Private-equity portfolio management platform. One database, one system, one process.
OCTO is a private-equity investment and portfolio management platform that consolidates fragmented data, systems, and operating processes into a cohesive Investment Ontology and a single IBOR (Investment Book of Record). It normalizes inconsistent data from CRMs, financial and market-data providers, fund administrators, internal systems, documents, and third-party or open-source feeds into a governed source of truth spanning prospects, funds, portfolio companies, LPs, and GPs.
The platform provides a configurable front end for deal sourcing, portfolio monitoring, analytics, reporting, and operational oversight. Firms can tailor workflows, permissions, metrics, models, and output artifacts to their business model while preserving a standardized organization-wide methodology. Outputs include tear sheets, portfolio-rebalancing analyses, valuation models, Investment Committee reports, disclosure materials, and reports for Limited Partners. Governed data can also be accessed through spreadsheets and external systems.
OCTO addresses the absence of a centralized view, duplicated processes across teams and locations, overlapping tools, inconsistent analyst methodologies, and discrepancies between vendor datasets. The result is simplicity, control, efficiency, and scalability.
Vendor-neutral by design. OCTO does not replace fund administrators, custodians, CRMs, or data providers. It operates above them in a multi-vendor environment and isolates provider-specific formats within ingestion adapters. Consolidation occurs at the data and process layer, so firms can add or replace providers without changing the core platform.
| Capability | Description |
|---|---|
| IBOR | Append-only transaction/cash-flow ledger; positions, commitments, and drawdowns are derived — never written |
| Investment Ontology | Dynamic OWL/SHACL canonical model in Git, versioned with SemVer; supports monitoring, analytics, reporting, and governed access from the platform, spreadsheets, and external systems |
| AI data integration | Vendor-isolated adapters for CRMs, financial and market-data providers, third-party and open-source feeds; multimodal extraction from pitch decks, PDFs, and other documents; built-in validation, lineage, and governance |
| Deal sourcing | Configurable inbound screening, standardized prospect pipeline, AI-assisted DDQs, and source-grounded Investment Committee reports with human review |
| Reconciliation | Continuous validation of source systems against the IBOR; discrepancies surface as workflow tasks |
| Onchain ingestion | Read-only adapters for Solana (Helius: RPC, webhooks, finality probe) and Arbitrum (EVM): tracked addresses, finalized-only token flows into instrument_flow, holdings snapshots for reconciliation |
| Look-through | Recursive exposure aggregation across fund → deal → portfolio company hierarchies |
| Modelling and analytics | Metrics engine (IRR, TVPI, MOIC, DPI) and low-code authoring for firm-specific metrics, ratios, forecasts, and valuation models |
| Personalized outputs | Configurable tear sheets, portfolio-rebalancing analyses, IC memos, disclosure materials, and LP reports |
| AI-driven applications | Proactive reporting, alert rules, natural-language data queries, memo/email drafting, news matching, and bespoke GP/LP interfaces |
| Permissions and collaboration | Fine-grained RBAC/ABAC controls for cross-functional teams, external users, entities, documents, fields, actions, and outputs |
| Security and governance | Data classification, encryption, lineage, retention, segregation of duties, tamper-evident audit, model governance, and controlled third-party processing |
| Self-hosted deployment | Supabase PostgreSQL, Auth, and Storage run on controlled infrastructure; the operator owns hardening, availability, monitoring, backups, PITR, upgrades, and disaster recovery |
| Workflow | Task assignment, approvals, and disclosure management; human approval gates outbound artifacts |
OCTO accelerates early-stage deal sourcing by enabling private-equity analyst and research teams to analyze inbound opportunities and manage prospects at scale. Firms can configure strategies, criteria, question libraries, report templates, and stage-transition rules while retaining a unified organization-wide methodology. This combines local flexibility with consistent analysis, evidence requirements, review controls, and decision records across teams.
Multimodal models extract text, tables, chart values, entities, and claims from pitch decks, PDFs, and other documents. The platform maps these results into the Investment Ontology and combines them with structured internal and third-party data. Every extracted claim retains a citation to its source location, confidence, model version, and review status.
Users define versioned screening criteria by strategy, such as minimum revenue, five-year growth, permitted countries, sector restrictions, or document recency. The platform evaluates inbound opportunities as Pass, Conditional, Fail, Unknown, or Conflicting and recommends a pipeline transition. Analysts review results, request missing evidence, and may override recommendations with a recorded rationale. The platform never silently converts missing information into a failed criterion.
The platform suggests answers to approved Due Diligence Questionnaire libraries covering company vision, differentiation, market position, financial quality, management, principal risks, and mitigants. It also proposes company-specific questions based on the submitted materials, inconsistencies, evidence gaps, and existing answers. Suggested answers include source citations and remain unverified until an analyst reviews them. Users can edit responses, export unanswered questions, and upload follow-up materials for extraction and reassessment.
OCTO generates reports from approved, versioned templates populated with governed Ontology data. Structured grounding and answer-level citations reduce unsupported model output. Reports can include the investment thesis, market analysis, financials, valuation and return scenarios, diligence findings, risks, mitigants, and open conditions. Users review and manually edit every report before submitting it for approval, export, or email distribution.
See Deal sourcing workflow for stages, controls, and the prospect-to-investment conversion flow.
Modular monolith: Kotlin on Java 21, Spring Boot 3, Gradle, Flyway, and self-hosted Supabase PostgreSQL/Auth/Storage. Spring Boot remains the domain and API boundary; sensitive domain writes are never exposed directly to clients. See ADR-0001 for the platform architecture and ADR-0002 for deployment, security, backup, and operational responsibilities.
- Whitepaper — the governed book of record, end to end
- System design — HLA, HLD, LLD, E2E architecture
- Platform architecture and end-to-end flow
- Self-hosted Supabase architecture and operations
- Neo4j graph store — schema:
ontology/octo-investment.cypher - Ontology concepts — reference mapping
- Ontology design guidelines
- AI architecture — platform capability mapping
- TypeSafe fit assessment — proposed judgment layer
- Application surface — ontology-aware app model
- Deal sourcing, screening, due diligence, and IC workflow
- User workflows and application flow
- User interfaces
- User journey
- User experience principles
- Quantitative methodology
- Data security and governance
- Reliability — SLOs, error budget, production readiness
- xStocks fit assessment — proposed instrument data source
- Arbitrum fit assessment — settlement network and onchain data source
- Arbitrum ingestion design and megaplan
- Helius fit assessment — Solana onchain data source
- Dune fit assessment — multi-chain analytics context
- Alpha Vantage fit assessment — market data
- TradingView MCP fit assessment — market data, fundamentals, news, calendars
- System map
- The Machine-Native Economy — BlackRock Digital Assets Research, Sept 2026: machine-native money needs machine-verifiable records
- US 10,872,067 — Creating data in a data store using a dynamic ontology — Palantir patent describing the canonical-ontology-plus-parser architecture Octo applies to the IBOR
- Concept-Centric Software Development — Wilczynski, Gregoire-Wright, Jackson: Palantir's experience report on governing shared concepts; motivates Octo's SemVer-versioned ontology and CTO-owned change path
This repo follows the TANTIRA SDLC — see AGENTS.md. Highlights: work starts from a GitHub Issue, conventional commits, PRs under 400 lines, risk tiers T0–T3, no commits to main, no deploys by agents.
modules/
api/ Spring Boot entry point — REST boundary, Flyway wiring, actuator
ibor-core/ Append-only ledger: commitments, transactions, cash flows, valuations
lookthrough/ Recursive exposure aggregation across fund → deal → company
ingestion/ Source adapters and document staging (vendor isolation boundary)
recon/ Source-vs-IBOR comparison, discrepancy surfacing
analytics/ Metrics engine: IRR, TVPI, MOIC, DPI; low-code metric authoring
deal-sourcing/ Screening, DDQ assistance, IC report assembly
workflow/ Tasks, approvals, outbound-artifact gates
control-panel/ Unified inbox: alerts, recon items, approvals, AI proposals
ontology/ OWL/SHACL validation and Cypher drift checks — CI gate for ontology/
db/migrations/ Flyway migrations (append-only, migration identity)
ontology/ Cypher schema (octo-investment.cypher), OWL/SHACL mirror, sample graphs (T2, SemVer)
infra/ Dokploy Compose definitions and Supabase override
docs/ Architecture, ADRs, product and methodology documentation
The web UI is a separate frontend concern (ADR-0001); infra/docker-compose.yml consumes it as a published octo-web image.
Backend scaffold in place — ./gradlew check compiles all modules and runs tests.
- Wired into the running application (Kotlin paths covered by Testcontainers ITs):
- Ingestion: Helius onchain (RPC, webhooks, finality probe), the EVM scan service, onchain staging → promotion → recon, and market-data sync (Alpha Vantage).
- Operations API: tenancy and admin (
/api/v1/me,/api/v1/admin), assets, prospects and screening rules (deal sourcing), reconciliations, compliance, reports with schedules and release (workflow), the company brain, and the anonymous contact form. - Reports compute PE performance (
analytics) and the GL journal (ibor-core). Compliance reads look-through exposure reports. - Agents sidecar (ADR-0005): Python DeepAgents workflows, tested with pytest and evals — screening/DD, IC memo, LP report, company brain, compliance rationale, equity bridge, operating review, DDQ response, calibration. Each workflow is behind a flag that is off by default.
- SIWS wallet sign-in ships in the web UI. The deployed GoTrue still needs
GOTRUE_EXTERNAL_WEB3_SOLANA_ENABLED(backlog item 1).
- Implemented as libraries, not yet wired into the running application: IBOR position derivation (deal, commitment and token positions) and the IBOR reader, the Kotlin decision-model client and what uses it (document classification, claim-support assessment), the rest of the analytics suite (Brinson attribution, return and risk measures, factor exposure, the value-creation bridge, DCF), the path-sum look-through computation, and the ontology validation gate.
- Still open: the reconciliation triage loop.
- ADRs in
docs/adr/: 0006 accepted (Neo4j bolt stays tunnel-only), 0003 superseded by 0004, the rest still Proposed.