Concept in active development — from the founder's working notes.

Meetings are where a lot of real agreement and real work actually happens — and a lot of that gets lost to note-taking instead of being lived. SenseMeet is an early-stage vision for a companion that opens straight into distraction-free capture and turns what's said into useful drafts as the conversation happens, without ever losing sight of who owns the sensing.

What mutual sensing means

Most meeting tools produce a transcript afterward — one recording, filed away, rarely reopened. SenseMeet starts from a different premise: what if the room itself could sense what's happening while it's happening, shared by everyone in it rather than owned by whoever hit record? The working notes call each meeting a momentary holon (a temporary, bounded whole) — a container for what a group builds together that opens when the gathering starts and closes when it ends, distinct from any one person's ongoing notes.

That's also why the current thinking starts with in-person gatherings and hand-drawn notes before it reaches for the fully digital case. Most transcript tools already assume a screen is watching; the harder, more useful version is a companion for rooms where people are actually looking at each other.

How a SenseMeet-supported gathering flows

Opening

Reaching for the tool should never cost a beat of attention that belonged to the people in the room. The starting design idea is simple: opening the app never shows the last thing you were doing — no old chat, no old draft flickering past, just a direct path into capture, closer to a lock-screen quick-capture shortcut than to opening a regular app. Whether to be alerted while recording is a setting, not a default.

During

While the conversation is still live, the vision reaches further than a passive transcript: SenseMeet should be able to help shape a draft, a suggested edit, or a piece of visual material directly from what's being said — the way turning to someone next to you and asking "can you write that down as we go" would work, if that someone were always paying full attention. In most meetings today, the founder's notes observe, the work of actually reading the transcript afterward falls to AI, not to the humans who were in the room — because the humans mostly don't do it either way.

After

Once the meeting closes, a further pass folds in more context: the fuller conversation, and — with explicit permission — relevant context from the individual workspaces of the people involved. This is also where trust and permission settings matter most, since this step reaches past what any one person said out loud.

Two layers of memory

The working notes describe two layers for anything captured: an immutable layer — what was actually said or recorded, kept as the record of truth — and a lighter, editable layer on top for clarifications, corrections, and comments. If something needs correcting, it happens as a comment that can be flagged as the clarified version, without erasing what was originally recorded. Permissioning — who can access which layer, and how — is treated as one of the harder design problems, not a detail to bolt on later.

How to store any of this so it's actually worth retrieving later is, in the founder's own words, still an open question — not yet a solved one.

Trust and permission come first

Anything that listens while people talk has to earn that access. Consent, permission settings, and a clear sense of who can see what are treated as core to the design, not an afterthought — with a strong pull toward local and sovereign processing rather than handing conversations to a black box. The founder's working notes name a preference for local models where possible, and for European sovereign infrastructure — Infomaniak and Proton, specifically — for anything that has to leave the device.

How it's taking shape

The early technical direction leans on existing open communication infrastructure rather than building messaging from scratch — prototyping on the Matrix/Element protocol as the base layer that meetings happen inside. On top of that, an outliner layer (in the spirit of Tana) captures structured notes, and a visual layer (in the spirit of Excalidraw) captures freeform sketches and diagrams — the two are meant to converge, over time, into the same kidur infrastructure that also underlies SilkNotes. Meetings recorded this way are meant to be shareable inside a shared space and surfaced in a synced calendar view, regardless of which video or chat tool people actually meet in.

None of this is built yet. It's the shape the founder currently thinks it should take.

Who this serves

The niche the founder keeps returning to: meetings are valuable, but the actual work of making them useful afterward — reading the transcript, writing the follow-up, extracting the decision — increasingly falls to AI, not to the humans who were in the room, because humans mostly don't do that work either way. SenseMeet is aimed at groups who want that AI-mediated layer to work with the room's mutual sensing rather than around it — gatherings where the conversation itself, not just its record, is the point.

Part of a larger flow

SenseMeet is one holon inside In-Flow(.info), an umbrella and meta-layer for tools meant to let information move without friction. It sits alongside three named siblings:

All of them are meant to converge on kidur, the shared infrastructure layer underneath.

Where this stands

SenseMeet is a concept, not a product. Nothing described on this page is built. Some of it is fairly settled in the founder's thinking — distraction-free opening, the immutable-plus-comments memory model, the pull toward local and sovereign processing. Other parts are genuinely open questions the notes pose and don't yet answer: how to run the AI work on idle compute rather than constant always-on inference, how to store what's sensed so it's actually worth retrieving later, and how ownership of what gets built should work at all — with an inclination toward commons-based, federated stewardship rather than conventional ownership, though how that would work in practice isn't settled either.

The name and the first working notes came from the founder directly; building it out is understood to be a co-creation with whoever joins the effort.

Curious where this is headed, or want to build part of it with us?

Visit idea2.life →