New: see what an AI assistant sees when it looks at your site. Run the check

neural partners

Every agent, one source of truth.

A shopping assistant on your storefront, Axon inside Core and the outside assistants that query you over MCP — all reading the same catalog, content and context, all logged like an employee, and none of them handing your customers’ data to anyone.

The shift

Five AI tools, and no idea what any of them know.

  1. 01

    Every tool brings its own bot.

    A chat widget, an email assistant, a copy tool — each with its own copy of your catalog, each slightly wrong.

  2. 02

    Customer data leaks by default.

    Paste a customer list into a chatbot and it lives somewhere else now, under somebody else’s retention policy.

  3. 03

    Nobody manages the agents.

    They act for you with no policy, no log and no budget — things you would never allow a new hire.

The Design Assistant on a lighting storefront, on a laptop: eight outdoor sconces and lanterns saved to a project, beside a chat where a shopper asks “How about black?” and the assistant suggests pieces under $500 and says stock should be confirmed with the showroom.

Privacy first

Customer data never leaves the app.

Agents inside Core can see a customer to help them. Anything connecting from outside sees your store, and nothing about the people who shop it. Zero data retention, by design.

Inside Core

  • Axon, for your team
  • The Design Assistant, for shoppers

Can read

  • Customer records
  • Orders and projects
  • Catalog and bookings
  • Your content and context

Your own agents see what your staff can, to do the same job.

Outside, over MCP

  • Claude
  • ChatGPT
  • Any assistant you connect

Can read

  • Catalog and bookings
  • Prices and availability
  • Your published content

Never

  • Names, emails or phone numbers
  • Orders or customer history

Outside assistants get the store, never the customers in it.

In the product

What's built in

What sits underneath.

  • PII stays in the app

    Agents inside Core can see customer records to do their job. Anything connecting from outside over MCP sees the store, never the people in it. Zero data retention, by design.

  • One router, the right model

    Each task goes to the model that fits it — Claude’s family directly or through AWS Bedrock, others through OpenRouter — so simple work stays cheap and hard work gets the best.

    Partners →
  • Agents managed like employees

    Every agent works under the same policies, logging and audit as your staff, with its AI usage metered against a limit.

    Automation →
  • Agents that ask each other

    When the Design Assistant needs something only the business knows, it asks Axon instead of guessing.

  • Your content, queryable

    Brand guidelines, voice, how-tos and manufacturer training uploaded to Core become what every agent answers from.

    Business Data & Context →
  • Products, bookings and listings, unified

    Products, services and materials enriched by AI into one structured catalog that every agent reads the same way.

    Products & Bookings →
  • Decision models for the fast calls

    High-volume yes-or-no and ranking calls go to decision models — TypeSafe’s Jev — which are faster, cheaper and more predictable than a full language model.

  • Readable by outside agents

    UCP and ACP profiles, WebMCP tools and semantic search, so the assistants your customers use can read your store.

    Agentic Commerce →
  • New protocols, folded in

    Agent standards move monthly. We track them and fold in the ones that matter — what we are watching is in Resources.

    Resources →

Questions

Asked before you ask.

Can outside AI assistants see my customers?

No. Assistants connecting over MCP see your catalog and content. Customer records stay inside Core, where only your own agents and staff can see them.

Which models do you use?

It depends on the task. Our router sends each job to the model that fits it — Claude’s family directly or through AWS Bedrock, and other models through OpenRouter where they are more efficient.

How do I know what an agent did?

Every action is logged and audited the way a staff member’s would be, and AI usage is metered by source against your plan.

Keep exploring

More in Neural Core.