Frame a clearly fictional opportunity and the diligence question.
Engage now
A diligence-workflow review covering source provenance, criteria, assumptions and the committee record.
Inspect
Development prototype
FastAPI and Next.js application surfaces with PostgreSQL, vector-search, and Redis boundaries.
Still to validate
No investment return, predictive accuracy, or decision-quality outcome is asserted.
Validate licensed data coverage, model and score behaviour, reviewer consistency, and operating controls without inferring investment performance.
源码审查日期
The decision behind the system
Can the investment committee follow the reasoning?
A diligence-workflow prototype connecting deal intake, research, scoring, scenarios, investment-committee reporting and portfolio monitoring.
Kairos focuses on investment diligence and committee records. Scores and scenarios support human judgement; they do not establish predictive accuracy or returns.
One diligence path
Move from intake and triage into research, scoring and scenario work without losing the decision context.
Review and reporting
Investment-committee, portfolio and LP reporting surfaces organise findings for human judgement.
Replaceable providers
Data and model adapters, including a deterministic mock path, separate workflow evaluation from live-provider configuration.
Read the complete system walkthrough
Triage
Frame a clearly fictional opportunity and the diligence question.
Research
Assemble authorised research, making provenance and missing sources visible.
Challenge
Review scenarios and criteria without presenting a score as a calibrated forecast.
Record
Preserve the committee’s reasoning and monitoring questions. Live data needs separate configuration and rights.
访客及客户数据的 AI 处理限于通过受治理的欧盟端点使用 Mistral。所有者授权的 Helio、搜索及 Product Council 流程已启用。独立的公开来源编辑流程使用 Anthropic，并以受治理的 Mistral 作为备用；该流程不接收网站咨询、客户文档或访客对话。远程 AI 不可用时，确定性工具仍可使用。
源码证据
源码实际能够证明什么
A full-stack diligence-workflow prototype with deal, research, scoring, portfolio, and reporting surfaces. Its source demonstrates workflow architecture and provider abstraction; it does not establish investment performance.
详细技术快照以英文维护。
成熟度
Development prototype
证据审查
源码快照截至
产品与架构决策
Separate the browser workflow, API modules, persistence, retrieval, and model-provider layer.
Use a deterministic mock provider so workflow and scoring behaviour can be exercised without a live model.
Position model-assisted scoring as decision support whose assumptions remain visible to an investor.
可核验的证据
FastAPI and Next.js application surfaces with PostgreSQL, vector-search, and Redis boundaries.
Backend tests spanning deal, intelligence, diligence, portfolio, and report workflows.
Provider adapters and local container configuration for repeatable development setup.
声明边界
No investment return, predictive accuracy, or decision-quality outcome is asserted.
External data coverage and model output depend on the services and credentials configured by an operator.
The repository does not establish a live customer deployment or a production operating record.