Build an AI chatbotwith AI.

OpenAI (GPT-4o / GPT-5). Default pick for general-purpose support and lead-gen bots. Cheaper per use, quicker to start replying, broadest model selection (4o-mini for cheap routing, 4o for quality, o1 for reasoning). The right default if you don't already have an opinion.

Anthropic Claude (Sonnet 4.6 / Opus 4.7). The right pick when conversation quality matters more than cost, sales chatbots, in-app product help, anything where tone and how it declines requests affect the experience. Claude also handles very long inputs (200K tokens), so it shines for "talk to my docs" use cases where you put a large knowledge base into the prompt.

Both, with routing. The mature pattern for serious chatbots: send short, simple questions to GPT-4o-mini at $0.15 per million tokens, and send complex or high-stakes questions to Claude Sonnet at $3 per million. WebTwizz builds the routing if you ask for it: you describe the rule ("if the message is over 500 characters or mentions account or billing, use Claude"), and the AI writes it.

Both integrations are ready in WebTwizz. Connect once and the keys are stored safely in your project, ready for your app to use.

If your chatbot needs to answer questions grounded in your own data (docs, support articles, product specs), you need AI search over that data, not just a long prompt.

The setup WebTwizz builds: your content is stored in Supabase in a searchable form, re-indexed whenever it changes, and before each answer the chat pulls the few most relevant pieces and hands them to the AI model. This is called retrieval-augmented generation, or RAG. It's how every serious doc chatbot is built.

For most B2B chatbots under 10K documents, keeping the search inside Supabase is the right call: one fewer service to run, and the search sits next to your logins and user data. Past that, you move to a dedicated search service like Pinecone or Qdrant. WebTwizz can connect either; Supabase is the default because it's simpler.

What you should NOT do: put your entire knowledge base into the chatbot's instructions on every message. Even with Claude's 200K token limit, you'll burn money and quality. Search first, then ask.