competitor. | Medium |
No fabricated traction, revenue, TAM, logos, or customer commitments are claimed. Financials below are planning assumptions only.
2. Product positioning
Claim this
Oracle Risk Operations Console for DeFi protocols: detect feed/data anomalies, map protocol exposure, and produce defensible action recommendations with replayable evidence.
Avoid this
Do not claim to be a new oracle provider, automated onchain defense system, compliance attestor, or generic crypto dashboard.
One-liner
“Oracle Risk Stack helps protocol operators detect bad market-data states, understand blast radius, and decide what to do before oracle issues become bad debt, wrong liquidations, or governance chaos.”
Why this wedge is sharper than Chainlink Trust Monitor
The older Chainlink Trust Monitor thesis is broader. For Oracle Risk, the sharper commercial wedge is narrower: incident triage and decision support for oracle-dependent protocols. Treasury/reserve monitoring can remain a later adjacent module, but it should not blur the first buyer.
3. Market opportunity
Market should be treated bottom-up, not TAM theatre. The reachable first market is small: protocols and operators with meaningful oracle dependency and enough TVL/OI/reputation risk that false liquidation or stale price incident is expensive.
| Segment | Pain | Ability to pay | Launch priority |
|---|---|---|---|
| Lending protocols | Collateral valuation, LTV changes, bad debt, liquidation safety. | High if TVL meaningful. | P0 |
| Perp / synthetic protocols | Mark price integrity, funding, oracle latency, manipulation risk. | High but demanding. | P0 |
| Vault / strategy operators | Rebalance errors, bad NAV, dependency opacity. | Medium. | P1 |
| Risk delegates / governance | Need defensible proposals and postmortem evidence. | Medium; often service/retainer. | P1 |
| Retail / general investors | Curiosity, not budget. | Low. | Avoid |
Market trap: “DeFi is large” does not mean this product has a large market. The paid market is only teams with operational responsibility and budget for risk tooling.
4. Problem
Oracle failure is rarely a single clean event. Operators need to distinguish: market dislocation vs stale feed vs source divergence vs manipulation vs safe recovery. Current internal PRD identifies four risk classes: stale data, divergent data, economically wrong but technically valid data, and no action layer.
Economically valuable jobs-to-be-done
- Tell me within minutes which feeds are unsafe or degraded.
- Tell me which markets/protocol functions are exposed.
- Show why this is an oracle/data problem versus real market movement.
- Produce an action recommendation with confidence and audit trail.
- Replay past incidents to justify thresholds and governance policy.
5. Product strategy
Product ladder
| Layer | Purpose | Launch scope | Revenue role |
|---|---|---|---|
| Watchtower | Freshness/deviation/divergence alerting. | Must launch first. | Demo wedge + monitoring value. |
| Risk Console | Dependency map, blast radius, incident explorer. | Launch as “manual-assisted” first. | Differentiation; supports higher ARPA. |
| Defense Engine | Policies, simulation, runbooks, approvals. | Simulation only. | Enterprise narrative + future moat. |
MVP that can be sold in 30 days
- 10–20 high-value feeds/assets across Chainlink/Pyth/reference sources.
- 3 anomaly detectors: heartbeat lag, cross-source deviation, confidence-band/volatility stress.
- Incident timeline with raw source snapshots.
- Manual blast-radius mapping for 5–10 protocols/markets.
- Weekly Oracle Risk Brief PDF/HTML plus urgent Telegram/webhook alert.
- Policy replay on 3 historical incidents: Mango-style manipulation, stale feed, source divergence.
Do not build yet
- Full multi-tenant permissioning.
- Automated onchain action.
- Broad all-chain/all-feed coverage.
- White-label dashboards.
- AI-generated actions without deterministic evidence.
6. Competitive landscape
| Player | What they do | Threat | Opening for Oracle Risk |
|---|---|---|---|
| Chainlink Data Streams / Data Feeds | Low-latency market data, DON-signed reports, risk indicators. | Could own data-quality layer. | Build on top as independent risk ops / impact layer, not compete. |
| Pyth Network | First-party price feeds, confidence intervals, high-frequency data. | Strong data provider and ecosystem. | Use as reference/secondary source in cross-source risk view. |
| Chaos Labs | Risk oracles, price/risk/proof feeds, Aave/Ethena-style integrations. | Most direct strategic competitor. | Avoid enterprise oracle-provider fight; focus nimble incident intelligence + advisory pilot. |
| Gauntlet | Risk management research/services for DeFi protocols. | Credibility and relationships. | Position as focused oracle-risk ops console, not broad risk consultancy. |
| OpenZeppelin Defender Monitor | Contract/event monitoring and ops tooling. | Can cover generic alerts. | Oracle-specific incident reasoning, blast-radius mapping, replay, and risk memo layer. |
Copyability objection: UI, alert rules, and basic feed tables are easy to copy. Defensibility must come from curated incident corpus, dependency map quality, replay methodology, and trusted operator workflow.
7. Business model & pricing
Pricing is assumption, not forecast. Start with paid pilots and custom reporting; only later package SaaS.
| Offer | Who | Price assumption | What is included |
|---|---|---|---|
| Validation Brief | Risk delegates / small protocols | $500–$1,500 one-off | One protocol/feed dependency audit + incident replay memo. |
| Pilot Watchtower | Small protocol / vault operator | $500–$2,000 / month | Tracked feeds, weekly brief, alerts, manual blast-radius notes. |
| Risk Console Beta | Protocol with real TVL/OI | $2,000–$7,500 / month | Dependency map, incident explorer, policy simulation, governance-ready reports. |
| Enterprise Defense Simulation | Large protocol / risk team | Custom $10k+ / month or retainer | Custom policy packs, replay benchmarks, integrations, SLA. Later only. |
Revenue sequencing
Do not wait for self-serve SaaS. Sell 3 paid audits/briefs first, then convert repeat pain into product.
8. Go-to-market plan
Founder-led GTM, not ads
- Compile 30 target teams: 10 lending, 10 perp/synthetics, 10 vault/strategy/risk delegates.
- Prepare one sharp demo: “LINK/USD deviation event → affected markets → suggested response → replay evidence.”
- Offer a paid “Oracle Risk Exposure Audit” for one feed/protocol pair.
- Publish 3 public incident teardown posts to establish credibility.
- Use Chainlink ecosystem thesis and >60k LINK context as credibility, but do not oversell it as partnership.
Demo script
- Start with a concrete incident, not product features.
- Show raw source anomaly.
- Show protocol exposure.
- Show action menu and confidence caveat.
- Ask: “Would this have changed your incident response? Who owns this today? What would you pay to avoid manual triage?”
9. Product roadmap tied to revenue
| Phase | Build | Revenue gate | Stop if |
|---|---|---|---|
| Days 1–14 | Incident corpus + 1 HTML/demo + 20 outreach targets. | 5 discovery calls booked. | No one agrees oracle risk is owned/budgeted. |
| Days 15–30 | Watchtower prototype + manual brief format. | 1 paid audit or explicit LOI. | Only “interesting” feedback, no weekly-use signal. |
| Days 31–60 | Dependency map v0 + 3 replay cases. | 2–3 paid pilots. | Blast-radius mapping cannot be made trustworthy. |
| Days 61–90 | Risk Console beta + policy simulation. | $2k+ MRR or service revenue. | Ops burden exceeds willingness-to-pay. |
10. Financial model assumptions
| Scenario | 6 months | 12 months | 18 months | Interpretation |
|---|---|---|---|---|
| Conservative | 0–2 paid briefs, $0–$2k MRR | $2k–$6k MRR | $5k–$12k MRR | Useful as niche service, not SaaS yet. |
| Base | 2–3 paid pilots, $2k–$6k MRR | $10k–$25k MRR | $25k–$60k MRR | Small B2B risk-tool business forming. |
| Upside | 1 enterprise beta + briefs, $8k–$15k MRR | $40k–$100k MRR | $100k+ MRR | Requires trust, integrations, and clear ROI proof. |
Cost buckets
- Data/API costs: Chainlink/Pyth/free sources first, paid sources only after buyer proof.
- Engineering: dashboard/backend/detectors/dependency graph.
- Research: incident corpus, protocol mapping, quality control.
- Sales: founder-led calls, content, demo prep.
- Legal/compliance: before claims around attestation or action recommendations.
11. Funding & treasury options
Use cash/light infra first. Do not sell or risk LINK treasury for unvalidated product build. Treasury is strategic credibility and customer-zero story, not default runway.
| Option | Use | Verdict |
|---|---|---|
| Self-funded micro build | 30–60 day prototype and outreach. | Best first move |
| Paid audit/service first | Fund development from customer pain. | Best validation |
| LINK builder reserve | Only for proven, small, capped expenses. | Guardrail required |
| External fundraise | Only after 3–5 paid pilots or strong LOIs. | Later |
| Leverage/collateralize LINK | Finance product before PMF. | Do not do |
12. Risks and mitigations
| Risk | Why it can kill the business | Mitigation |
|---|---|---|
| False positives | Customers lose revenue or ignore alerts. | Confidence bands, severity tiers, replay benchmarks, manual approval. |
| False negatives | Tool misses the only incident that matters. | Clear scope, source coverage labels, no safety overclaim. |
| Data licensing | Cannot redistribute/reference paid data. | Use official/free data first; legal review before paid feeds. |
| Competitor absorption | Chaos/Chainlink/Gauntlet can copy simple features. | Own narrow workflow + incident corpus + customer-specific mapping. |
| Service trap | Every customer needs custom work, margins collapse. | Turn repeated custom work into templates; cap pilot scope. |
| Trust burden | Risk teams won’t trust new vendor. | Start as decision-support/reporting, not execution authority. |
13. 90-day execution plan
Week 1–2
- Build incident corpus v0: Mango, Synthetix sKRW, Cream, TWAP cases, Chainlink heartbeat/market stress cases.
- Create target list and outreach script.
- Define 10 feeds/assets and 5 protocol dependency examples.
Week 3–4
- Prototype Watchtower detectors with raw event logging.
- Create weekly Oracle Risk Brief template.
- Run 5 demos; ask direct WTP questions.
Month 2
- Add dependency graph v0 and blast-radius confidence badges.
- Sell 1–2 paid audits.
- Convert repeated audit format into product UI.
Month 3
- Policy replay module for 3 incident classes.
- Beta Risk Console for 2 pilots.
- Decision gate: productize, service-pivot, or stop.
14. KPI dashboard
| KPI | Target before “real business” claim | Why it matters |
|---|---|---|
| Qualified discovery calls | 15+ | Validates buyer and pain. |
| Paid pilots / audits | 2–3 | Validates willingness to pay. |
| Weekly active pilot usage | 70%+ of pilots | Validates habit/retention. |
| Alert precision | >80% “useful / not noise” pilot rating | Prevents alert fatigue. |
| Incident replay coverage | 10+ high-quality cases | Builds moat and confidence. |
| Mapping confidence | Every blast-radius claim has source/confidence | Prevents fake intelligence. |
15. What must be true
- Oracle risk is owned by a named person/team with budget.
- Manual triage cost/reputation risk is high enough to pay for tooling.
- Customers trust a new vendor as decision support if evidence is transparent.
- Dependency maps can be maintained without impossible manual burden.
- Incident replay creates enough differentiation versus generic alerts.
- Team can resist building broad Chainlink platform before sales proof.
16. Bottom-line recommendation
Recommendation: Proceed for 30 days as a paid-validation sprint. Build only what is needed to sell an Oracle Risk Exposure Audit + Watchtower pilot. Do not build a full SaaS, do not promise auto-action, do not spend heavily on paid data until buyer proof exists.
Immediate next actions
- Compile incident corpus v0 and protocol/feed map v0.
- Create one demo scenario from raw event to recommended action.
- Book 15 discovery calls with lending/perp/vault/risk contacts.
- Offer paid audit first; use revenue to decide productization.
Artifact generated by Lưu Bị on 2026-05-12. Static HTML is self-contained and dashboard-embedded. The plan deliberately labels assumptions and refuses unverified traction claims.