The digital twin of your business, built in four moves.
- 01Ontology
The nouns and verbs of the business, plant, order, customer, ships, supplies, as one model rather than a list of tables.
- 02Mapping
Every system and every handoff between a person and a system, drawn onto it, so the re-keying and the waiting are visible.
- 03Data unification
One record per real thing, however many systems spell it differently, with the merge reversible and every field pointing back to its source.
- 04Automation
The work that runs on the model once it is true: agents and platforms do the work on it, and a named person presses send.
How the twin gets built, and what it returns.
One live, permission-aware model, built in five steps: what your people know, what your systems hold, what each process costs, what can run on the model, and what comes back. Nothing trains on your data, and the model behind the agents is swappable.
- 1 Knowledge mapping
How the work actually gets done.
Who knows what, what is decided where, which customer pays late: most of it lives in people’s heads. We capture it in interviews and write it into the model as knowledge, each claim with who said it and when it was true. The Operational Audit is where this starts, and it is written in the same shape.
- 2 Data mapping
One record per real thing.
The systems and the organisational data are drawn onto the same model: the ERP, the accounts, the CRM, the shared drive, the mail. A customer is one customer, not five spellings in five systems, and every field points back to where it came from. Permission is resolved as the data comes in, so a reader with no right to the source sees nothing; it fails closed.
- 3 Financial mapping
What each process costs to run today.
Cost is laid onto every process: the labour, the slice of each licence it consumes, the overhead that comes with a desk. The business can see exactly what each thing costs to run, per run and per year, process by process. Each line carries its source, the same as every other claim in the model.
- 4 Automation
What runs on the model once it is true.
With the model true and costed, the expensive processes are in plain view and the work can move onto it. Agents work on the model and automation platforms act through it, and nothing reaches a client or posts to a ledger without a named person pressing send. 0 auto-sent is a number on the screen, and it stays at zero.
- 5 Return on investment
The same process, before and after.
The cost to run today from step three, the cost to run on the model, and the difference, with the number in the open. It is measured at the desk that owns the process, not estimated in the round. Clear return from AI, per process, per year.
Built to be read by the tools you already use.
Orchid is not another screen to open. It is the model that automation platforms and assistants are built to read from, each with the permissions of the person or system asking.
Automation platforms
Built to be the record these platforms act from: the customer, the contract, the tracker just gone Overdue, permission already resolved, one connection instead of six exports.
Make
Power Automate
Assistants
Built so the assistant a person already uses reads the twin with that person's permissions and answers from what the business knows, not from a guess.
ChatGPT
Gemini
Copilot
Perplexity
Meta AI
The ERP, in time
Once the twin reads a system fully, the path is to mirror it, shadow it, then replace it as the system of record: orders, stock and planning first, finance last.
The digital twin of the business
Four primitives and no fifth. Everything above reads from here.
- 01Entities
Customers, contracts, people, sites: one record each.
- 02Knowledge
Every claim carries its source.
- 03Trackers
What is outstanding, in one of five states, never maybe.
- 04Actions
Drafted on the model, sent by a named person.
Permission resolves at ingest and fails closed. Nothing trains on your data. The model underneath is swappable.