What is a digital twin of a company?
A financial digital twin is a conversational surface over a single company, assembled from the market read, the knowledge graph, the name's track record, and the agent-readiness gate. Instead of searching, you converse with the name.
Converse instead of search
A digital twin lets you ask a name plain-English questions: how is it doing, who does it compete with, have you been right about it, is my agent clear to act on it. Each question routes to the right underlying engine and returns a grounded answer with the trust envelope.
The twin is not a new model on top of the stack. It is a thin conversational layer over engines that already exist: the market brain, the graph, Institution Memory, and readiness.
Discipline built in
A profile view is ledger-neutral, meaning it reads without recording a decision. Only act-oriented questions trigger a recorded readiness check, which itself carries the multi-model verification consensus and writes to the Trust Ledger.
A buy, sell, or hold question is never answered as advice. It is redirected to the data-readiness verdict and the read, with a note to decide using your own judgment or a licensed advisor. The twin is educational only.
How LyraMind implements it
The router is deterministic keyword intent-routing, so it is offline-safe and needs no LLM. It classifies a question into intents like peers, suppliers, ETFs, sector, macro, track_record, verification, read, risks, act, or advice, and answers from the matching engine. An LLM answerer can plug in later behind the same ask contract.
The profile card assembles the headline read, the graph summary of connections, the memory brief on the name's history, and a set of suggested questions, all wrapped in the trust envelope.
Related concepts
The twin sits on top of the knowledge graph (connections), Institution Memory and the Trust Ledger (track record), and the agent-readiness gate (act questions), which carries multi-model verification.
- A conversational surface over one company, not a new model
- Assembled from the market read, knowledge graph, track record, and readiness gate
- Deterministic keyword intent-routing; offline-safe, no LLM required
- Profile views are ledger-neutral; only act questions record a decision
- Buy/sell/hold questions are redirected to the readiness verdict, never answered as advice
- An LLM answerer can plug in later behind the same ask contract
Questions
Does a digital twin give investment advice?
No. A buy, sell, or hold question is redirected to the data-readiness verdict and the current read, with a note to decide using your own judgment or a licensed advisor.
Does asking the twin a question require an LLM?
No. The router uses deterministic keyword intent-routing and is offline-safe. An LLM answerer can be added later behind the same contract.
Does viewing a twin's profile record anything?
No. A profile view is ledger-neutral. Only act-oriented questions trigger a recorded readiness check.