Right knowledge in, more sales out.

An AI knowledge base is the layer of business content — websites, files, and articles — that an AI agent or chatbot uses to answer questions. Unlike a traditional knowledge base built for customers to read, an AI knowledge base is structured to be read by large language models. Knowledge hub is Text's AI knowledge management tool, built specifically to power AI Agent and Copilot with accurate, on-brand answers.

Traditional knowledge base software (like Zendesk Guide or HelpScout Docs) helps you publish a customer-facing help center. Knowledge hub doesn't publish anything — it's the source of truth your AI reads from. Whether your customers talk to AI Agent directly, or your human agents rely on Copilot for reply suggestions, Knowledge hub is what makes those answers accurate.

Three types of sources:

• Websites — up to 5,000 subpages per site, with automatic re-scanning every 5 to 365 days so your AI stays current.
• Files — PDF, Word, PowerPoint, TXT, TSV, CSV, and Markdown, up to 50 MB each (PDFs capped at 200 pages).
• Articles — custom text you write or paste directly, useful for internal policies or content that doesn't live anywhere else yet.

No. Generic RAG knowledge base tools like Pinecone or Weaviate require engineers to build the embedding pipeline, retrieval logic, and integration with your support stack. Knowledge hub is customer-support-native — connect a website or upload a file, and AI Agent and Copilot start using it immediately. No vector databases to manage, no code to write.

For AI Agent, Knowledge hub is the layer of business content the bot actually works from. When a customer asks a question, AI Agent retrieves the relevant content and generates an answer grounded in your business — not in the model's training data. Any chatbot you run on Text taps into the same knowledge layer in the same way. That setup is what makes Knowledge hub function as an AI agent knowledge base in the first place.

Copilot can work from a different slice of the same Hub. You can split sources by audience: internal-only material that should never reach a customer (escalation playbooks, agent-only SOPs, sensitive pricing logic) stays available to Copilot for your human agents, while the customer-facing layer (public policies, product docs, help articles) is what AI Agent and chatbots use in conversations with customers. One source. Two audiences, cleanly separated.

Answers are only as accurate as the sources you connect — that's why Knowledge hub treats source quality as the core feature. Because AI Agent and Copilot generate responses from your Knowledge hub content (a retrieval-based approach, not pure generative AI knowledge management), they cite real business content instead of inventing answers. If a question isn't covered by your sources, the AI escalates instead of guessing.

Files and articles update the moment you save changes. Websites re-scan automatically on a schedule you set — anywhere from every 5 days to once a year. The same fresh content flows everywhere it's used: AI Agent and chatbots pull from it in customer conversations, and Copilot pulls from it when suggesting replies and answering questions for your human agents. Update your pricing page once, and every AI touchpoint reflects it — no manual sync, no agents quoting outdated policy, no customers misled.

Yes. Knowledge hub functions as the chatbot knowledge base for AI Agent and any chat widget powered by Text. Add your sources once and every chatbot conversation pulls from the same up-to-date layer.