Private AI assistant for small business
A private AI assistant for small business is an internal tool that answers questions from your own company's knowledge — your pricing logic, your vendor decisions, and the way your team actually runs jobs. The owner and staff ask it questions in plain language and get answers drawn only from the documents the business has uploaded, each one cited back to its source file.
It is not a customer-facing chatbot and not a lead-capture widget. It is a reference your own people use instead of interrupting the owner, and in production nothing it holds is sent to a public AI service.
How is it different from ChatGPT Team?
ChatGPT Team is a general-purpose chatbot with a shared workspace. It answers from a large public model plus whatever someone pastes into a given conversation, and its knowledge of your business ends when that chat is closed. A private AI assistant for small business works the other way around. It is grounded in a fixed library of your documents — your price book, your vendor list, your standard operating procedures — and every answer is retrieved from that library and cited back to the file it came from, so staff never have to know which document holds the answer or paste it in first. Access is scoped to your business, so one company's pricing cannot surface in another's workspace. In production the assistant runs against a privately hosted model: your documents, and the questions asked about them, are not sent to a public AI service and are not visible to a model vendor.
How is it different from a chatbot builder?
A chatbot builder is a tool for creating a scripted assistant that talks to your customers. It lives on your website, follows conversation flows you design, and exists to deflect support messages or capture leads. A private AI assistant for small business is pointed inward. Its users are the owner and the staff, not the public, and its purpose is to answer the internal questions that today interrupt the owner: how a job prices, which supplier to call first, what the warranty actually covers. There are no flows to build and no intents to train. You upload the documents that hold your operating knowledge, and the assistant retrieves answers from them on demand. Because it is an internal reference rather than a public endpoint, it is never the front door to your business, and it never collects or qualifies customer contact information.
Do I need a developer to set it up?
No. Setup is an onboarding step handled for you, not an engineering project. Your workspace is provisioned, your staff accounts are created, and your first documents — a price sheet, a vendor list, a procedures file — are uploaded and indexed so the assistant can answer from them. After that, keeping it useful is ordinary office work: when a price changes or a policy is updated, someone uploads the new version and the old content is replaced. Staff use it through a normal web login and ask questions in plain language, the same way they would type a message. There is no model to configure, no code to write, and no infrastructure for you to run. The one thing that stays with you is judgment about what belongs in the library — which documents represent how the business actually operates — and that is the owner's call, not a developer's.
Is my company's data used to train AI models?
No. In production your documents, and the questions asked about them, are never sent to a public AI service and are never used as training data by a model vendor. The assistant runs against a privately hosted model, and retrieval happens inside your own isolated workspace. Your files are stored for one purpose — answering your team's questions — and are not pooled with other customers' data or exposed to them. This is enforced in the system's design rather than left to a settings page: real business documents are blocked from reaching any public AI service, workspace isolation is checked on every request, and every question and change is written to an audit log the owner can review. If a document is removed, its content is deleted from the index and stops appearing in answers.
What does it cost to get started?
Getting started is a fixed onboarding engagement followed by an ongoing subscription. The onboarding covers provisioning your private workspace, creating staff accounts, and loading and indexing your first set of documents so the assistant returns useful answers from the start. The subscription covers hosting the private model, running retrieval, storing your documents, and support. Pricing scales with the size of your team and the volume of documents and questions rather than with a per-seat public-chatbot model, and there is no separate charge for underlying public AI usage because production does not use a public AI service. Exact figures depend on your trade, your document library, and how many staff will use it, so pricing is quoted per business after a short conversation about scope. There is no long build phase to pay for before the assistant is answering questions.
Put what only you know where your team can reach it
Upload the documents that hold how your business prices, sources, and runs jobs. Your staff ask questions in plain language from the portal and get your answer, grounded in your files and cited — while every document stays inside your private workspace, never a public AI service.