CopilotKit
CopilotKit builds the enterprise agentic frontend stack: open-source SDKs and protocols that connect AI agents to user-facing applications, letting teams add chat, generative/interactive UI, shared state, human approvals, persistence and agent experiences inside real products. Its stated vision is that all UI is becoming AI and that CopilotKit should be the layer where humans and agents collaborate.
At a Glance
- Enterprise software teams
- Developers and product teams building agentic applications
- Fortune 500 and Global 50 companies
- Companies deploying employee AI coworkers and agent workspaces
- +1 more
AI Tools by CopilotKit
(3)
OpenDots
Self Hosted AI Coworker Template
Channels SDK
AI Agent Messaging Platform SDK
CopilotKit
React SDK for Agentic AI Apps
Discussions
No discussions yet
Be the first to start a discussion about CopilotKit
Latest News
Introducing OpenDots, the open-source alternative to OpenAI Dots
Introducing AG-UI 1.0, a stable spec for connecting any agent to any application
New and improved CopilotKit Inspector with Automatic Learning
Introducing WebMCP for CopilotKit
Products & Services
Framework for embedding AI agents in applications with chat, generative UI, in-app actions, shared state, human-in-the-loop approvals and headless UI; supports React, Angular, React Native and other surfaces.
Open Agent-User Interaction protocol carrying streamed messages, tool calls, agent state, multimodal results, subagents, metadata, interrupts and token usage between agent backends and user-facing applications.
Persistence layer and managed/self-hosted platform for AG-UI Streams, rich threads, cross-device sync, user memories, channels, observability, analytics and automatic learning/self-improvement.
SDK and managed connections for bringing the same agent to Slack, Microsoft Teams, text, calls and other channels, with tools, memory and cross-platform conversation history.
Market Position
CopilotKit positions itself as an open, self-hostable agentic frontend stack rather than a model provider or closed copilot builder: it connects any agent backend to any user-facing app through AG-UI and renders generative UI, approvals and shared state. Its competitive set includes frontend/agent UI frameworks such as Vercel AI SDK and assistant-ui, agent application platforms such as LangChain/LangGraph and Mastra, and closed enterprise builders such as Microsoft Copilot Studio; CopilotKit differentiates through its open protocol, broad backend/framework compatibility, in-product interaction layer and enterprise self-hosting.
Leadership
Founders
Atai Barkai
Co-founder and CEO. Previously worked on media infrastructure at Meta and led development of flagship iOS apps at Doximity; studied physics at the University of Pennsylvania (BA and MSc).
Uli Barkai
Co-founder and Head of Growth & Partnerships. He leads growth, partnerships and ecosystem efforts; he and Atai are brothers and previously co-founded tawkitAI, an AI-powered podcast platform.
Executive Team
Atai Barkai
Co-founder & CEO
Former Meta media-infrastructure engineer and Doximity iOS-app development lead.
Uli Barkai
Co-founder & Head of Growth & Partnerships
Leads growth, partnerships and ecosystem efforts; co-organizes AI Tinkerers San Francisco.
Board of Directors
Founding Story
The brothers originally built tawkitAI, an AI-powered podcast platform, then open-sourced internal infrastructure for copilots and saw strong developer interest. They joined Techstars Seattle's 2023 cohort and pivoted/renamed the company to CopilotKit, pursuing the vision that agents and humans would collaborate inside software and that an interaction layer was needed between agents and users.
Business Model
Revenue Model
Open-source SDKs and protocols are offered freely, while CopilotKit monetizes hosted/managed CopilotKit Intelligence and enterprise services through subscription plans, per-seat team pricing, custom enterprise contracts and AWS Marketplace availability.
Pricing Tiers
One developer; 3-day thread retention, 200 max threads, 1 GB multimodal storage, 1 Slack/Teams org, 5 channels and 500 total channel credits.
5-day thread retention, 5,000 max threads, 10 GB storage, 1 org, 10 channels and 2,000 channel credits/month.
Up to 5 seats; 14-day thread retention, 25,000 max threads, 100 GB storage, 2 orgs, 20 channels and 5,000 channel credits/month; includes self-hosted database and dedicated Slack support.
Unlimited max threads, custom retention/storage/channels/credits, VPC/on-prem included, analytics and self-learning early access, dedicated engineering support and SLA; available through AWS Marketplace.
Target Markets
- Enterprise software teams
- Developers and product teams building agentic applications
- Fortune 500 and Global 50 companies
- Companies deploying employee AI coworkers and agent workspaces
- Industries including financial services, telecom, healthcare, technology and consumer products
- Embedding AI agents and copilots inside enterprise web and mobile products
- Generative UI for workflows that require forms, tables, charts or multi-step interactions
- Human approval and auditable agent actions
- Persistent employee agent workspaces and AI coworkers
- Bringing agents to Slack, Teams, text and voice calls
- Research, drafting, document editing and other specialist-agent workflows
- DocuSign
- S&P Global
- Cisco
- Deutsche Telekom