AFFiNE started as an open-source project, and it has grown into a community far beyond the codebase: 68.3k GitHub stars, more than 7 million downloads from all over the world.
Many of you use AFFiNE to write, draw, plan, organize databases, build personal knowledge systems, and think across documents and canvases. Thank you for trusting AFFiNE with that work.
Today, we want to share an important team update and, more importantly, what comes next for AFFiNE.
As part of this next stage, JiaChen He is stepping away from his role at AFFiNE to build Bridge, a new product focused on agent-native workflows.
AFFiNE will continue under Vincent's leadership. Vincent has been part of AFFINE's early team and has worked closely on the product’s direction from the beginning.
This is a transition, but not a reset. AFFiNE will continue with the same commitment to open-source, local-first, user-owned knowledge work, and a clearer focus on becoming a more open, interoperable workspace.
AFFiNE's core product will continue.
Docs, whiteboards, databases, local-first workflows, and open-source development remain central to AFFiNE. Many users choose AFFiNE because it brings writing, visual thinking, and structured information into one workspace. That foundation is not changing.
The upcoming changes are not a move away from AFFiNE's core. They are meant to make AFFiNE more open, easier to extend, and easier to connect with the tools users already trust.
The next stage of AFFiNE will focus on four areas: open files, flexible AI, agent interoperability, and a more open-source foundation.
We are working toward direct, real-time reading and writing of Markdown files.
Markdown should not be treated only as an import or export format. The goal is for users to work with .md files directly while still using AFFiNE's editor, database, and canvas experience.
For users who already rely on file-based workflows, this should make AFFiNE easier to use alongside tools like Obsidian, code editors, local agents, and other knowledge systems.
AFFiNE should not lock your knowledge into one app. It should make your knowledge easier to move, edit, connect, and reuse.
We also want AFFiNE to work better with the AI tools users already choose.
BYOK is part of this direction. Users should be able to bring their own API keys for supported AI providers, choose the models they prefer, and keep clearer control over cost, access, and data flow.
This is not about turning AFFiNE into a closed AI workspace. It is about giving users more control over how AI fits into their existing workflow.
This direction includes:
BYOK for supported AI providers
More flexible model choices
Clearer control over what AI can access
Better integration between AFFiNE AI and user-owned workspaces
We are also exploring better integration with third-party agents.
As agent tools become more capable, users need workspaces that agents can read from, write to, and help organize without taking control away from the user.
AFFiNE is well positioned for this because it already combines documents, databases, and whiteboards. The next step is to make that workspace more accessible to external agent workflows while keeping user control and data ownership at the center.
In practice, this means AFFiNE should become easier to use with the following:
Local Markdown files
Third-party AI agents
External automation workflows
User-controlled storage and backend options
We are also planning changes to AFFiNE's open-source structure.
Our direction is to open up third-party storage and backend options, separate enterprise permission and admin controls from the core product, and move AFFiNE from an open-core model toward a fully open-source repository under Apache 2.0.
This should make AFFiNE easier to self-host, easier to inspect, easier to extend, and easier for the community to trust.
We will share more details as these changes are ready.
We will also make Bridge invite codes available to AFFiNE users who registered before this announcement. You can claim your invite code by following the instructions in upcoming emails.
This is a thank-you to the community, not a change to AFFiNE's product direction. AFFiNE and Bridge will move forward as separate products.
Bridge is focused on helping AI agents complete real tasks on your computer. Some AFFiNE users may be interested in that direction, especially as knowledge work and agent workflows become more connected.
We will share invite details with eligible users soon.
AFFiNE is continuing.
The next chapter is about open files, open-source infrastructure, flexible AI choices, and agent-ready workflows.
Thank you for using AFFiNE, contributing to the project, sharing feedback, and helping shape what an open knowledge workspace can become.
Vincent Chen