Built For Product And Engineering Teams
Sometimes the website widget isn't where you need the chatbot. You need it inside your own product, mobile app, internal tool or backend job. FastBots is an AI chatbot trained on your own content, exposed through a single REST endpoint that returns the bot's reply. Pair it with Zapier MCP or Make workflows and the same bot can act as an AI agent, not just answer questions.
curl -X POST https://app.fastbots.ai/api/public/chat \ -H "X-BOT-API-KEY: sk_live_••••••••" \ -H "Content-Type: application/json" \ -d '{"message": "What are your opening hours?"}'
One endpoint. Send a message, get your bot's reply and a chatId to keep the conversation going.
Providing AI Chatbots to Trusted Brands Around The World
Quick Answers
It's a single REST endpoint, POST https://app.fastbots.ai/api/public/chat, that sends a message to a FastBots chatbot and returns the bot's reply. The bot is trained inside the dashboard on your own website, documents and files, then exposed through the API to anywhere you can make an HTTPS call.
Product, engineering and ops teams who want a chat experience inside their own product, mobile app, internal tool or backend job rather than only on the FastBots website widget. It's also useful for adding bots to channels we don't natively support.
Train a chatbot in the dashboard, generate an API key from the bot's Settings page, and call the endpoint with the X-BOT-API-KEY header. There's a free plan to build and test on; see the pricing page for API availability and paid plans, which start at $39/month.
No, they sit side by side. The same trained bot can power the website widget, the direct chat link, supported channels like WhatsApp and Slack, and your own custom client over the API, all at once. Conversations from every surface land in the same Chat History.
Why FastBots
POST /api/public/chat takes a message and returns a reply. Same JSON shape every request, no endpoint sprawl, no surprises when you wire it into your client.
Each chatbot has its own X-BOT-API-KEY generated from the bot's Settings page. Rotate them independently and scope different bots to different products without one leaked key affecting the rest.
Pass the chatId returned in the previous response back on the next request and the bot answers with full context. Leave it out and a new conversation starts cleanly.
Point the bot at your website, docs, PDFs, Google Sheets or YouTube videos in the dashboard. The API serves whatever the bot has been trained on, no retrieval pipeline to build or maintain on your side.
INVALID_API_KEY, BOT_DISABLED, CHAT_NOT_FOUND, BOT_UNAVAILABLE and the rest are documented up front. Your client knows which ones to retry with backoff and which mean the request or config needs fixing.
Use your own UI: a React panel inside your SaaS, a mobile screen, a CLI, a voice or IVR menu. The endpoint is headless, so the look, feel and channel are entirely yours.
Retrain on new content, edit the prompt, swap the AI model or add guardrails inside the dashboard. Your API client keeps working, no migration, no client release needed.
Connect Zapier MCP or Make to the same bot and it can act as an AI agent: booking, ticket creation, CRM updates and order lookups can fire from inside the conversation rather than after it ends.
Real Scenarios
Embed a chat panel inside your SaaS that's trained on your help centre, changelog and pricing. Users get answers without leaving the screen they're working on.
Wire the endpoint into your iOS or Android app so users can ask questions in-app instead of emailing support. Same trained content, native UI.
Drop a chat box into Retool, an admin panel or an internal Next.js app so staff can query your trained internal documentation without switching tabs.
When the standard widget doesn't fit your brand or layout, build your own chat UI and call the API behind it. Keep the bot logic in FastBots, keep the front-end yours.
Pipe transcribed speech into the endpoint and read the reply back through your voice or kiosk stack. The bot stays text-based; the surface around it can be anything.
Want the bot in Discord, Microsoft Teams or a niche messaging platform? Use the API to bridge inbound messages into FastBots and post the reply back.
Use a trained bot as a private, content-grounded LLM inside scripts and scheduled jobs, summarising tickets, drafting replies or answering questions from a queue.
Hit the endpoint with a set of canned questions to check the bot still answers correctly after you retrain it, swap a model or change the system prompt.
Try It Free
Need A Chatbot Inside Your Own Product?
Free plan, no credit card. Build your first chatbot in minutes.
Setup
Crawl your website, upload PDFs, paste help docs or connect a Google Sheet. The bot learns from your real content rather than generic AI knowledge.
House voice
Answer like a developer-relations engineer. Keep replies concise, prefer short code examples over prose, and link to /docs when the question goes beyond a quick answer. If the user needs account-specific data or billing changes, tell them to open a ticket.
Write a system prompt, choose answer sources, pick the AI model and define what the bot should refuse or escalate. This is what your API client will return on every call.
Open the bot's Settings page, generate an X-BOT-API-KEY, then POST to /api/public/chat with a message. You're back a reply and a chatId in seconds.
Honest Comparison
Most teams building a chat experience inside their own product weigh up three options: call the FastBots API, build the whole pipeline from scratch on a raw LLM API, or stitch together a generic chatbot API with their own retrieval layer. Here's how they actually stack up.
FastBots From $39 / month | DIY On A Raw LLM API Dev time + token costs | Generic Chatbot API Varies | |
|---|---|---|---|
Trained On Your Own Content Website, files, sheets, YouTube | Yes, in the dashboard | You build the RAG pipeline | Depends on the platform |
One Documented Endpoint POST /api/public/chat | Yes | You design and version it | Usually yes |
Conversation Continuity Pass chatId back to keep context | Built in | You manage state and history | Varies |
Knowledge Management UI Update training without a redeploy | Included | Build your own admin | ~Sometimes |
Live Chat Handover Available Hand to a real person when needed | Business plan and above | Build it yourself | Often a separate add-on |
Same Bot On Other Channels Widget, WhatsApp, Instagram, Slack | Included | Build per channel | Varies |
Comparison reflects typical alternatives. Other tools may differ, so always check current pricing and features before deciding.
Try It Free
Spin Up A Bot Free And Hit The Endpoint Today
Free plan, no credit card. Build your first chatbot in minutes.
Very easy-to-use service. It takes 10 minutes to set up a bot. Service also very good value for money.
Got Questions?
The ones we hear most before teams sign up. If yours isn’t here, our support team replies within hours during business days.
It's a single REST endpoint: POST https://app.fastbots.ai/api/public/chat. Send a JSON body with the user's message and an optional chatId, and the response gives you the bot's reply plus a chatId to pass back on the next request to maintain conversation context.
Each chatbot has its own API key, generated from the bot's Settings page. Pass it on every request in the X-BOT-API-KEY header. Keys are per-bot, not per-account, so you can scope different bots to different products and rotate them independently.
Yes. The first response includes a chatId. Pass that back as chatId on the next request and the bot answers with full conversation context. Leave chatId out and a new conversation starts. Past conversations are also visible inside the dashboard's Chat History.
Not currently. The endpoint returns a single reply per request rather than a token stream. If you want a typing effect in your UI, render the returned text character by character on the client side after the response arrives.
No. The Bot Chat API is chat-only: messages in, replies out. Training updates, prompts, model choice, appearance, integrations and live-chat handover are all configured inside the FastBots dashboard, not over the API.
Documented codes include INVALID_JSON, MESSAGE_REQUIRED, MISSING_API_KEY, INVALID_API_KEY, BOT_DISABLED, CHAT_NOT_FOUND, BOT_UNAVAILABLE and INTERNAL_ERROR. BOT_UNAVAILABLE and INTERNAL_ERROR are worth retrying with backoff; the rest mean the request or configuration needs fixing on your side.
You build and train your bot inside the FastBots dashboard, and there's a free plan to start on (one chatbot, 50 messages a month, no credit card). Paid plans start at $39/month and add more bots, messages, training capacity and channels. Check the current pricing page for where API access sits on each plan.
Yes. The same bot can be embedded as a website widget, shared via a direct link, connected to WhatsApp, Instagram, Messenger, Telegram or Slack, and called from your own client over the API, all at the same time. Each surface uses the same trained content.
Not through this endpoint. It accepts a text message and returns a text reply. File and image uploads inside conversations are supported on the chat widget on Essential plan and above, but the public chat API itself is text only today.
Train a chatbot on your content, generate a key and call the endpoint from your own code in minutes. No credit card needed for the free plan.
