If you’re building an AI agent without writing code, two names come up again and again: Voiceflow and Botpress. Both let you drag boxes on a canvas, connect a knowledge base, and ship a working chatbot in an afternoon. But they pull in different directions, and picking the wrong one means either a beautiful agent your developers can’t extend, or a powerful agent your non-technical teammates are afraid to touch.
We build agents on both platforms for clients every week. This is the comparison we wish existed before we learned the hard way: what each tool is genuinely good at, where it falls down, what it actually costs in 2026, and how to choose in under five minutes.
The one-sentence version
Choose Voiceflow if a designer, marketer, or CX person owns the agent and you care most about clean conversation design across chat and voice. Choose Botpress if a developer owns the agent and you need deep logic, custom integrations, and the freedom to drop into code when the canvas runs out of room.
Neither is “better.” They’re built for different people. Read on for why that matters in practice.
How each one feels to build in
Voiceflow: the designer’s canvas
Voiceflow’s editor is the cleanest in the category. Conversation steps read left-to-right like a storyboard, so a stakeholder who has never seen the tool can follow how a chat moves from greeting to intent to handoff. That clarity is the whole point: it’s built for teams where the person designing the conversation isn’t the person who can write a Python function.
In practice, a typical Voiceflow build looks like this:
- Drop a knowledge base (upload PDFs, paste URLs, or sync a help center) so the agent can answer from your real content.
- Lay out the happy path with Talk and Listen blocks, then branch with intents and conditions.
- Use an AI step (the “Response AI” / generative block) to let the LLM answer freely where you don’t want a scripted reply.
- Test in the built-in runner, then publish to a web widget, WhatsApp, or a voice channel.
Where Voiceflow shines is anything voice-shaped or design-led: IVR-style phone agents, voice assistants, and chat flows that multiple stakeholders need to review and sign off on. Its prototyping and collaboration features are a real advantage when the agent is a team deliverable, not a solo project.
Botpress: the developer’s workbench
Botpress also has a visual flow editor, but its center of gravity is logic and extensibility. You get an Autonomous Node (an LLM-driven step that can decide which tools to call), first-class custom code actions in JavaScript, and a clean path to wire the agent into your own APIs and databases. It is fully LLM-agnostic, so you can route to OpenAI, Anthropic, Mistral, or a model of your choice without fighting the platform.
A typical Botpress build leans on:
- Knowledge Bases for retrieval, same idea as Voiceflow.
- The Autonomous Node plus Tools, so the agent can take actions (look up an order, create a ticket) instead of only talking.
- Code execution cards when you need real logic: transform an API response, validate input, branch on a calculation.
- Deployment to web, WhatsApp, Telegram, Slack, or your own app via the API.
Botpress is the right call when the agent has to do things in other systems, when business logic gets genuinely complicated, or when a developer wants an escape hatch into code without rebuilding the whole thing from scratch.
Pricing in 2026 (read this carefully)
This is where the two diverge most, and where people get surprised by their bill. The headline subscription is only half the story for both tools, because the underlying LLM usage costs money on top.
Voiceflow uses a credit model. Roughly: every user message to your agent consumes 1 credit, and phone/voice consumes about 10 credits per minute. Plans bundle a credit allowance; LLM token usage is layered on top depending on the model you pick. The Pro plan starts around $60/month, Business sits higher (around $150–$250/month depending on volume), and additional builder/editor seats are about $50/month each. The mental model: your bill is predictable right up to the credit cap, where the agent stops until you top up or upgrade.
Botpress changed its pricing in May 2026 and now bills per conversation (a conversation being an exchange with at least two end-user messages), with AI Spend bundled in for newer workspaces instead of billed as a separate floating line. There’s a pay-as-you-go tier (around 500 incoming messages/month, one seat, a small ~$5 AI credit), then Plus at ~$89/month, Team at ~$495/month, and a Managed tier around $1,495/month. The mental model: your bill drifts up with traffic rather than hard-stopping.
| Voiceflow | Botpress | |
|---|---|---|
| Best for | Designers, CX, marketers | Developers, technical teams |
| Superpower | Clean canvas; voice + chat design | Custom logic, code, integrations |
| Billing model | Seats + credits (predictable, caps out) | Per conversation, AI bundled (drifts with traffic) |
| Entry paid plan | ~$60/mo (Pro) | ~$89/mo (Plus) + pay-as-you-go option |
| Extra builder seat | ~$50/mo | Included differently per plan |
| Code escape hatch | Limited | Strong (JS actions) |
| Native voice/phone | Strong | Available, less of a focus |
Prices and limits shift; always confirm on each vendor’s live pricing page before committing. But the shape of the two models is stable and is the thing to plan around: Voiceflow = caps, Botpress = creep.
When each tool is the WRONG choice
Honesty matters more than balance here.
- Skip Voiceflow if your agent’s main job is taking actions in other systems with messy conditional logic. You’ll hit the ceiling of the canvas and end up bolting on external services to do what Botpress does natively.
- Skip Botpress if no one on the team can or wants to think like a developer. The power is real, but a marketer handed a Botpress project often stalls. The clean review-and-collaborate workflow Voiceflow gives you isn’t the priority here.
- Skip both if you only need a simple FAQ bot on one website. A lighter, cheaper help-desk widget will get you there faster. And if your real goal is a deeply custom product agent with full control over every response, a code framework (or an agent SDK) may serve you better than either no-code tool.
- Watch the traffic math. A viral spike on Botpress means a bigger invoice; the same spike on Voiceflow means the agent quietly stops at the credit cap. Match the failure mode you can live with.
A practical way to decide in 5 minutes
Answer these three questions:
- Who maintains this agent six months from now? A designer/CX person → Voiceflow. A developer → Botpress.
- Does the agent mostly talk, or mostly do? Mostly talk (answer questions, qualify, route) → either works, lean Voiceflow for voice. Mostly do (hit APIs, run logic, take actions) → Botpress.
- Which bill surprise can you tolerate? “It stopped at the cap” → Voiceflow. “It cost more this month” → Botpress.
If two of three point the same way, you have your answer. The fastest validation is to build the same small slice in both: a greeting, one knowledge-base lookup, and one branch. An hour in each free tier tells you more than any review, including this one.
FAQ
Can I build a voice agent (phone calls) on both?
Yes, but Voiceflow treats voice as a first-class design surface, which makes IVR-style and voice-assistant flows noticeably smoother to build and review. Botpress can do voice via integrations, but if voice is the core of your project, Voiceflow is the more natural fit. Remember Voiceflow charges voice at roughly 10 credits per minute, so model your minutes before launch.
Do I still pay for OpenAI or Anthropic separately?
It depends on the platform and plan. Voiceflow layers LLM token usage on top of your credit plan, so heavy generative use raises costs. Botpress, since its May 2026 change, bundles AI Spend into newer workspaces’ per-conversation pricing, so for many teams it’s now one line rather than two. Always check whether model usage is included or billed on top for the specific plan you choose.
Which is easier for a complete beginner?
Voiceflow, in most cases. The canvas is more intuitive and the learning curve is gentler for someone who has never built a bot. Botpress rewards a bit more technical comfort, and pays that back with far more power once you’re past the basics.
Your next step
Don’t agonize over the choice on paper. Open both free tiers today and rebuild one real conversation from your business: the question your customers ask most. Time how long each takes and notice which editor you actually enjoy. The tool you’ll still be maintaining happily in six months is the right one, and an hour of hands-on building will tell you which that is far better than any spec sheet.