File
SMF·KUMOSPACE
Received on
31.07.2026
Reviewed on
28.09.2026
Exhibits annexed
3
Questions
4

SMF·KUMOSPACE

Can a prompt replace Kumospace?

Community platforms — online communities and team communication

Not yet Verdict recorded on 28.09.2026 · Verified on 31.07.2026
Price
$16/moSource: kumospace.com · Checked on July 31, 2026
Per year
$192
Build time
One sitting
Votes
0 votes
YesAlmostNot yet (checked)

Exhibit tracking slip

Exhibit A The prompt
Exhibit B What you lose
Exhibit C Why people still pay: network, moderation, and communication infrastructure
Exhibit Q Questions

Verdict

Proximity audio for a handful of people is reachable: put avatars on a map, open a peer connection when two get close, close it when they separate. The wall is arithmetic. A full mesh needs every participant connected to every other, which collapses somewhere around six people, and going beyond that means running a selective forwarding server plus TURN relays for anyone behind a restrictive network. That is real-time media infrastructure with a bandwidth bill, and it is the product.

Exhibit B — What you lose

Exhibit A — The prompt

Received on31.07.2026
Build a proximity-audio virtual room for a small team. Audio only, mesh only, and honest about the ceiling.

Room: a 2D floor plan rendered on a canvas, with a background image the operator supplies and named zones drawn as polygons. Each participant has an avatar moved with arrow keys or by clicking, with position broadcast over a WebSocket at a modest tick rate and interpolated between ticks so movement looks smooth without flooding the channel.

Audio, which is the whole product. Every pair of participants within a hearing radius holds a WebRTC peer connection. Volume is a function of distance: full inside an inner radius, falling to silence at the outer radius, using a curve that is gentle near the speaker and steep at the edge so walking away is a clear signal. Open a connection when two avatars come within the outer radius plus a margin, and close it after they have been separated for a few seconds — hysteresis, so people standing at the boundary do not thrash connections.

Zones: entering a named zone puts you in a group where everyone hears everyone at full volume regardless of distance, which is how a meeting room works. Leaving reverts to distance-based audio.

Signalling and connectivity: a small signalling server over WebSocket. Configure a STUN server, and document that a TURN relay is required for participants whose network blocks direct connections — roughly one in five in corporate environments — and that TURN relays all their audio through a server you pay for. Include the TURN configuration and say plainly what it costs.

The ceiling, stated in the interface rather than hidden: a full mesh needs n(n-1)/2 connections, so eight people is 28 streams per participant to encode and receive. Enforce a hard participant cap, default six, and display why. Adding more means a selective forwarding unit, which is a different system; name mediasoup or LiveKit in the README as what that step actually requires.

Presence: a status per participant (available, focused, away) and a knock action that requests a private conversation, which is the interaction that makes a virtual office feel less rude than a call.

Also build: push-to-mute, a device picker, an audio-level meter per participant, and a persistent room state so the floor plan survives a restart.

Out of scope: video of any kind, screen sharing, recording, text chat, and any hosted multi-tenant deployment.

Opening prefills the prompt — press enter to run it.

Exhibit B — What you lose

  • B.1 more than about six people in a room at once
  • B.2 video, which multiplies the bandwidth problem by an order of magnitude
  • B.3 connection reliability for participants behind corporate firewalls
  • B.4 screen sharing, recording and the meeting features an office needs

Prior art

Exhibit C — Why people still pay: network, moderation, and communication infrastructure

Because real-time media is a service, not a feature. Servers close to every participant, relays for the ones whose network blocks direct connections, and capacity that does not degrade at the eleventh person are what the subscription actually buys.

Questions

Is there anything to migrate from Kumospace?

No. A virtual office holds no durable data beyond its floor plan, so switching means rebuilding the room layout, which is an afternoon. Nothing of value is trapped.

Why audio only?

Because video is what makes the mesh collapse. Audio streams are tens of kilobits; video is megabits, and a full mesh multiplies that by every participant. Audio-only also happens to be what most people actually leave running in a virtual office all day.

What does it cost to run?

A small VPS for signalling is a few dollars. TURN is the term that bites: relayed audio for participants who cannot connect directly is metered bandwidth, and a team where several people are behind restrictive firewalls will relay most of the traffic. Budget for it rather than assuming STUN suffices.

What is the one thing that does not survive the rebuild?

Everyone being able to join. Kumospace works from a hotel wifi, a corporate VPN and a phone tether, because it runs relay infrastructure in enough places. A mesh with one STUN server works beautifully for the people whose networks cooperate and not at all for the ones whose networks do not.

Receipt

Already built this yourself?