AI survey & feedback agent
A survey that feels like a chat, not a form: it asks a couple of questions, follows up on the interesting answers, and hands you structured data.
The scenario
Traditional surveys have a response-rate problem: a link to a 20-question form gets ignored, and the few who start it drop off halfway. You end up with thin, skewed data — and no idea why the unhappy customers were unhappy, because a rating scale cannot ask “what went wrong?”.
How SMS-24 handles it
SMS-24 runs the survey as a natural conversation on WhatsApp and Telegram, only where the recipient agreed to be contacted. It asks a couple of questions, follows up on a low score or an interesting comment to get the real reason, and writes structured results (score plus the free-text why) back to your systems over MCP. Response rates climb because it feels human; it answers in the respondent’s language and discloses that it is an AI.
What it can (and can’t) do
An honest capability list — no autonomy theatre:
- Auto-reply on Telegram and WhatsApp in your persona’s tone of voice, around the clock.
- Real-time AI voice calls — Telegram voice today, PSTN through your own Twilio number.
- Answers grounded in your knowledge base and uploaded documents (PDF, DOCX, TXT, MD).
- Calls your own tools mid-conversation — CRM, calendar, order lookup — over MCP and function calling.
- Discloses that it is an AI and hands off to a human on request, in line with the EU AI Act.
- Keeps chat and voice-note history per conversation, encrypted at rest; you bring your own LLM key.
CRM & tools (MCP / function calling)
Connect your CRM, calendar or order system once and the agent uses it live. SMS-24 speaks the Model Context Protocol (MCP), so any tool you already expose as an MCP server becomes a function the agent can call mid-conversation — look up an order, check availability, create a ticket — and read the result back to the customer in their language. No bespoke glue code per integration.
Pricing
Plans start at €29/month. You bring your own LLM key (OpenAI, Anthropic or Gemini), so you pay your provider directly with no markup from us. Voice minutes and messenger accounts scale with the tier.
Start free See the Telegram AI agent
FAQ
Why does conversational beat a form?
People answer a short chat they would ignore as a link, and the agent can ask a follow-up (“what went wrong?”) that a static rating scale never can — so you get the reason, not just the number.
Where do the results go?
Structured — score plus the free-text reason — written to your CRM, analytics or spreadsheet as an MCP tool, ready to analyse.
Is it consent-first?
Yes. It only surveys contacts who opted in, discloses it is an AI, and stops immediately on request.