Botpress vs Voiceflow: Voice, Pricing and Handoff Compared

Botpress and Voiceflow both build AI agents for customer conversations, but they part ways on three things a buyer feels early: voice, price and release control. Botpress prints its price list, from a $0 Free plan with 100 conversations up to Team at $750 a month billed annually, and bills per conversation ^[1]. Voiceflow's pricing page prints no prices; it bills usage in credits and quotes business plans on request ^[2]. Pick Botpress for a published price, a built-in helpdesk and WhatsApp on every plan. Pick Voiceflow when phone agents and staged releases sit at the center of the project. Neither wins outright.

Three Botpress builders against Voiceflow's playbooks and workflows

Botpress puts three builders on one plan and one price list. Botpress Studio is a browser-based visual builder whose Autonomous Node runs on LLMz, Botpress's own engine, while workflows bypass the model wherever you need strict control, such as payments or escalation. Botpress Viber skips nodes entirely: you describe the agent in plain language and shape it through Instructions, Playbooks and Escalation. The ADK is the code route, a TypeScript CLI driven by adk init, adk dev and adk deploy, with evals and traces built in.

Voiceflow builds every agent from instructions, playbooks, workflows, knowledge and tools. Playbooks are goal-based and let the model reason freely, while workflows follow a defined step-by-step path and can embed playbooks mid-flow. Projects run on the Agentic framework, which Voiceflow recommends, or the Conversation flow framework, and the switch runs one way only. Same word, different weight: Botpress Viber uses playbooks for reusable procedures inside one agent, while in Voiceflow they are one of two core building blocks, next to workflows.

On models, Voiceflow sets a default LLM per project, lets each playbook override it, and, on Business, can switch to a fallback LLM provider during an outage; its pricing page for businesses also says you can bring your own model. Botpress Studio supports models from OpenAI, Anthropic, Google, xAI and Mistral plus open models through Fireworks AI, Cerebras and Groq, but your own LLM through the SDK is Enterprise only. Voiceflow lists nine prebuilt integration tools plus API, hosted MCP and JavaScript function tools; Botpress counts 190+ integrations. Our Botpress AI agents guide covers the three builders in depth.

Voice decides the shortlist for many teams

This axis is lopsided. Botpress's pricing page lists the voice channel only on Enterprise, which carries custom pricing. On Enterprise, the only plan with the voice channel, 3 minutes of a call count as one conversation. Below Enterprise, a Botpress agent works in text: webchat and WhatsApp on every plan, Free included, plus Telegram, Messenger, Instagram, Slack, Teams and other channels through Hub integrations. So if phone calls are the core job, Botpress starts with a sales conversation.

Voice is core at Voiceflow. Its agents take speech-to-text from providers such as Deepgram, AssemblyAI or Google and voices from ElevenLabs, Cartesia, Amazon, Microsoft and others. Voiceflow can provision a US or Canadian number directly, includes one phone number per workspace, and connects Twilio, Vonage, Telnyx or your own phone system over SIP. Outbound calling needs one of those providers, because Voiceflow's own numbers do not dial out. Call recording covers phone calls only. Concurrent call limits depend on the plan, and inbound SMS is still labeled beta.

One vendor prints a price, the other sends a quote

Botpress bills per conversation: any exchange with at least two end-user messages in the billing month, counted the same whether AI-only, human-assisted or run in the emulator. Plus costs $150 a month billed annually or $189 a month billed monthly for 250 conversations a month. Team costs $750 a month billed annually or $939 a month billed monthly for 1,500 conversations a month. At 95% of the quota Botpress adds a pack of 100 extra conversations automatically, at $65 per pack on Plus and $50 per pack on Team, charged each month the pack stays on. This auto-recharge cannot be turned off.

By our arithmetic, a 1,300-conversation month on Plus billed monthly comes to $189 plus 11 packs at $65 per pack, or $904 for the month, and above 1,350 conversations a month billed monthly the Team base price is the cheaper route. At annual rates, that line sits above 1,150 conversations a month. By the FAQ's 95% rule, the real bill can be one pack higher, so near that line Team can already be cheaper.

AI usage is bundled into each paid plan; past 95% of that limit Botpress buys a $10 credit grant at cost each time the limit is reached, kept up to 12 months and not refundable. The Free plan stops at its 100 conversations, with no packs, top-ups or overages; the widget then shows Quota exceeded. One catch matters. Packs keep billing each month until you remove them in billing settings, and unused pack conversations expire at month end.

Voiceflow's docs name four plans, Free, Pro, Business and Enterprise, billed at the organization level. Paid plans carry a monthly credit allotment that resets each cycle and is shared across workspaces, while the Free plan gets a one-time grant. Auto top-ups are optional: when the balance hits zero they buy credits for an amount you set. Credits also drain outside the conversation itself, since LLM chunking on knowledge syncs, transcript evaluations, voice features and file processing all consume them. So Botpress is easy to price from the published list but hard to cap, because packs and AI grants recharge automatically, while Voiceflow's unit price stays unknown until sales quotes it.

Handing a conversation to a human

New Botpress workspaces hand off to Botpress Desk, the AI-native helpdesk launched in June 2026 and included on all four plans. In Desk the AI agent stays in the conversation, and the human sees the history and an AI-generated summary of what was already tried. Routing and assignment need Team or higher. Older workspaces can keep the legacy HITL integration on Plus or higher, and teams in another helpdesk can hand off through Hub integrations for Zendesk, Intercom, HubSpot, Freshchat and others. Our human handoff guide walks through each route.

Voiceflow offers native handoff connections for Genesys, UJET, Kustomer, Dixa and Zendesk . Anything else goes through Custom Handoff, which needs a developer and an HTTPS service implementing Voiceflow's connection API. You choose what context travels: the full transcript, a custom variable or nothing. The support platform owns agent availability and assignment, and handoff also works on SMS. The difference is structural. Botpress ships a helpdesk, while Voiceflow connects to yours.

Releases, testing and who can touch what

Voiceflow's deepest tooling sits here. Each project supports up to 10 environments, Main included, each with its own draft, live version and history . Protection can lock an environment so only admins and owners publish to it, and traffic splitting sends a share of users to different environments for A/B tests. Tests start manually and run in parallel. Evaluations score transcripts automatically once enabled, but they consume credits.

Botpress tests differently. In Botpress Viber you write question tests and multi-turn flow tests with mockable tools, and the ADK adds evals with LLM-judge assertions. Every emulator conversation counts toward the quota and draws on AI usage, so heavy testing has a cost. Workspace roles, per-agent access control and Allowed Origins need Team or higher, and Botpress includes 3 seats on Free and Plus. White-label webchat starts on Plus, whereas Voiceflow removes widget branding on Business and gives free, unlimited viewer seats on every plan.

Which team each one fits

Botpress fits teams that want to read the price before talking to anyone, keep a helpdesk inside the same plan, and run WhatsApp from the Free plan up. It also suits developers, since the ADK gives a code path next to the visual and plain-language builders. Voiceflow fits teams whose agent answers the phone, or who need staged releases with protected environments and A/B splits, and who are comfortable requesting a quote. If neither shape matches, our Botpress alternatives roundup covers other options.

On G2, reviewers mostly praise how easy Botpress is to use, its clean interface, its wide choice of integrations and its developer-friendly design; the recurring complaints are connecting workflows to each other, the quality of AI voices, testing an agent with specific users, and an interface and documentation that beginners find hard. Read that list against your own project. A voice-heavy build meets the voice point first.

Price list or phone line: settling the choice

Start from the channel. If the agent must talk on the phone and you are not buying an enterprise contract, Botpress drops out and Voiceflow's telephony options carry the decision. If the agent lives in webchat and WhatsApp and a human team answers what it cannot, Botpress puts the builder, the helpdesk and a printed price under one plan. Ask Voiceflow for a quote anyway, then set it against the Botpress conversation math for your real monthly volume, including the packs added automatically at 95% of the quota.