Overview
Botpress is a platform for building, deploying, and managing AI agents and chatbots for customer support and automation. It combines generative AI with structured logic through its visual Agent Studio and autonomous engine.
Our take
Botpress is a developer-oriented platform for building conversational agents with real control over flow, tools and deployment, priced with a base subscription plus AI usage on top. That two-part pricing is the thing to model carefully, because usage is the variable that grows with success rather than with your plan.
Worth it if
Technical teams building custom conversational agents who need control over logic and deployment.
Look elsewhere if
Non-technical teams wanting a bot configured in an afternoon — the flexibility costs setup effort.
✨ Features
Agent Studio
Visually design conversation flows and agent logic with drag-and-drop.
Knowledge Bases
Ingest and index content from websites and documents for accurate answers.
Autonomous Engine
Blends generative AI with structured logic for reliable behavior.
Human Handoff
Escalate complex conversations from the agent to a live human.
⚖️ Pros & Cons
Pros
Cons
💰 Pricing
Pay-as-you-go
$0
- Start free
- 500 incoming messages/month
- 1 collaborator seat
- $5 monthly AI credit
- Unlimited bots
Plus
$89 /mo
- $89/mo plus usage
- Live agent handoff
- Additional collaborator seats
- Higher message limits
- Analytics
Team
$495 /mo
- $495/mo plus LLM & overage usage
- Up to 50,000 incoming messages/month
- Team permissions & collaboration
- Advanced analytics
- Priority support
Enterprise
Custom
- Contact sales for a quote
- Custom message volume
- SSO & advanced security
- Dedicated support & SLA
- Custom integrations
Categories
❓ Frequently Asked Questions
How should I budget when pricing is subscription plus usage?
Estimate conversations per month times messages per conversation times model cost, then double it for the first months while you tune. Usage scales with adoption, which means a successful launch raises the bill. Set spend alerts before going live rather than discovering the shape of the curve from an invoice.
How do I stop an agent answering things it shouldn't?
Constrain it structurally rather than by instruction. Ground answers in your own documentation, define explicit refusal-and-escalate paths for billing, refunds, legal and account security, and audit real transcripts weekly. Prompt instructions alone are not a boundary — they are a preference the model can drift away from.