About IvreetMeet

IvreetMeet (Hebrew: עברית מיט, pronounced "ivreet meet") is an Israeli software service that turns Hebrew recordings — meetings, calls, lectures, interviews — into a transcript split by speaker, a Hebrew summary with decisions and action items, and documents you can export or share. It runs its own speech-recognition model trained on spoken Hebrew, works in the browser with no installation, offers a free plan, and exposes a REST API and an MCP server so AI assistants such as Claude and ChatGPT can search your meetings. It was founded in 2025 in Jerusalem by Maor Yehezkel and operates under the trade name TetraZPE.

Facts

Why it exists

Spoken Hebrew is a hard case for general speech models: no vowels in writing, prefixes glued to words, several accepted spellings, slang, and English terms in the middle of most sentences. General tools handle clean, read Hebrew and fall apart in a real meeting. IvreetMeet trains and runs its own model on spoken Hebrew, builds speaker separation into the pipeline rather than bolting it on, writes the summary in Hebrew, and keeps recordings in Israel. The product is the whole path from a recording to a document a team can act on.

What it is not

• Not a bot that joins calls. You record in Zoom, Google Meet, Teams or on a phone and upload the file.
• Not a general-language tool. It is tuned for Hebrew, with English mixed in.
• Not a scoring system for people. Speakers get names; nobody gets categorised, and no emotional state is inferred about a person.

Milestones

• 2025 — founded in Jerusalem; first version of the service and the site.
• September 2026 — the in-house Hebrew speech model replaces the previous one in production; speaker consent and deletion rules; the export studio; chat with your meetings; the REST API and the MCP server with connectors for Claude, ChatGPT, Cursor, VS Code, Codex CLI and Gemini CLI; the English site; recordings moved to storage in Israel; published terms, privacy and refund policies.