We believe that the future of customer experience is agentic — that engagements with customers will be fluid, proactive, intelligent, and capable. And who better to build the solutions for their customers than the businesses themselves. Our philosophy is that businesses need to own their AI solutions to achieve the best results, building internal skills that allow them to build, manage, and iterate upon their AI solutions on their own.

V4 looks to the future. A future where agents can execute tasks, react to complex conversations, and make the right choices for businesses and their customers. V4 is our biggest product launch for enterprise yet. It’s the complete framework for enterprise AI customer experience from development to deployment; from measurement to iteration.

Build agents with Voiceflow’s enterprise agent framework

Agent: The Agent screen is the hub for everything your agent can do. In V4, you build a single agent and give it capabilities — rather than building dozens of individual agents and wiring them together. It starts with a global prompt that defines your agent's identity: its persona, response guidelines, and guardrails. From there, you layer on system tools and skills. Between identity and capabilities, you can have an agent up and handling real customer questions in minutes.

Skills: Skills are how your agent completes complex tasks. There are two types: Playbooks and Workflows. Playbooks are natural language instructions that give the agent a broad framework for completing open-ended work, like troubleshooting or product recommendations. Workflows are step-by-step sequences that execute exactly as you've designed them, guaranteeing order for tasks that need it. The two work together: a Workflow can contain a Playbook, giving you AI reasoning at any step without sacrificing the guarantee that every step executes in sequence. The agent decides which skills to use and when — you define the what, it handles the how.

Knowledge: The knowledge base is your agent's information management system — trained on your business, your content, and your sources. Connect Zendesk articles, internal docs, web pages, Shopify data, and more. When a new source is added or updated, the agent can use it immediately. The agent pulls from the Knowledge Base in real time, deciding what's relevant, and surfacing the right answer.

Context Engine: The Context Engine is the runtime powering every agent conversation. It synthesizes everything the agent needs to respond — business data and tools, customer context, conversation history, and short and long-term memory — all in real time, while streaming responses token by token simultaneously. The agent doesn't stop to think. It reasons, retrieves, and responds all at once. When an agent moves from one task to the next, the Context Engine hot-swaps instructions, loading only what's needed and clearing what isn't — keeping the context window lean and the experience fast.

V4 under the hood

Asynchronous tool calls: In enterprise environments, your agent has to work across complex systems — and not every API responds in milliseconds. In V4, tool calls run asynchronously — the agent keeps the conversation going while work happens in the background. No dead air, no awkward pauses. And when a tool call is unpredictable in length, you can set custom messages that fire on start, on completion, or on a delay, so the customer always knows what's happening.

Model selection: V4 lets you mix and match models across your agent — using a fast, lightweight model for greetings and routing, and bringing in a more powerful reasoning model only where the conversation demands it. The right model for the right job, at every step. At scale, that optimization compounds. Every new project also comes with Voiceflow-recommended model stacks — curated for your use case and balanced for cost and performance — and in-app model evaluations let you test and compare models against your own data before you ship.

Memory: V4's Context Engine includes full memory management, designed to keep costs down and performance high as conversations grow. Short-term memory is configurable — dial it up for complex multi-step conversations, dial it back for high-volume, fast-resolution use cases. Long-term memory extracts and stores key details separately, so your agent stays intelligent even after conversation history fades. 

Deploy customer experiences to web, phone, and mobile

Multimodal web widget: Embed a voice or chat agent directly on your website or mobile app — no custom engineering required. V4's Context Engine is natively multimodal, powering both voice and chat from a single API, so building for either channel — or both — is seamless out of the box.

Native number provisioning: Provision a phone number directly inside Voiceflow. No third-party telephony setup, no additional configuration — go from building your voice agent to deploying it live in minutes.

Telephony: V4 includes native integrations for Twilio, Vonage, and Telnyx for enterprise voice deployments. Whether you're running high-volume inbound support or proactive outbound experiences, the infrastructure is built in and ready to scale.

Custom SIP trunking: For teams with existing telephony infrastructure, V4 supports custom SIP trunking — so you can bring your own technology and connect your voice agent to the systems you already have in place.

Measure ROI with the new observability suite

Analytics: The Analytics dashboard gives you a high-level view of how your agent is performing across conversations. Monitor evaluation results, usage patterns, costs, and operational metrics all in one place — surfacing the signals that matter most to your business.

Transcripts: Transcripts let you monitor conversations in real time or look back at every past conversation in granular detail. Custom properties make it easy to tag and filter at scale, and in-depth action logs are tied to every transcript — so debugging is fast and finding gaps is even faster.

Evaluations: Evaluations use AI to analyze transcripts and score them against criteria you define — whether that's resolution rate, customer satisfaction, or any custom metric your business cares about. Evaluations run automatically on every new transcript, and can be applied retroactively to historical conversations so you can track trends over time. Instead of manually reviewing conversations, you get insights at scale, automatically.

Collaborate with the entire team on the next iteration

Environments V4 supports multiple deployment environments, so teams can build, test, and iterate with confidence — without the risk of changes affecting live customer conversations. Move from development to staging to production on your own terms.

Real-time collaboration Just as Voiceflow brought real-time collaboration to conversation design in V2, V4 brings it to the full agent-building workflow. Teams can build and iterate together across agent frameworks, Skills, and knowledge — with the visibility and version control that enterprise teams need to move fast without breaking things.

Get early access to V4

There’s so much to discover with Voiceflow V4. Experience the next generation of AI customer experience with Voiceflow. We’re rolling out access on an invitation basis. To receive your invitation, visit the following link: www.voiceflow.com/v4