LiveProduction foundation
14-module foundation
Screening, trust operations, Scout, graph, founder/team, signals, memos, maps, funding, screeners, analytics, valuation, marketplace, and data products now have production surfaces.
Venture Decision OS
From deck screening to a defensible venture intelligence workflow.
Startup Screener now has a production Venture OS foundation around the evidence graph: the durable record of decks, claims, reviewer decisions, annotations, benchmarks, signals, sources, and human judgment that explains why a venture decision was made.
North-star metric
Evidence-backed decisions completed per organization per month
Reduce time from raw startup material to evidence-backed decision.
Attach citations, confidence, source provenance, and model metadata to AI-assisted outputs.
Turn decks, reviewer decisions, annotations, signals, and benchmarks into reusable venture intelligence.
LiveImport-ready
Provider-neutral records
CSV/JSON imports and internal evidence are live now; licensed private-market data providers and CRM sync remain governed adapter paths.
LiveDemo-safe
Public preview
A demo-safe preview shows all modules with sample records, graph nodes, analytics, valuation caveats, marketplace governance, and data products.
GovernedControlled
Bounded intelligence
Predictive and valuation examples carry confidence, sample size, calibration status, limitation text, and no investment-advice framing.
Super-app strategy
Production foundation live, governed adapters still explicit.
All 14 modules now have production foundations in the authenticated workspace and API surface. Public examples use demo-safe data. External private-market data, CRM sync, webhooks, and other third-party adapters remain governed integration records until licensing, provenance, scopes, and entitlements are configured.
Current wedge: Screening OS with evidence-linked reports and cohorts.
Compounding layer: Venture Graph, Scout, decisions, signals, and benchmarks.
Platform layer: APIs, plugins, integrations, data products, and governance.
Module foundations
All 14 modules have deployed production foundations.
All 14 modules now have production foundations in the authenticated workspace and API surface. Public examples use demo-safe data. External private-market data, CRM sync, webhooks, and other third-party adapters remain governed integration records until licensing, provenance, scopes, and entitlements are configured.
Core live2 modules
Intelligence live5 modules
Platform live7 modules
Venture Constellation
Semantic evidence graph with progressive disclosure.
12 nodes / 9 edges. Filter by entity, search evidence, then inspect why each node matters.
company86% confidence
SignalForge
Strongest match when the thesis prioritizes industrial workflow automation, measurable customer ROI, and evidence-backed enterprise conversion.
Why this matters
SignalForge is a high-confidence company node with 1 source and 2 graph relationships. It matters because downstream screening decisions become more defensible when this node stays tied to permission scope, citations, and reviewer-visible context.
Sources
1
Edges
2
Scope
public demo
Connected relationships
supports - 86% confidence
maps to - 70% confidence
Evaluation harness
Future-readiness checks are visible, not implied.
No accessible workspace reports were evaluated, so the harness is showing public demo and roadmap checks.
Sample evidence grounding
passDemo Scout candidates include cited sample-report evidence with freshness, permission, confidence, and origin metadata.
Workspace calibration data
warnA stronger evaluation run requires completed workspace reports with human decisions and feedback labels.
Conceptual architecture
A modular system above private-market data.
Personas -> Super-App Shell -> Product Modules: Screen | Scout | Profiles | Signals | Market Maps | Memos | Exports | Admin -> Platform APIs: BFF | public API | webhooks | plugin SDK | entitlement checks -> Domain Services: identity | intake | entity resolution | scout | scoring | audit -> Data Foundation: DB | object storage | search | vector store | graph | warehouse -> AI Layer: model gateway | prompt registry | eval harness | provenance ledger
01Personas
02Super-App Shell
03Modules: Screening OS, Scout, Venture Graph, Founder/Team Intelligence, Signals, Market Maps, Advanced Screeners, Decision Memos, Exports/Data Products, Admin/Governance
04Platform APIs: BFF, versioned public API, webhooks, plugin SDK, adapter contracts, entitlement checks
05Domain Services: identity, intake, entity resolution, evidence ledger, scoring, Scout, signals, memos, audit, policy
06Data Foundation: relational database, object storage, search, vector index, graph projection, warehouse
07AI/Eval/Governance Layer: model gateway, prompt registry, tool registry, evaluation harness, human labels, provenance ledger, policy engine
Evidence contract
Every claim needs provenance.
CompanyFounderSignalSourceClaimMarketDecisionBenchmarkIntegrationEntitlementConsentPolicyAuditEvent
The current report surface already exposes citations, confidence, provider metadata, fallback markers, and reviewer annotations. The Venture OS layer now extends that provenance model to Scout, market maps, founder profiles, signals, memos, APIs, and benchmark products.
Evidence contract
claimsource idsource typecaptured timestampconfidencepermission scopemodel or reviewer origin
Persona value flows
One evidence graph, different workflows.
Founder
Revise deck evidence, track readiness, and opt into sharing paths.
Investor
Convert a thesis into a ranked shortlist and diligence memo.
Accelerator
Standardize scoring, decisions, mentor notes, and exports across cohorts.
Corporate
Map buy/build/partner options with cited startup evidence.
Analyst
Maintain reusable screens, market maps, and decision-ready deliverables.
Admin
Control retention, permissions, providers, audit, integrations, and data policy.
Platform interfaces
APIs and adapters remain governed by provenance.
Evidence Graph API
foundationExpose canonical nodes, edges, claims, sources, decisions, and permission scopes.
Every node and edge must preserve source, timestamp, confidence, permission scope, and origin.
Scout API
existingAccept thesis input and return ranked results with rationale, gaps, freshness, and cited evidence.
Reviewer access is required; results must not cross organization or cohort boundaries.
Signals API
roadmapPublish company events, freshness, source, confidence, and alert subscriptions.
External signals remain gated until licensing, freshness, and source provenance are configured.
Market Map API
roadmapReturn market taxonomy, company clusters, competitors, emerging spaces, and map exports.
Maps should disclose evidence coverage and not imply exhaustive market coverage.
Decision Memo API
roadmapGenerate editable cited memos for diligence, selection, rejection, mentoring, and partnerships.
Memos cannot introduce uncited factual claims and should separate human decisions from AI drafts.
Webhooks
roadmapNotify approved integrations when reports complete, decisions are created, signals fire, or exports are ready.
Payloads require signatures, idempotency keys, retries, and scoped delivery.
Plugin SDK
gatedAllow governed UI and workflow extensions inside the super-app shell.
Plugins need sandboxing, scope approval, audit logging, and tenant-level enablement.
CRM and data-provider adapters
gatedSync approved evidence to CRMs and ingest licensed external market, company, funding, and people data.
Adapters can enrich canonical entities but cannot overwrite uploaded or reviewer provenance.
AI governance
Future capabilities are explicitly gated.
Production foundations are live. Claims that depend on licensed external data, calibrated prediction, or third-party sync remain gated by provenance, entitlement, and adapter governance.
No displayed factual claim appears without cited evidence or an explicit missing-evidence marker.
Human decisions remain stored separately from AI-generated scores and recommendations.
Provider, model, fallback, warning, confidence, and source metadata stay visible downstream.
Predictions, valuations, and benchmarks remain gated until sample size, calibration, and bias checks are defensible.
External adapter records must keep source freshness, permission scope, and origin metadata.
Key risk gates
Private-market data ambition exceeds current resources.
Phase licensed adapters after provenance, entitlement, and entity-resolution foundations.
AI scores become overtrusted.
Keep citations, confidence, warnings, model metadata, fallback status, and human decisions visible.
Benchmarks mislead at low sample size.
Preserve minimum-sample gates and benchmark confidence labels.
External adapters create data leakage.
Require adapter scopes, source-level permissions, audit logs, data-use policies, and admin approval.
Scout foundation
The first Scout workspace is deterministic and provenance-first.
The reviewer-only Scout flow accepts a thesis and ranks completed workspace reports the reviewer can access. Each result returns fit score, confidence, decision hint, gaps, and cited source freshness. External provider adapters remain intentionally gated behind licensing and governance work.
Example thesis
AI workflow automation for industrial manufacturing.
Expected top result
SignalForge, ranked for traction evidence, industrial AI fit, and diligence-ready customer proof.