SlabOS workflow article · 33Documentation review · September 15, 2026

Connect an owner’s AI through MCP

Give an authorized AI client a defined read-only view of shop analysis.

SlabOS’s MCP page documents an owner-only connection for an external AI client. It describes a read-only analytical interface, shop isolation, expiring credentials, revocation and a visible call history. This is a separate access path from the built-in assistant.

The important documentation is the interface boundary. The public page says the connection returns analysis rather than bulk customer, quote or job rows, and that it cannot create or change records. An AI client receiving results is an external destination chosen by the owner; the client’s own handling of those results remains relevant to that choice.

Inputs, outputs and an example to evaluate

Inputs: An owner-authorized AI client and a question about the shop’s business data.

Outputs: Read-only analytical results within the documented MCP access boundary.

Compare one permitted aggregate answer with its underlying report, then show the owner-visible call history without exposing credentials.

Scope to understand

The public page describes the access contract. A reproducible allow/deny test is still needed to substantiate each security boundary in a published report.

Public capability guides and product sources

Published by SlabOS. Documentation review: September 15, 2026. Basis: Public product and operating documentation. This article is a workflow explanation, not a newly performed product test.

Where this fits in the SlabOS workflow

Follow this capability into the connected quote-to-install journey:

Review results & plan the next job: Feed reviewed results back into pricing, estimating, scheduling and the next customer conversation.

Can I connect my own AI assistant to SlabOS?

SlabOS provides an owner-facing MCP connection for supported access to shop context from an owner-selected AI. The public MCP page and capability explanation describe that connection within the SlabOS permissions model. It gives the owner another way to work with supported business information alongside the platform’s own AI features.

What should I evaluate before using an AI assistant with shop data?

Start with a supported question that has a known answer in the shop’s records. Compare the assistant’s response with those records and review the connected account’s permissions. Confirm which tools and actions the current connection exposes rather than assume unrestricted access. This helps distinguish the available connection from a promise that every AI model can perform every shop task.

Current product capabilities

The public MCP connection is owner-only and read-only. It returns analytical results rather than bulk job, quote or customer rows, with expiring and revocable credentials and a visible call history. Connection details stay inside the owner’s account. Read the MCP for owners page →

The built-in analyst answers questions about the shop’s invoice balances, job status, sales pipeline and materials from its own records. The same public page describes reviewed vendor PDF, CSV and Excel imports for colors, thicknesses and prices. Read the AI CFO page →

Explore the capability in context

Continue into the related public explanations for inputs, outputs and scope.

Role-based access for countertop shop teams

Connected product pages and tools

MCP for owners
Review the public scope and boundaries of connecting an owner-selected AI to SlabOS.
AI CFO
Explore owner-facing questions about shop records, costs and reporting.
Features
Place this capability within SlabOS estimating, production, inventory and office work.
Pricing
Review the published commercial model and what is included; confirm current terms for your shop.