SELF-HOSTED · SOURCE-AVAILABLE
The AI that sits in your meetings and files the follow-up.
Echo joins your calls, tells every voice apart — even in a shared room — and files a structured follow-up straight into your wiki and your team's inboxes. Your servers, your model keys, one-time license.
MaraLet's ship the beta on Friday.
TheoI'll own the migration.
AdaLoop in support before we launch.
- DecisionShip the beta on Friday
- ActionTheo — own the migration
- ActionAda — loop in support before launch
BUILT FOR OURSELVES FIRST
Echo started as a fix for our own worst habit: nobody writing the meeting down. It sat in our calls for months, got sharp, and now it's yours to run.
THE PROBLEM
Cloud notetakers make your meetings someone else's data.
Most cloud tools send every word to a vendor and bill you per seat. The private ones can't even join the call.
A copy of every conversation on a vendor's servers.
A per-seat bill that grows with your team.
Desktop recorders can't tell who's speaking in a shared room.
HOW IT WORKS
From calendar invite to filed follow-up
- 01
Connect your calendar
Point Echo at Google Calendar. It joins the meetings you invite it to.
- 02
It joins and transcribes
Echo sits in the call and labels each speaker by voice — even people sharing one microphone.
- 03
It writes the follow-up
At the end, your own model turns the transcript into a follow-up in your template.
- 04
It delivers the result
The follow-up lands in your own tools over MCP — a wiki, a tracker — or in attendees' inboxes.
CAPABILITIES
What it actually does
Joins the call
A participant on the meeting, driven by your calendar — not a recorder on one laptop.
Labels every speaker
Voice-based separation tells people apart even on a single shared microphone.
Decisions, not transcripts
It pulls out the decisions, action items and owners — never a raw wall of text.
A format per meeting type
Give standups, sales calls and 1:1s each their own follow-up template.
Any language, in and out
It detects the languages spoken and writes the follow-up in whichever language you set.
Delivers — and cross-checks
Sends the follow-up to your tools over MCP or by email. Filed into Kosmos, it's checked against past follow-ups and flags contradictions.
One console to run it all
A web admin panel for the day-to-day — no config files to hand-edit.
- Access rules — org-wide or per meeting
- Raw recording & full transcript on hand
- Voice-sample bank for in-room labels
HOW IT COMPARES
Echo vs. a typical cloud notetaker
| Feature | Echo | Typical cloud notetaker |
|---|---|---|
| Where recordings live | Your servers | A vendor's cloud |
| Pricing | One-time license | Per seat, every month |
| In-room speaker labels | Yes — by voice | No |
| Model keys | Yours | The vendor's |
| Source code | Available (BSL) | Closed |
OWNERSHIP
You own it — code included
Echo runs on your servers and your model keys — recordings and follow-ups never reach us. And you get the source we run (BSL): wire it into your stack, or let your engineers (or an afternoon of vibe-coding) bend it to how your team actually meets. Yours to keep — just not to resell.
Your servers · your keys · your code
PRICING
One license. Yours to keep.
Save $1,000 while the early-bird lasts.
- One-time — no subscription
- Source-available (BSL)
- Free updates through v1.0
- Install help included
Self-serve checkout coming soon — email us to buy now.
THE SUITE
Own the whole DelosX suite
Echo
Meeting follow-ups that stay on your servers — this product.
You're here Workspace · KosmosKosmos
Self-hosted tasks + wiki. Echo files follow-ups straight into it.
Explore Kosmos Support · HarmoniaHarmonia
Self-hosted AI support on your own model key.
Explore HarmoniaFAQ
Questions, answered
What does "source-available (BSL)" mean?
You get the source code we run internally under a Business Source License: run it, change it, self-host it — you just can't resell it as a competing product.
What does the one-time price cover?
A perpetual license for one legal entity, install help, and free updates through v1.0. After that you maintain your copy.
Do recordings ever reach DelosX?
No. Echo runs on your servers and your model key; audio, transcripts and follow-ups stay with you.
Is there an admin interface?
Yes — a web console to set access rules org-wide or per meeting, review the raw recording and full transcript, and manage the voice-sample bank used for in-room labels.
How does it label speakers in one room?
It separates voices from the audio and matches them against your voice-sample bank, so it can tell people apart even on a single shared microphone.
Where can it send the follow-up?
Anywhere reachable over MCP — a wiki, a tracker — or straight to attendees' inboxes by email.