Product story
Building an AI that remembers
Whisperer began as an overlay that answered questions mid-call. It now holds the context around your work — meetings, knowledge, tasks, calendar — and acts on it. Every release since May 2026, in order.
- 31Releases
- 5Months
- 16Languages
The three that changed the product
Everything else follows from these.
Live translation, with nothing joining the call
Simultaneous voice translation on macOS and Windows. No bot appears in the meeting and nobody is asked to approve one: the translation runs on the audio already on your machine.
Read the storyThe assistant is called Leo
A name, a memory that survives between conversations, tools it can run in other services, and a permission gate on every one of them. Leo knows who you are and what time zone you are in.
Read the storyWork gets a place: the Hub
Meetings, tasks, files and knowledge gather around the thing they are actually about. A meeting finds its own hub from what was said in it, and the hub shows what is missing rather than what happened.
Read the storySix threads
Each one runs the whole length of the timeline. Follow a single thread from its first version to what it is now.
Everything, in order
Dates come from the change that shipped the work and are shown by month: the day code lands and the day you can use it are rarely the same day, and we would rather be honest than precise.
September 2026
-
The Hub starts reasoning
A ledger of what was promised, three machines that read it, and a synthesis after each meeting that Leo can speak from.
-
The Hub opens to every plan
Understanding a meeting stopped being a paid gate and became a setting.
-
Leo reaches you with the app closed
Push notifications, so a reminder that matters does not sit waiting for you to open a tab.
August 2026
-
Work gets a place: the Hub
Meetings, tasks, files and knowledge gather around the thing they are actually about. A meeting finds its own hub from what was said in it, and the hub shows what is missing rather than what happened.
-
Leo joins the live call
A rule for when interrupting is worth it, and memory of the conversation as it happens rather than a replay afterwards.
-
Send the result to someone without an account
What a meeting produced opens by link for people who do not use Whisperer, with the visibility you choose.
-
Obsidian plugs in
Your vault becomes a knowledge source through a real plugin, not a folder you remember to upload.
-
The assistant is called Leo
A name, a memory that survives between conversations, tools it can run in other services, and a permission gate on every one of them. Leo knows who you are and what time zone you are in.
-
The knowledge base opens outward
Forty-five sources — your documents, drives and wikis — indexed and searched by meaning, then mixed into answers as they are written.
-
Streaming speech recognition
Two streaming models for every paid plan and a premium recogniser for Max. The tail of a recording stopped going missing.
-
Translated subtitles, live
One streaming model translates while the sentence is still being spoken. The line closes on a pause in the text, not a pause in the speech — so a short reply no longer waits for the next one.
-
A teleprompter in the overlay
Your script scrolls itself while you speak, in the panel you already have. No second window, and nothing new on screen for anyone watching.
-
A shared surface in the call
Draw together inside the video room — and the board ends up in the recording with everything else.
July 2026
-
Notion, Slack and HubSpot connect
Connect once in your account, and a meeting note lands on the right record without anyone copying it across.
-
Skills you can attach
Reusable instruction sets — ready-made, or pulled from your own GitHub repository — attached to the prompts and roles you already use.
-
From chat to agent
Rebuilt on LangGraph: one memory, tools it can actually run, an audit trail behind every action, and server-side confirmation before anything touches your data.
-
The assistant gets a voice
Talk to it instead of typing, streamed through the backend rather than whatever speech engine the browser happens to ship.
-
Live translation, with nothing joining the call
Simultaneous voice translation on macOS and Windows. No bot appears in the meeting and nobody is asked to approve one: the translation runs on the audio already on your machine.
-
Video rooms inside the event
The room lives in the calendar event, guests join by link, and the recording becomes a meeting the moment it ends.
-
Files live here too
Folders, search and drag-and-drop. A room recording lands in storage and comes back as a meeting with a transcript.
-
Decisions become cards
A kanban tracker with projects, due dates, priorities and labels — fed by what was actually agreed on the call.
-
A calendar that syncs both ways
Events, invitations and reminders, with two-way sync to Google Calendar, Outlook and CalDAV.
June 2026
-
Claude, ChatGPT and Cursor get a door in
An MCP server on your account: outside AI tools read your meetings and act on them, inside the boundary you set and no wider.
-
Meetings become a place, not a log
Folders and tags, notes, templates, AI settings per meeting, recording straight from the browser, and import of audio or video you already have.
-
Calls become maps
Topics, decisions and action items pulled out of the transcript and laid out as a mind map you can edit.
-
Install it from the browser
Whisperer on the home screen as a standalone app. No store, no download page.
-
The overlay learns to translate
A redesigned macOS panel with a translation strip: the other side speaks their language, you read yours while they are still talking.
-
Windows gets a native client
Not a wrapper around the web app: C# on WPF, WASAPI loopback for the other side's audio, DPAPI for tokens, and an overlay that Windows itself keeps out of screen captures.
-
Sixteen languages
Interface, website and documentation in sixteen languages, Arabic included, right to left.
-
Every model, one subscription
Claude, GPT, Gemini, DeepSeek, Qwen, Grok and Llama behind a single gateway — with a different model assigned to each job: answers, code, vision, transcription.
May 2026
-
Whisperer starts listening
A desktop overlay that hears both sides of a call, keeps a live transcript and answers while the conversation is still going. Invisible when you share your screen, tied to no video service.
Nothing in this area yet.