⚠️ SCOPE SUPERSEDED: The M2 team-competency-assessment approach described in this ADR (ProfileHarvester, ClaimVerifier, DigitalPresenceAnalyzer, TrackRecordResearcher) has been superseded by ADR-0025, which redefines M2 as an Unfair Advantage audit. This document is retained for historical context and will not be implemented as originally designed.

ADR-0013: M2 Team Implementation Navigator

Status

accepted

Context

M1 (Investment Navigator, ADR-0009-m1-product-vision) answers “is this niche worth entering” but does not assess whether a specific team can actually implement the opportunity. Founders and investors need a complementary tool that answers “can this team pull it off?” Early consultations revealed that the highest-value differentiator would be Combined M1+M2 mode: integrated niche + team fit assessment, which no competitor currently offers.

Competitor landscape:

• Perplexity / ChatGPT Deep Research — answer questions, no structured team assessment.
• Consulting due diligence — $15K-$50K, 2-4 weeks per engagement.
• Existing team assessment SaaS — checklist-based, no verification layer.
• No existing product combines niche analysis with evidence-based team assessment in a single verdict.

Governance risk: team assessment touches third-party reputation (the assessed team), triggering Constitution Law 3 (Reputation Over Speed), Law 5 (Human Veto on Irreversible/Legally Sensitive), Law 9 (Default to Safety). Without explicit framing, an AI-driven assessment could produce defamatory or legally non-compliant output (GDPR, 152-ФЗ, PDPL).

Decision

Add Module M2 — Team Implementation Navigator to the Synth Nova roadmap as a strategic document, ratified now; implementation sprint deferred.

Core commitments:

1. Evidence-Based Team Assessment. All claims are either verified via public sources (under explicit team consent) or labeled “not verifiable from public sources.” No claim accepted at face value. Governed by EvidenceVerificationPolicy.
2. Delta-based discrepancy classification. <10% normal variance, 10-30% yellow flag, >30% red flag, unverifiable grey flag. Neutral language always (“discrepancy detected,” never “founder lied”) — protects assessed reputation per Law 3.
3. Consent-only model (v1). Team members authorize specific platforms per session. No scraping without consent. No access to private/locked accounts. GDPR Article 6 lawful basis = consent.
4. v1 scope: LinkedIn (user-provided URLs/PDF exports), GitHub (API), Crunchbase (API). v1.1: Telegram via TGStat (РУ market). v2+: broader social media via official Business APIs.
5. Combined M1+M2 mode is priority. Standalone M2 is a subset of Combined and emerges automatically when M1 context is skipped. Combined is the moat; Standalone is commodity.
6. Implementation gate: sprint begins only after M1 has been tested by ≥3 external users and stabilized based on feedback. Strategic asset ratified now; code later.

Alternatives Considered

Option A: Skip team assessment, focus only on M1

• Pros: narrower scope, faster M1 polish.
• Cons: leaves half of the investment question unanswered (“is the team capable?”). Reduces product value and differentiation. A founder or investor needs both signals.

Option B: Build as external scraping service without consent

• Pros: no friction for end-user; auto-populates team profiles.
• Cons: GDPR / 152-ФЗ / PDPL violation; reputation risk (doxxing framing); Constitution Law 5 (legally sensitive) and Law 9 (default to safety) violation. Platforms (LinkedIn especially) actively litigate against scrapers. Non-starter.

Option C: Build generic team checklist tool

• Pros: simple; cheap; ships fast.
• Cons: commodity. Competitors already exist. No differentiation. Doesn’t earn trust — a checklist is just self-reporting, which investors already distrust.

Option D: Evidence-Based M2 with consent-only model and Combined integration ← chosen

• Pros: establishes a defensible product moat (M1+M2 Combined is unique); consent-only model is legally defensible in all three jurisdictions (EU, RU, UAE); evidence verification layer earns investor trust; reuses M1 infrastructure (CEO, Director, Judge, Policy Engine, python_calc).
• Cons: consent friction (user must provide URLs and checkbox); verification takes additional time and API calls vs. pure self-report; Crunchbase Basic API may cost $299/mo when first triggered.
• Why chosen: directly addresses the gap (investment decision = niche + team), complies with Constitution Laws 3/5/9 by construction, and the Combined mode becomes the strongest differentiator in the category.

Consequences

Positive:

• New strategic asset in manifest: 07-Roadmap/Team-Implementation-Module.md.
• New operational policy: 05-Rules/EvidenceVerificationPolicy.md.
• 4+ new agents planned for implementation sprint (ProfileHarvester, ClaimVerifier, DigitalPresenceAnalyzer, TrackRecordResearcher) — ADRs deferred to sprint time.
• Combined M1+M2 mode is positioned as the product’s unique differentiator vs. Perplexity / ChatGPT / consulting.
• Law 3 (Reputation Over Speed) operationalized for third parties via neutral-language rules in the Policy.

Negative / Trade-offs:

• Anthropic API costs will increase (additional agent calls per team assessment) — smaller than M1, but non-zero.
• Potential need for paid APIs (Crunchbase Basic $299/mo) triggered by actual usage, not pre-emptively (per IntegrationTriagePolicy / ADR-0011-integration-triage-policy).
• Legal review recommended before first external use of M2 (GDPR / 152-ФЗ / PDPL compliance of consent form).
• North Star metric extends: “80% autonomous closure” applies to M2 tasks too — by Q4 2026, team assessments should achieve same autonomy rate as M1.

Mitigations:

• Implementation gate (M1 tested by 3+ external users first) prevents premature resource commitment.
• Consent-only model from v1 avoids legal risk.
• Integration Triage Policy gates paid API adoption on observed pain point, not speculation.

Follow-ups

• Consent form draft (legal review required before M2 launch).
• Detailed per-agent ADRs created during M2 implementation sprint (ProfileHarvester, ClaimVerifier, DigitalPresenceAnalyzer, TrackRecordResearcher).
• Judge calibration for M2 output quality (new reference set needed).
• M1+M2 Combined flow prototype (Director-level decision logic).
• Evaluate Crunchbase Basic API when first real team assessment requires it.
• TGStat integration (v1.1) triggered by first РУ market test.
• Constitutional review: does Law 3 (Reputation Over Speed) require additional safeguards for assessed third parties?

References

• Constitution — supreme governance (Laws 3, 5, 6, 7, 9 directly apply).
• Team-Implementation-Module — M2 strategic document (07-Roadmap).
• EvidenceVerificationPolicy — operational rules for M2 (05-Rules).
• Niche-Evaluation-Module — M1 strategic document; M2 integrates in Combined mode.
• ADR-0009-m1-product-vision — M1 product vision.
• ADR-0011-integration-triage-policy — governs paid API adoption (Crunchbase, TGStat).
• ADR-0012-constitution — Constitutional foundation.
• ObservabilityContract — retention defaults (overridden for raw profile data).
• DecisionRights — consent escalation.
• Decision-Log — ADR index.