Ink & Switch at Ten: Rethinking Who Holds Data and How Tools Change
Ink & Switch has reached ten years of operation in 2026. The organization describes itself as an independent research lab exploring the future of tools for thought. Ink & Switch
Its stated vision is a new computer that amplifies human intelligence and helps people think more clearly and collaborate more effectively. The emphasis is on cognition and coordination, rather than throughput or automation.
That vision is organized into four research areas. Local-first software covers software architecture that returns data to users and enables collaboration in every tool, with the primary copy on user devices. Malleable software covers design of software environments where people can customize tools in the moment, during use. Programmable ink covers a dynamic medium for sketching ideas where adding behaviors is as natural as applying ink to paper. Universal version control covers tools to help people explore alternatives, keep track of history, and collaborate better across media.
The current prototypes map to those areas. Ambsheets is described as spreadsheets for exploring scenarios and possibility spaces for making better decisions. Embark is described as dynamic documents for making plans, exploring how live data and computation can enrich informal travel plans. Inkbase is described as programmable ink, exploring what would be possible if hand-drawn sketches were programmable like spreadsheets.
Livelymerge continues the collaboration work at the systems level. It is described as work toward a collaborative programming kernel, a shared core where the entire object memory, the live data a program is working with, is shared simultaneously by multiple users. Keyhive addresses trust. It is described as a local-first access control system with capabilities, cryptographic permissions that can be passed between users, and end-to-end encryption, which keeps data readable only to participants, for secure collaboration. The lab also lists its local-first software essay, titled "You own your data, in spite of the cloud," about collaborative software that lets users retain ownership of their data.
Two efforts are listed as production software. Allume is a digital workspace to brainstorm, organize, and connect ideas visually using a flexible canvas for notes, sketches, and PDFs. Automerge is a library for building collaborative applications that sync changes across devices, even offline, using conflict-free replicated data types (CRDTs), data structures built to merge concurrent edits without a central server. Ink & Switch
The lab states its ongoing research is supported by individual and corporate sponsors and grants from bodies around the world. To mark the tenth year, it states it designed an interactive art piece with help from friends as a celebration. Visitors are invited to play with the piece by clicking and dragging everywhere.
The broader context here for working technologists is ownership plus mutability plus history. Local-first moves authoritative state to user devices. Capabilities and end-to-end encryption move enforcement to the edge with it. CRDTs supply the merge rules to keep that edge state consistent without a central arbiter. Malleable software and programmable ink then ask what interfaces should sit above that base when behavior can be defined during use rather than before it.
In my view, that stack deserves close attention because it reverses assumptions that still shape most product architectures. Cloud sync, permission checks on a server, and version history as an afterthought become edge data, cryptographic delegation, and history as a basic building block. The prototypes give concrete places to test the idea, from scenario modeling in Ambsheets to live planning in Embark, sketch logic in Inkbase, and shared runtime state in Livelymerge. None require accepting the full vision at once. Each can be judged as an architectural choice about where state lives, who can change it, and how divergence is reconciled.
On a practical level for teams building collaborative tools, the path from research to dependency is the point to watch. A canvas or a demo is easy to admire and discard. A sync library with offline support and CRDT conflict resolution enters build files and incident reviews. That makes Automerge and Keyhive the pieces to prototype with first, while the ink and malleability work can guide longer-term interaction design. If the next decade of tools for thought gives users more direct control over their data and their tools, it will likely arrive through that kind of step-by-step adoption rather than a wholesale new computer.
