Voiceflow
by Voiceflow, Inc.
AI agent platform for customer support teams, across chat, phone, SMS and WhatsApp
Voiceflow is a platform for building and running AI customer support agents on web chat, phone, SMS and WhatsApp. It is aimed at CX, support and engineering teams that want to control how an agent behaves, test it before release and hand conversations to their existing help desk or contact center.
Voiceflow is a Toronto company that sells a platform for building and running AI agents for customer support. It was founded in 2018 as Storyflow, a tool for interactive Alexa stories, and renamed later that year. Its focus has since moved from voice-assistant apps to customer service agents, and TechCrunch reported in August 2023 that sign-ups had roughly doubled after ChatGPT launched. An agent in Voiceflow combines a global prompt with goal-driven playbooks, where the model decides how to reach an outcome, and fixed workflows on a canvas for steps such as identity checks or refund rules. Agents call tools (prebuilt Zendesk, Salesforce, HubSpot and Shopify integrations, generic API calls, JavaScript functions and MCP servers) and answer from a knowledge base built from a company's documents. Teams choose models per agent or playbook from providers including Anthropic and Google, or use Voiceflow Core, the company's own model, released in May 2026 and updated to Core 4.2 that September. The same agent runs in a web widget with a voice mode, on phone lines, over SMS and through a REST API, and passes live conversations to Zendesk, Genesys, UJET, Kustomer or Dixa. Release tooling includes separate environments, traffic splits for A/B tests, simulated-persona tests and LLM-graded transcript evaluations. Voiceflow has raised $35M, most recently a $15M round led by OpenView at a $105M post-money valuation, and claims more than 3,000 customers. It competes with Google Dialogflow, Amazon Lex, Rasa and Botpress.
A CX or support operations lead with a technical owner on the team (an engineer or CX ops builder) who wants one customer-facing agent across chat and phone and wants to control its logic.
One agent definition that runs on web chat, phone and SMS, with scripted steps for compliance-sensitive flows and model-driven playbooks for open-ended questions, tested in staging before it reaches customers.
At a Glance
- Category
- AI Agents & Orchestration
- Pricing
- Subscription, Usage-based, Freemium
- Target Market
- CX and support leaders, Contact center operations, Enterprise developers, Agencies
- Deployment
- Cloud-first, API-based
- Founded
- 2018
- Headquarters
- Toronto, Canada
- Team Size
- 51-200
- Customers
- 3,000+ (vendor-reported, 2026); 450 per TechCrunch in August 2023
Key Features
- ✓Playbooks
Goal-driven sub-agents written as instructions, such as processing a return, where the model calls tools until the goal is met.
- ✓Workflows
A canvas of fixed steps for processes that must run the same way every time, like authentication or an IVR menu, and can embed playbooks.
- ✓Multichannel runtime
One agent runs in a web chat widget with voice mode, on phone numbers provisioned by Voiceflow or via Twilio, Vonage or Telnyx, over SMS and through a REST API.
- ✓Live agent handoff
Passes chat conversations to Zendesk, Genesys, UJET, Kustomer or Dixa, and forwards phone calls to a number or SIP address.
- ✓Model choice and fallback
Pick the model per agent or playbook from providers including Anthropic and Google, or Voiceflow Core, with fallback to another model during a provider outage.
- ✓Environments and testing
Separate development, staging and production environments with traffic splits, version rollback, simulated-persona tests and LLM-graded transcript evaluations.
- ✓Tools and integrations
Prebuilt Zendesk, Salesforce, HubSpot and Shopify actions plus API calls, JavaScript functions and MCP tools so the agent can check orders or update tickets.
Capabilities
Use Cases
- •Phone support automation
A contact center puts an AI agent on its support line that answers routine calls and forwards the rest to a human queue by number or SIP address.
- •Order and returns handling
An ecommerce team lets the agent look up orders in Shopify, check the return policy in a workflow step and create or update a Zendesk ticket.
- •Chat deflection with handoff
A support team answers common questions from its own help-center content in the web widget and hands unresolved chats to agents in Genesys or Kustomer.
- •Agency client deployments
An agency builds support agents for several clients in separate workspaces, white-labels the widget and hands each finished agent to the client.
- •Safe release of agent changes
A CX ops team tests a new returns playbook on 10 percent of live traffic in a separate environment before merging it into production.
Ideal For
Best For
- ✓Contact centers that need the same AI agent on the phone line and the website chat widget
- ✓Support teams handing escalations to Genesys, UJET, Kustomer, Dixa or Zendesk
- ✓Regulated businesses that want deterministic workflow steps for identity checks and refund eligibility
- ✓Agencies building and white-labeling support agents for several clients
- ✓Teams that want to switch LLM providers per agent without rebuilding the agent
Not Ideal For
- ✗Support teams whose volume arrives as email tickets inside a help desk, because Voiceflow runs its own chat and phone front door and its docs do not describe working inbound email tickets
- ✗Non-technical teams with no one to own playbooks, environments and evaluations; Product Hunt reviewers split on whether the learning curve is too steep
- ✗Buyers who must model cost before a sales call, since the pricing page publishes no plan prices and usage is billed in credits that vary by model
- ✗Teams on Intercom, Freshdesk, Gorgias, Front or Salesforce Service Cloud that need native live-agent handoff, which the handoff step does not list
Deployment
Market & Ratings
3,000+ (vendor-reported, 2026); 450 per TechCrunch in August 2023
Market Analysis
Pros
- ✓Mixes scripted workflow steps with model-driven playbooks, so risky steps such as refunds can be fixed while open questions go to the model
- ✓Production release tooling: separate environments, traffic splits for A/B tests, version rollback and LLM-graded transcript evaluations
- ✓Voice, SMS and chat share one agent, and phone numbers can be provisioned in-platform in the US and Canada
- ✓Vendor-stated SOC 2 Type II and ISO/IEC 27001:2022, with HIPAA-compliant deployments supported
Cons
- ✗No published business pricing, and credit consumption depends on model, conversation length and voice minutes, so cost is hard to forecast before real traffic
- ✗Handoff covers five platforms, and in September 2026 the Zendesk column of the handoff capability matrix was empty for extras such as wait-time messages and inactivity timeouts (per Macha's reading of the docs)
- ✗Learning curve: one Product Hunt reviewer called it too steep for most people, and getting value from playbooks and evaluations takes a technical owner
- ✗Trustpilot reviews are split between five-star and one-star, with the one-star reviews complaining about support at launch
- ✗Editor seats are a paid add-on, so the cost of the team maintaining the agent grows with headcount
Pricing
Free
$0
- ✓One-time credit grant that does not renew
Pro
From $60/mo
- ✓Monthly credits equal to the plan fee (third-party reported)
- ✓One editor seat included; extra seats reported at $50/mo
Business
From $150/mo
- ✓LLM fallback models
- ✓Priority support
- ✓Removal of Voiceflow branding on the widget
Enterprise
Contact for pricing
- ✓SSO
- ✓Private cloud hosting
- ✓Custom credit allotments
Voiceflow's pricing page lists no prices for businesses and routes them to a demo. Its billing docs describe a monthly plan fee, add-ons (editor seats, phone numbers, concurrent calls) and credit-based usage. Macha, a competing vendor, read Voiceflow's credits table in September 2026 as $0.005 per customer message and $0.05 per phone minute plus model tokens, and cited the Botpress review for Pro at $60 and Business at $150 a month. The same analysis quotes Vendr's median negotiated contract at $258,625 a year. Agencies get a free trial with no credit card.
Security & Compliance
Sources
This page was written from 9 sources, 4 on domains other than voiceflow.com.
- 1.voiceflow.com — voiceflow.comvendor
- 2.voiceflow.com — aboutvendor
- 3.voiceflow.com — ai infovendor
- 4.voiceflow.com — pricingvendor
- 5.voiceflow.com — billingvendor
- 6.techcrunch.com — voiceflow a platform for building conversational ai experien
- 7.betakit.com — ai startup voiceflow raises 15 million usd to help companies
- 8.getmacha.com — voiceflow complete guide
- 9.producthunt.com — reviews
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
Strawberry
AI browser for sales, recruiting and support teams, with shareable agent 'companions' that work inside your logged-in tabs
Temporal
Open-source durable execution platform that keeps long-running AI agents and workflows alive through crashes, timeouts and failed tool calls
Town
AI work assistants ("Townies") for every employee that handle email, calendar and routine work, and now hand tasks to each other across a team
Notion Agents
Autonomous AI agents inside the Notion workspace that run on triggers and schedules for up to 24 hours, with a model router and credit-based pricing