loading...
OpenCopilot lets any cloud service embed a conversational AI assistant that can answer questions and take actions – without the user ever leaving the page.
ENTRY ANGLES
Vertical-specific AI copilot platforms optimized for particular SaaS categories · Pre-trained AI assistants leveraging domain-specific vocabulary and data schemas · Category-specific copilot as the default developer choice for adding AI assistants
VERTICALS
CAPABILITIES
Domain-specific AI model training and fine-tuning, Understanding of vertical-specific user actions and data schemas, Developer integration and distribution capabilities
OPENCOPILOT FOUNDER
“add a 10% discount to all underperforming products.”
OpenCopilot lets cloud service developers add AI assistants to their products. The assistant appears as a chatbot that expands when a user clicks a button on the service's page.
Users can ask the assistant questions or instruct it to take action. On an e-commerce platform, for instance, a user could ask "what's the status of order #1234," request "give me last month's sales data as a table," or even say "add a 10% discount to all underperforming products."
Embedding the assistant window takes 10 lines of JavaScript in the page template.
To integrate the AI core with their service, developers can use OpenCopilot's API or SDK, the hosted cloud version, or install the software on their own servers.
Self-hosting is possible because OpenCopilot is open source – available on GitHub. Open source means developers can inspect what they're installing and have confidence about what data they're feeding the system. It also means all OpenCopilot users can contribute bug fixes and new features.
The free tier is capped at 1,000 requests per month. Higher volumes are available on custom pricing – contact the team directly.
Major cloud platforms have already started integrating OpenCopilot. API hosting platform Postman built a chatbot that helps developers add APIs to the platform, write tests, and generate documentation. A Shopify integration helps store owners monitor and grow their sales.
OpenCopilot incorporated this year and has just been accepted into Y Combinator, receiving $550K in seed funding.
The core reason developers would want to embed OpenCopilot is to reduce user churn – by making the product easier to adopt and easier to use every day.
The idea of in-product guidance isn't new. WalkMe pioneered it – building a platform for embedding contextual arrows, tooltips, and step-by-step walkthroughs onto service screens. B2B services became its biggest audience, as companies needed new hires and existing employees to adopt software quickly. WalkMe went public in 2021 at a $2.56B valuation and was later acquired by SAP for $1.5B – evidence that the underlying problem is real and valuable enough to support an enterprise-scale company.
But ChatGPT shifted expectations. Users now expect to type a question and get a useful answer – not hunt through a series of arrows and tooltips. OpenCopilot's positioning is simple: the chat interface is the better interaction model, and it's time to build a platform around that assumption rather than retrofitting it as a feature.
The positioning bet is sharpest when you consider the competitive dynamics. Intercom, Zendesk, and similar platforms have added AI chatbot capabilities – but as layers on top of legacy architectures built for ticketing and documentation. For them, AI is one feature. For OpenCopilot, it's the entire product. The incumbents are weighed down by existing contracts and infrastructure; a chat-native entrant starting clean has structural freedom they don't.
Direqt, [covered here](/review/tam-odno-mesto-a-zdes-mnogo) previously, is playing a similar positioning game in a different vertical: a platform for online publications to embed AI chatbots on their pages, explicitly to keep readers on site longer. That startup raised $5.9 million.
It feels increasingly likely that every serious web application will have an AI assistant within the next few years. The question is who captures the "platform" position in each category – the default choice when a developer wants to add an assistant without building one from scratch.
OpenCopilot is trying to own that position for cloud services broadly. Direqt is trying to own it for online publications. The pattern suggests the space will segment further: a platform purpose-built for e-commerce stores, another optimized for online learning, another for developer tools, and so on. Each vertical has its own vocabulary, its own common user actions, its own data schemas – and an assistant pre-trained on those specifics will outperform a generic one.
That segmentation is the opportunity. The horizontal platform (build an AI assistant for anything) is a hard fight against well-resourced incumbents. The vertical platform (build the AI assistant specifically for SaaS accounting tools, or for project management, or for HR software) is a more defensible starting position – and the category is early enough that most of those verticals are unclaimed.