Last updated: 2026-08-26
Open is a YC-backed (W24) AI customer support platform that automates 77% of support tickets on average across voice, chat, email, WhatsApp and SMS. It uses Agentic Workflows that call a company's own APIs to complete actions like refunds, not just answer questions, and includes an MCP Server for Claude Desktop, Cursor and VS Code.
About Open
Open is an AI-powered customer support platform built by Y Combinator company Open (YC W24, founded 2024 by Mohammad Gharbat), used in production by Print.com, Mollie.com, Viva.com, More.com, Orderchamp.com and OneCarNow.com. The platform automates an average of 77 percent of a company's support ticket volume across voice, chat, email, WhatsApp, SMS, Slack and social channels, aimed at high-volume support queues that would otherwise need a growing team of human agents. Instead of scripted chatbot flows, Open trains a single AI agent from more than 50 connected knowledge sources (Notion, Confluence, Zendesk Help Center, Google Docs, Slack, Snowflake, BigQuery and more) and gives it Agentic Workflows that call a company's own internal APIs to take real actions like issuing a refund, updating an account or rebooking a shipment, not just answer questions. The agent runs across every channel from one unified inbox, supports 100+ languages with mid-call language switching on voice, and connects into existing helpdesks (Zendesk, Salesforce Service Cloud, Intercom, Freshdesk, HubSpot Service Hub) instead of requiring a migration off them. Open is built for high-volume support operations at banks, airlines, e-commerce platforms and payments companies where ticket volume outpaces headcount, and for support operations leaders trying to raise automation rates without lowering CSAT. Customer TicketSwap reports automating more than 67 percent of support across 16 million users in 42 countries using the platform. Engineers using Claude Desktop, Cursor or VS Code can reach that same agent through Open's MCP Server without switching tools, an integration path most customer-support AI vendors do not expose. Open bills on a usage basis tied to confirmed resolutions rather than a flat monthly fee, with full pricing detail in the FAQ below and a custom rate for customers above 100,000 resolutions a month. There are no seat fees and no monthly platform fee. The platform is web-based, and voice support runs on Open's own cloud telephony stack with local numbers available in more than 100 countries. Open raised a $1.52 million pre-seed round that closed in January 2025 (Y Combinator, Spacetime, Mollie CEO Koen Koppen and Mollie co-founder Adriaan Mol among the backers) and a later $7 million round co-led by Y Combinator and X by Unifonic with Shorooq and Sadu Capital participating. On the compliance side, Open holds a SOC 2 Type I certification with Type II in progress, is ISO 27001 certified, states GDPR compliance with EU data residency, and describes its platform as supporting HIPAA-compliant patient support for healthcare customers.
Pricing
Usage-based: $0.75-$0.90 per fully resolved ticket, with the rate decreasing automatically as monthly volume rises. No seat fees, no monthly platform fee, no per-message fees; carrier minutes for voice are billed at cost (about $0.10/resolution). Tickets escalated to a human agent are not billed. Enterprise volume above 100,000 resolutions/month gets a custom rate. Backed by a stated $2 million refund guarantee tied to implementation and service.
Key Features
- Agentic Workflows call your own APIs: The agent triggers real actions inside a company's own systems, such as issuing a refund or updating an account, through a no-code workflow builder that ships with more than 50 templates.
- Six-plus channel deployment from one agent: A single AI agent handles voice, web chat, email, WhatsApp, SMS and Slack from one unified inbox instead of a separate bot per channel.
- MCP Server for Claude Desktop, Cursor and VS Code: Engineering teams configure and query the support agent directly from Claude Desktop, Cursor or VS Code through Open's Model Context Protocol server.
- 50+ knowledge source connectors: The agent trains directly from Notion, Confluence, Google Docs, Slack, Zendesk Help Center, Snowflake and BigQuery among more than 40 other sources.
- Cloud telephony in 100+ countries: Voice support runs on Open's own telephony stack with local phone numbers in more than 100 countries and mid-call switching across 100+ languages.
- Outcome-based pricing: Customers are billed solely on confirmed ticket resolution, with human handoffs excluded from the bill entirely (see price_notes for exact rates).
Pros
- Bills solely on confirmed resolution and human-escalated tickets cost nothing, unlike per-seat helpdesk add-ons that charge regardless of outcome (exact rate in price_notes).
- Runs one agent across 6+ channels (voice, chat, email, WhatsApp, SMS, Slack) from a single inbox rather than needing a separate bot per channel.
- Exposes an MCP Server so support-agent setup lives in the same developer tools (Cursor, VS Code, Claude Desktop) engineers already use, instead of a separate vendor admin panel.
- Trains its agent from more than 40 knowledge-source connectors (Notion, Confluence, Zendesk Help Center, Snowflake, BigQuery), so existing documentation rarely needs re-authoring to go live.
Cons
- Pricing is entirely usage-based with no published flat monthly tier, so budgeting for spiky ticket volume (launches, outages) means forecasting per-resolution spend rather than a fixed cap.
- SOC 2 Type II is still in progress (only Type I is certified as of this review), which can slow enterprise security reviews that require the completed Type II report.
- Vendor material and third-party coverage both note the agent still hands off nuanced or highly complex L2/L3 issues to a human, and automation rates climb only after the agent is trained on existing knowledge sources.
- Independent third-party review volume (G2, Trustpilot) is thin for a company only two years past its YC batch, making it harder to benchmark satisfaction against established players like Intercom Fin or Ada.
Frequently Asked Questions
How much does Open (open.cx) cost in 2026?
Open charges $0.75 to $0.90 per fully resolved ticket, with the rate dropping automatically as monthly volume rises. There are no seat fees, no monthly platform fee and no per-message fees, and carrier minutes for voice are billed at cost separately, around $0.10 per resolution. Customers above 100,000 resolutions a month get a custom enterprise rate through sales.
Is Open free to use?
No. Open has no free tier or published free trial; every plan runs on the pay-per-resolution model. Tickets that the AI cannot resolve and hands off to a human agent are not billed, so cost only accrues on confirmed AI resolutions.
What are the best alternatives to Open?
Intercom Fin suits teams that want AI bundled directly into their existing helpdesk. Decagon fits large enterprises needing deep Salesforce or Zendesk-side integration. Ada is a fit for large multilingual retail or telecom deployments at scale.
How does Open compare to Intercom Fin in 2026?
Intercom Fin runs as part of Intercom's own helpdesk, so it suits teams already standardized on that platform. Open instead sits on top of whatever helpdesk a company already uses (Zendesk, Salesforce, Freshdesk, HubSpot) and adds voice, phone numbers in 100+ countries and Agentic Workflows that call internal APIs, billed per resolution rather than per seat.
How do you get started with Open?
Open's own quickstart claims a live AI agent can be on a company's site in under 10 minutes, connecting to one of 50+ knowledge sources such as Notion, Confluence or a Zendesk Help Center. Enterprise rollouts across voice, phone numbers and existing helpdesk integrations go through a named solutions engineer included in the standard rate.
Top Alternatives
- Intercom: Pick Intercom if you want AI bundled with the helpdesk itself; pick Open if you want an agent that works on top of the helpdesk you already run.
- Decagon: Pick Decagon for a large-enterprise chat/email deployment with deep Salesforce or Zendesk integration; pick Open if voice and outcome-based pricing matter more than enterprise CCaaS depth.
- Ada: Pick Ada for a mature multilingual deployment at large retail or telecom scale; pick Open for agentic actions across voice plus chat with usage-based billing.
- Sierra: Pick Sierra if you want a fully managed, white-glove AI support service; pick Open if you want self-serve setup and transparent per-resolution pricing.