Build a custom GPT-style agent in 10 minutes. Connect Gmail, WhatsApp, SMS, Telegram, and your website. One agent answers everywhere — in the channel's native tone.
Stop juggling five SaaS dashboards. Chat Cactus connects to Gmail, Twilio SMS, WhatsApp, Telegram, and the embedded chat widget — and the same agent answers them all.
Drag and drop PDFs, Word docs, CSVs, or plain text. We chunk, embed, and index automatically. The agent only answers from what you uploaded — and cites every source.
The same question gets answered three different ways. Long-form HTML email. Tight 160-char SMS. WhatsApp markdown with emoji. You don't configure it. The model knows.
Hi Alex,
Thanks for reaching out! Your Pro plan renews on March 15, 2026. The renewal charge will be $19, billed to the Visa ending in 4242.
Let me know if you'd like a copy of the invoice in advance.
Best regards,
Acme Support
New integrations ship every month. Custom webhooks for anything else.
Your agent answers refund, shipping, and "where's my order" questions instantly. Customers love the speed, your team loves the relief.
The pre-chat form captures contact details before the conversation. The agent qualifies (budget, timeline, fit), then routes hot leads to your team.
Train an agent on your company wiki, HR docs, IT runbooks. Drop it in Slack-via-webhook. Suddenly your senior people get their afternoons back.
How support and ops teams put Chat Cactus to work.
A single agent answers on the website, over email, and by SMS — with the full conversation context retained across channels, so customers never repeat themselves.
Upload your docs, write a one-paragraph persona, paste one line of HTML — the first agent is typically answering real questions in under 30 minutes, formatted natively for each channel.
Replies cite the exact documents they came from, so teams can review what the agent said and verify it pulled from approved sources — useful anywhere compliance matters.
What you get out of the box vs. what you build over six months.
Every account starts free for building and testing — pick a plan when you're ready to go live. Card only charged after your trial ends.
For getting your first agents in front of customers.
Start 14-day free trialFor teams running real customer conversations at volume.
Start 14-day free trialScale, compliance, and custom integrations — priced for your volume.
Talk to salesOpenAI (GPT-4o, GPT-4o-mini), Anthropic (Claude 3.5 Sonnet), or a fallback chain across both. The model router automatically downgrades when you approach your monthly cost cap, so you never get a surprise bill.
Under 10 minutes for the first agent. Upload your docs, write a 1-paragraph persona, pick a model. The widget snippet is one line of HTML. Channel integrations (Gmail OAuth, Twilio API key, Telegram bot token) take 2-5 minutes each.
Knowledge documents are embedded once on upload, then only the relevant chunks are sent to the LLM at query time. Integration credentials (OAuth tokens, API keys) are AES-256-GCM encrypted at rest. We never train models on your data.
Yes — human escalation is built in. When a customer asks for a person (or the AI can’t answer reliably), the conversation moves to your team’s Human Escalations queue: the AI stops replying, your team is notified, and a teammate answers on the customer’s own channel straight from the dashboard — then resolves it or hands it back to the AI.
Yes. Every action in the dashboard has a REST endpoint. The embedded widget itself runs on the same public API. Full docs live at /docs after sign-in.
Yes. The codebase runs as two processes (Next.js + FastAPI) against MySQL — no managed services required. Enterprise plan includes the self-host license + setup help.
10 minutes from sign-up to first answered customer message. Free forever plan.