AI assistant, omnichannel support desk, knowledge base. Everything you need for seamless customer support, all in one place.
In 2017, a small team of developers built a customer support tool because they needed one. Two years later they did something that, in software, still counts as a small act of faith: they gave the code away. Chatwoot went open source under the MIT license, and the decision quietly reset the terms of the product. Anyone could download it, read every line, run it on their own servers, and never send a customer conversation to someone else's cloud. Today more than 15,000 businesses use it, the project has passed 35,000 stars on GitHub, and the three founders have taken a seat at a table crowded with companies that raised a hundred times more money.
The pitch is easy to state and harder to build: a modern support platform that competes with Intercom and Zendesk, except you own it. Chatwoot consolidates the places customers reach out - website live chat, email, WhatsApp, Instagram, Facebook, Telegram, Line, SMS - into one shared inbox where a support team works the way they already work in Slack. Assign a conversation. Tag it. Leave a private note. Hand it off. The novelty is not any single feature. It is that the whole thing is inspectable, self-hostable, and free at the core.
Customer messages arrive from too many directions. A shopper asks a question on the website widget in the morning, follows up over WhatsApp at lunch, then emails a receipt request at night. To the customer it is one relationship. To most companies it is three disconnected tools, three logins, and an agent who has no idea the three threads are the same person. Chatwoot's answer is consolidation done carefully: pull every channel into a single timeline, attach it to a customer profile, and let the whole team see the history.
The self-hosted angle solves a second, quieter problem. Support conversations are among the most sensitive data a company holds - names, order histories, complaints, sometimes health or financial details. Handing all of it to a third-party SaaS is a compliance question in some industries and a philosophical one in others. Chatwoot lets a company keep that data on its own infrastructure, which is why regulated businesses and privacy-minded teams keep showing up in its community.
The incumbents in customer support are formidable. Zendesk, Intercom, Freshworks, and Salesforce Service Cloud have deep feature sets and large sales teams. Chatwoot does not try to out-spend them. It tries to out-open them. Where the incumbents are cloud-only and priced per seat, Chatwoot ships its source code, lets you run it yourself, and keeps the core free. That single structural choice changes who trusts it and why.
There is a counterintuitive lesson buried in the model. Most SaaS tools try to become the system you can never leave; the lock-in is the business plan. Chatwoot handed customers the keys instead. Making it easy to walk away turns out to be a strange kind of retention - people stay because they chose to, not because they are trapped.
In 2024 Chatwoot shipped Captain, its AI system, and split it into two jobs. The Assistant is the bot customers actually talk to; it answers common questions using the company's own help content and resolves routine queries before they reach a person. The Copilot works the other side of the desk - it sits with human agents, drafting replies, translating messages, and summarizing long threads. By early 2026 Captain had moved directly into the reply editor, so an agent reads, drafts, refines, and sends without leaving the box.
The framing matters. Captain is not sold as a replacement for support teams. It is aimed at the repetitive slice of tickets - the "where is my order," the "how do I reset my password" - that never deserved a human's full attention. Take that off the queue and the harder conversations get the time they need.
Around Captain sits the rest of the platform: a customizable live-chat widget, a help center and knowledge base with AI-powered search, mobile apps for agents on the move, a voice channel added in 2025, and a developer API for teams that want to wire in their own workflows. It is a broad product for a company its size, and much of that breadth is a dividend of being open - contributors extend it in directions the core team never had to fund.
Chatwoot was founded by Pranav Raj S, who serves as CEO, Sojan Jose, the CTO, and Nithin David Thomas. The team has roots in India's FOSS community, and the company grew up distributed, with people across the United States and India. The tool began in 2017 as a proprietary project, went open source in 2019, and the founders formed a company around it before joining Y Combinator's Winter 2021 batch.
The money followed the community, not the other way around. After YC, Chatwoot raised a $1.6 million seed round in August 2021 from Goat Capital, Uncorrelated Ventures, Y Combinator, Hack VC, and angels including Lambda School's Austen Allred. It is a modest number next to competitors that have raised into the hundreds of millions, and that constraint reads less like a limitation than a design choice. When you cannot outspend the field, the open-source repo becomes the growth team.
Chatwoot runs on an open-core model. The heart of the platform is free and open source, so any team with the appetite can self-host it at no license cost. The company makes money on the parts people would rather not run themselves: a hosted cloud version with tiered subscriptions, usage-based credits for Captain AI, and enterprise plans with support. Free earns the trust and the audience; the hosted product monetizes the segment that values not touching a server.
It is a well-worn path - the same shape that turned other open-source projects into companies - but it works here because the free tier is genuinely complete, not a crippled demo. A startup can run the whole thing for years without paying, and that is precisely what makes the paid tiers credible when the startup grows up.
The founders' deepest skill is not a particular technology. It is the discipline of building in public. Every design decision lives in a repository other people can question, fork, and improve, which imposes a rigor that closed teams can skip. A support platform touches messy edges - inbound email parsing, WhatsApp's business rules, real-time chat at scale, agent permissions, reporting - and Chatwoot has spent years hardening those edges with help from contributors who run the software in production and file the bugs they hit. That feedback loop, thousands of real deployments reporting back, is a form of expertise money cannot buy quickly.
It also shapes the roadmap. Voice arrived because customers asked for it. Captain's move into the reply editor came from watching agents toggle between windows. The company's advantage is proximity to the people actually doing support work, and a codebase open enough that those people can shorten the distance themselves.
Chatwoot occupies a specific corner: teams that want a serious, omnichannel support platform but also want control - over cost, over data, over the code itself. That includes developers who trust software they can read, companies in regulated industries that cannot ship conversations to a third party, and cost-conscious startups scaling their support without scaling a per-seat bill. Against Intercom's polish and Zendesk's enterprise reach, Chatwoot's edge is not a longer feature list. It is a different answer to a question the incumbents stopped asking: whose software is this, really?