← Back
Nexel platform

AI needs more than access to your files.

For AI to work inside a business, it needs context: what things are, how they relate, who may access them, which rules apply and what happens next.

Nexel creates that operating layer across the systems your company already uses.

Keep the systems. Connect the context.

Nexel is designed to sit across the software already running the company rather than require another system-of-record migration.

Documents, communication and structured operational data can feed the environment while the source applications remain in place.

connect
  • Documents
  • Email
  • CRM
  • Finance
  • Practice systems
  • Databases
Nexel

Ingestion, identity resolution, permission mapping.

Build a model of how the business works.

Every industry operates around a set of real things and relationships.

A wealth firm has clients, accounts, entities, mandates, advice, holdings and obligations.

An accounting firm has clients, entities, periods, returns, transactions, advice and deadlines.

Nexel models these explicitly instead of asking a language model to infer the structure every time someone asks a question.

the model
Client
  • Accountholds
  • Entityowns
  • Mandateconstrained by
  • Advicereceived
  • Obligationsubject to
  • Documentevidenced by
  • Transactionrecorded in
  • Personrepresented by
For technical teams

Nexel uses declarative ontology schemas to define entity types, relationships, identity-resolution behaviour and permission semantics.

Resolve the context before the model sees it.

Relevant information is assembled around the entities and relationships involved in the workflow.

The model receives the context required for that piece of work rather than unrestricted access to an undifferentiated corpus.

Build workflows, not prompts.

A Nexel workflow defines:

  • what starts the work;
  • which context is required;
  • which rules apply;
  • what the model is being asked to produce;
  • what evidence must accompany it;
  • who reviews the result;
  • and what happens next.
a workflow, end to end
  1. Trigger

    What starts the work.

  2. Context

    The entities and relationships the work depends on.

  3. Analysis

    The rules that apply, and the task the model is given.

  4. Output

    The work product, and the evidence attached to it.

  5. Approval

    Who reviews the result, and what happens next.

Models are infrastructure.

The foundation model is important.

It is not the durable part of the system.

The business-specific assets are the model of the organisation, its data preparation, workflows, permissions, evaluations and operating history.

Nexel's architecture is designed so model choice can change without rebuilding those assets from scratch.

Control how AI is allowed to work.

workflow
Prepare annual review
May read
Client recordsRelated entitiesApproved source documents
May produce
Draft review
May change external systems
No
Approval
Partner required
Evaluation
Production threshold configured

Authority belongs to the workflow, not to the model.

Start with one workflow.

Find the first workflow