Botpress
Enterprise-grade AI agent platform for customer support. Its agents issue refunds, update accounts and run multi-step workflows instead of only answering, billed per conversation rather than per seat, on top of an existing helpdesk or standalone.
What is Botpress?
Botpress is an enterprise-grade platform for building and running AI agents, aimed squarely at customer support. It comes in two halves that can be used together or separately. Botpress Studio is the build environment: a drag-and-drop canvas for conversation flows and branching logic, a code editor alongside it for custom actions and bespoke integrations, and a conversation emulator for testing before anything goes live. Botpress Desk is the helpdesk side, handling L1, L2 and L3 tickets with teams and capacity-based routing, business hours, priorities, custom statuses, tags, custom fields, merging, snoozing and threaded replies.
The reasoning layer is the Autonomous Engine, which blends LLM reasoning with structured logic to decide what to do next, which tools to call and how to handle inputs it has never seen, within constraints and fallback rules the builder defines. Agents draw on Knowledge Bases built from documents, websites, spreadsheets and CSV files, and on Tables, which hold structured data that persists across sessions and scales to millions of rows. A multi-LLM strategy lets each part of a workflow run on a different model, traded off for cost, speed or capability.
The commercial argument is that the agents act rather than only answer: refunds, account updates, order corrections and multi-step workflows across a customer's stack, escalating to a human with full context when they should. Botpress connects to an existing Zendesk or Intercom helpdesk over OAuth without a migration, or replaces it. Channels include webchat, email, WhatsApp, Slack and Viber, with voice reserved for the Enterprise tier.
Botpress claims ten years of AI infrastructure work, 750,000 agents built, more than a million end users across 180 to 190 countries, and use by over 35% of the Fortune 500. Security rests on SOC II certification, GDPR compliance, penetration testing by KPMG and AWS hosting. Enterprise adds a private LLM gateway that strips personal data before it reaches a model provider, prompt sandboxing, audit and conversation logs, and an MCP server exposing helpdesk operations.
What it does
- Resolve a support ticket end to end, without a human ever touching it
- Issue a refund under the same conditions a human agent would apply
- Update a customer account: change a plan, correct billing details, fix an account error
- Run a multi-step workflow that coordinates several systems in one motion
- Hand off to a human with the full conversation history, account data and troubleshooting steps already attempted
- Reach out proactively before a limit, a renewal or a change affects the customer
- Publish and maintain a self-service help centre that the AI also draws on
When to use Botpress / When not to
A quick filter to help you decide if Botpress is the right fit.
When to use Botpress
- Support teams that have already deployed an AI tool and hit a ceiling on the tickets that require an actual action rather than an answer
- Organisations that want to keep their existing Zendesk or Intercom helpdesk and connect it over OAuth instead of migrating away from it
- Large support operations where per-seat licensing has become the dominant cost, since Botpress charges by conversation volume and not by headcount
- Regulated sectors such as healthcare, fintech, insurance and banking that need audit logging, escalation rules and a HIPAA BAA
- High-volume e-commerce and higher-education teams facing seasonal surges: order tracking, returns and refunds, or enrolment and financial-aid enquiries
When not to use Botpress
- Individuals: the terms of service require the customer to be operating a business and exclude personal or household use
- Anyone under 16, since the service does not knowingly collect personal data from children below that age
- Small teams looking for an inexpensive FAQ chatbot, as the first paid tier starts at USD 150 per month on annual billing
- Buyers who need a voice channel, custom data residency, custom retention or an uptime SLA without moving to the undisclosed Enterprise tier
- Teams that want a native mobile app or a browser extension, neither of which Botpress publishes
How to use Botpress
A typical end-to-end flow, from setup to results.
- Create a free account at sso.botpress.cloud, with no credit card required
- Connect your existing helpdesk over OAuth if you already run Zendesk or Intercom, or start from Botpress Desk as a standalone helpdesk
- Build the agent in natural language or on the Studio's drag-and-drop canvas, without needing developer resources
- Feed it knowledge: upload documents, ingest websites, connect spreadsheets and CSV files into the Knowledge Bases
- Add Tables where the agent needs to remember or update structured records mid-conversation
- Set the constraints, escalation rules and fallback logic that decide when the AI acts and when it hands over
- Write custom code actions for the internal APIs that no prebuilt integration covers
- Test everything in the built-in conversation emulator before deployment
- Deploy to the channels you need from the Hub: webchat, WhatsApp, Slack, email, or voice on Enterprise
- Monitor the KPI dashboard and CSAT analytics, and roll back through version history when a change misfires
Pros & Cons
Pros
- No per-seat cost on any paid plan, so the bill does not grow with the size of the support team
- AI spend is included: Botpress covers USD 0.10 of LLM usage per conversation, with USD 25 included on Plus and USD 150 on Team
- Sits on top of an existing helpdesk over OAuth, so there is no migration and no rip-and-replace
- Agents take action — refunds, account updates, order corrections — rather than only answering, and hand off with full context
- Every human correction feeds back into the agent, so resolution rates are meant to compound instead of plateauing
- Serious compliance posture: SOC II, GDPR, KPMG penetration testing, a published DPA, an Article 27 representative and a HIPAA BAA on Enterprise
- Permanent free plan with 100 conversations, three seats and three agents, usable without a credit card
Cons
- The first paid tier is USD 150 per month on annual billing, a high entry point for a small team
- Month-to-month prices are rendered in JavaScript and are not published in the static page, so only the annual commitment figures are verifiable
- Auto-recharge on conversation packs cannot be switched off, and the FAQ answers that question with a plain no
- Unused conversations in a pack expire at the end of the month and do not roll over, while a conversation spanning two months is counted twice
- Data residency, custom retention, security review, uptime SLA and the voice channel are all Enterprise-only, and Enterprise pricing is not public
- No support email address is published: technical support runs through live chat from the Plus plan, Discord and a contact form
- No native mobile app, no browser extension, and no published list of product interface languages
Pricing & Plans
A permanent free plan is available at no cost, covering 100 conversations, three seats and three AI agents, and requiring no credit card. The lowest paid entry point is the Plus plan at USD 150 per month, billed annually, which includes 250 conversations per month and USD 25 of AI usage. Month-to-month pricing is advertised as 20% higher than the annual commitment but is not published in the page source.
- Free — USD 0 — 100 conversations
- 3 seats
- 3 AI agents
- Botpress Studio and Desk
- webchat and WhatsApp
- Help Center
- 1
- 000 table rows
- Plus — USD 150 per month billed annually — 250 conversations per month
- USD 25 of AI usage included
- unlimited AI agents
- 3 seats
- white-label webchat
- Help Center with a custom domain
- live-chat technical support
- 100
- Team — USD 750 per month billed annually — 1
- 500 conversations per month
- USD 150 of AI usage included
- unlimited seats
- teams
- routing and RBAC
- team analytics
- real-time collaboration
- Enterprise — custom pricing — custom conversation volume
- dedicated support
- uptime SLA
- security review
- custom storage
- voice channel
- custom data retention and residency policies
- Managed plan (Built by Botpress) — custom pricing — agents designed
- built and maintained by the Botpress team
- Botpress for Good — reduced pricing for nonprofits
- educational institutions and open-source projects
- granted on application
Data, GDPR & hosting
A consolidated view of how Botpress handles your data.
GDPR overview
GDPR compliance is stated explicitly and backed by concrete arrangements. The privacy statement, last updated on 17 August 2026, says Botpress adheres to the GDPR and takes part in an EU Data Protection Representative Program. A Data Protection Officer is designated and reachable at legal@botpress.com, and VeraSafe Netherlands BV in Amsterdam is named as the Article 27 representative in the European Union. Botpress is self-certified under the EU-U.S. Data Privacy Framework, its UK extension and the Swiss-U.S. Framework, and is subject to FTC enforcement; other transfers rely on standard contractual clauses. A data processing agreement is published and can be signed online, using SCC modules 2 and 3 with ten days' notice on sub-processor changes. Complaints are answered within 45 days, with VeraSafe dispute resolution and cooperation with EU authorities, the ICO and the FDPIC.
Who owns the data?
Under the terms of service the customer keeps all worldwide right, title and interest, including intellectual property rights, in its Customer Data, and Botpress acquires no ownership beyond the licence needed to run the service. The customer also owns the intellectual property in the Customer Bots it builds; where such rights would otherwise vest in Botpress or its staff, Botpress undertakes to assign them. Botpress retains ownership of the platform itself, the software, its interface, the documentation and any improvements, as well as of the Analytics Data it derives, which it is not obliged to share. Under the data processing agreement the customer is the controller and Botpress the processor.
Reuse rights
Customers may reuse their own Customer Data freely: they own it outright and Botpress holds only the licence required to deliver the service, so no permission is needed to export, reuse or exploit it. The same applies to the Customer Bots they build, whose intellectual property is theirs. What they cannot reuse is the platform side of the equation: the software, its interface and the documentation remain Botpress property, deliverables produced under Professional Services belong to Botpress unless they are Customer Bots, and Analytics Data derived from usage stays with Botpress, which has no obligation to hand it over. On its own side Botpress limits itself to a defined list of purposes: improving features, providing technical support, auditing security, and identifying trends without reference to a specific customer. Algorithm and model training is listed as a permitted use of Analytics Data, but Conversation Data and anything derived from it is expressly carved out and never used for that purpose.
Data retention & training
Hosting summary
Amazon Web Services hosts the Botpress cloud services and all Customer Data, and also processes usage analytics for those services. Customer Data sits in a shared environment but is logically segregated from other data on the same infrastructure. No AWS region or specific hosting country is named anywhere on the site. The privacy statement warns that personal data may be processed, transferred or disclosed in the United States and in other countries where affiliates and service providers operate or hold servers. Transfers out of the EEA, the United Kingdom and Switzerland rely on the relevant Data Privacy Framework certification, held by the US entity Botpress Inc. in Newark, Delaware, or on back-to-back standard contractual clauses. Google Analytics data is stored on servers controlled by Google. Custom data residency policies are available, but only on the Enterprise plan. Buyers with a regional hosting requirement will therefore need to raise it with the sales team, since nothing on the public pages settles the question.
Things to keep in mind
Risks and trade-offs to weigh before adopting Botpress.
- Conversation packs recharge automatically and the option cannot be disabled, so a busy month can push spending well past the plan you chose
- The counting rules work against you at the margins: unused pack conversations expire monthly, a conversation spanning two months is billed twice, and separate anonymous sessions each count
- Delegating refunds and account changes to an agent means delegating judgement; the escalation rules and fallback logic you write are the only thing standing between automation and a costly mistake
- Two contracting entities apply depending on where you are — Technologies Botpress Inc. in Canada, Botpress, Inc. in Delaware elsewhere — which changes whether Quebec or Delaware law governs your agreement
- OpenAI and Microsoft Azure appear among the sub-processors handling service data, which may conflict with internal policies in some organisations
- Backup is your responsibility: the terms state the service does not include backup of Customer Data, and Botpress may permanently delete it 30 days after termination
- Automating the routine tickets erodes the everyday practice through which support staff learn a product, so the human expertise you still need for the hard cases has fewer occasions to form
Setup & Integrations
Technical difficulty
Low to start, higher to master. No developer resources are needed for a first deployment: agents are built in natural language or on a drag-and-drop canvas, and an existing helpdesk connects over OAuth in minutes. Botpress advertises going live in days, and one customer reports being operational the same day. A built-in emulator lets you test before deploying. Complexity rises with multi-system workflows and custom code actions, where the embedded code editor and API come into play. Teams without support operations can have Botpress handle the full implementation, or hire a certified partner.
Deployment
Integrations
Behind Botpress
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Botpress.
Frequently asked questions
What counts as a conversation?
What happens when I reach my conversation limit?
Can I turn off auto-recharge?
Is there a free plan?
Do I have to migrate my existing helpdesk?
Is my data used to train AI models?
Is a data processing agreement available?
Where is the data hosted, and what security is certified?
Should you pick Botpress?
Botpress is a mature product rather than a recent arrival: ten years of infrastructure work, 750,000 agents built, and a USD 25 million Series B closed in June 2025. Its argument is narrow and clearly made. Most AI support tools deflect the easy tickets and hand the rest back; Botpress is built for the ones that require an action — a refund, an account change, a workflow spanning several systems — and escalates with the full context when a human is genuinely needed. The second argument is commercial: conversation-based pricing with no per-seat charge, which inverts the usual incentive since a better agent lowers the bill instead of raising it.
It suits support teams that have already hit a ceiling with a general-purpose AI tool, organisations wanting to keep their existing Zendesk or Intercom setup, and regulated sectors that need audit trails and a HIPAA BAA. The legal posture is unusually solid for the category: a published DPA, an Article 27 representative in the EU, Data Privacy Framework certification, SOC II, and a contractual commitment never to train on conversation data.
The reservations are mostly about money and tiering. The entry paid plan is USD 150 per month on annual billing, which puts it out of reach of small operations, and month-to-month prices are not visible in the page source. Auto-recharge on conversation packs cannot be disabled. Data residency, custom retention, uptime SLA and the voice channel all sit behind an Enterprise tier whose price is not published. Prospective buyers should establish the month-to-month rate, the Enterprise terms and the AWS hosting region before committing.
- Choosing a selection results in a full page refresh.
- Opens in a new window.