· Feedbot team
Feedbot v2: One AI Chatbot for Support, Sales and Feedback
Why we rebuilt Feedbot around three roles in one bot, flat pricing, and a feedback loop into your coding agent.
Most AI chatbots for websites do one job: they answer questions. That’s useful, but for a small SaaS the chat window sees much more than questions. People ask about pricing. They report bugs. They tell you what’s missing.
Feedbot v2 treats all of that as one conversation handled by one agent with three roles.
Supports
The bot answers from your knowledge base: your site, docs and files. When a question depends on the user’s account, it can call your API through Actions, for example to check an order or a subscription. When it can’t help, it flags the conversation and pings you in Telegram.
Sells
When a visitor shows buying intent, the bot asks the qualifying questions you define, recommends a plan and shows buttons to start a trial, buy or book a demo. Leads get a score and a stage, and hot ones reach you right away.
Listens
This is the part we care about most. Complaints, bugs and ideas are grouped into issues with user quotes, frequency and the pages where they happened. Your coding agent can pick them up:
- through MCP in any coding agent that supports MCP;
- with
npx feedbot pull, which writes open issues into your repo as Markdown files; - or as an agent-ready report you copy from the dashboard.
When an issue is marked fixed, the bot can tell the users who reported it. That message does more for retention than most emails we’ve ever sent.
Flat pricing
Models got cheap, so we don’t charge per resolution or per seat. There’s a free plan with 50 conversations a month, and paid plans cost $5, $15 and $39. Yearly billing gives you two months free.
Works without a website
Every bot has a hosted chat link. Put it in a mobile app, an onboarding email, a QR code or a Telegram channel.
Existing Feedbot widgets keep working as before. If you’re new, start free and paste one script tag.