RU

Tiro by The Playto: $3M ARR and a wiki built from meetings instead of transcripts

Published: 2026-11-15 · Author: AI Release · @ai_release1
Tiro by The Playto: $3M ARR and a wiki built from meetings instead of transcripts

⚡ The gist

🔍 What happened

Korean outlet AI Times sat down with Lim Eung-sung, founder of The Playto and creator of Tiro. Born in 1999, he has launched five startups and over 40 services since age 20, and Tiro has racked up $3M ARR in two years. The service processes 170,000 hours of audio a month and runs at more than 400 companies, some of them deployed company-wide.

The product logic goes like this: first transcription, then "contextual correction" — fixing proper names and numbers, the places where ordinary transcripts fall apart. From there, everything flows into Tiro Wiki: wiki pages with a table of contents plus the original recordings. If several meetings happen in parallel, new information gets pulled into the base in real time. From there you can search it, push it to Notion and Slack, and convert it into a corporate document template. Via MCP, the wiki plugs into Claude and Gemini — which then mop up the leftover tasks.

There's also access control. The wiki only opens up meetings an employee attended or that were shared with them. Closed meetings drop out automatically, and access is revoked the moment someone leaves the team.

🧠 Our take

Lim says it outright: STT quality has leveled off, and competing on recognition accuracy is a dead end. While Naver Clova Note, Notion AI, and Google Meet integrations fight over tenths of a percent of error rate, Tiro is moving the goalposts: it doesn't sell a transcript, it sells corporate memory. Our read: this is an honest attempt to escape a commodity market where the cheapest vendor sets the price.

What changes is what companies are actually paying for. A transcriptionist is a consumable; a knowledge base built from meetings is an asset that appreciates with every month of use. Hence the 90% renewal rate the company reports: ripping out a wiki where a year's worth of calls has been dumped is far harder than canceling a transcription subscription. That's also the biggest risk — a product like this retains customers not through quality but through accumulated content.

Who wins: large companies with heavy meeting flows and access requirements — both the permission granularity and the MCP bridges to external models appeal to them. Who loses: small players selling "recording and summary" on a single prompt. We think part of the AI notes market will collapse around two or three wiki platforms, and the rest will get absorbed into free messenger features.

⚠️ What we don't know

All the key numbers — $3M ARR, 400+ companies, 90% renewals — come from the company itself; there's no independent verification. No pricing, no average contract size, no churn in dollar terms, no breakdown of how many of the 400 customers actually pay versus sit on a pilot. The quality of "contextual correction" is also just talk: the article offers no accuracy metrics on names and numbers. And one more unknown: how the wiki handles Russian and other languages beyond Korean and English. If you're eyeing this product, I'd suggest calculating for yourself how many tokens an active wiki burns per month — Lim himself admits that overly detailed storage slows the system down and drives up costs.

💬 Worth debating

  • Should you pay a meeting service for transcription accuracy or for the accumulated knowledge base — which matters more to a team a year in?
  • Access control based on meeting attendance: solid security, or a convenient excuse when part of the context is simply invisible to an employee?
  • Tying into MCP and external models — a bet on openness, or a dependency on Anthropic and Google that will hurt the product the moment their policies change?

Forecast: within a year, Tiro will roll out a paid tier priced by knowledge volume rather than audio hours — otherwise revenue growth will hit the ceiling of how many meetings a customer can hold.

Disclaimer: The figures — $3M ARR, 400+ customers, and 90% renewal rate — come from The Playto itself in an interview with a single outlet (AI ); there is no independent verification and no data on revenue or churn.

Source: AI

🤖 AI summary
  • Date: 2026-11-15
  • Source: www.aitimes.com
  • Tags: Tiro, ИИ-встречи, MCP, база, AI, Notes, productivity, стартап
  • Key facts:
    • AI meeting service Tiro from The Playto has hit $3M ARR (roughly 4 billion won) two years after launch.
    • More than 400 companies, including large corporations and banks; per the company's own numbers, renewal rates are above 90%.
    • The bet isn't on transcription but on Tiro Wiki — a knowledge base built from meetings, hooked up to Claude and Gemini via MCP.
    • Korean outlet AI Times sat down with Lim Eung-sung, founder of The Playto and creator of Tiro.
    • Born in 1999, he has launched five startups and over 40 services since age 20, and Tiro has racked up $3M ARR in two years.
    • The service processes 170,000 hours of audio a month and runs at more than 400 companies, some of them deployed company-wide.
    • The product logic goes like this: first transcription, then "contextual correction" — fixing proper names and numbers, the places where ordinary transcripts fall apart.
    • From there, everything flows into Tiro Wiki: wiki pages with a table of contents plus the original recordings.
#Tiro#ИИ-встречи#MCP#база#AI#заметки#продуктивность#стартап
← PreviousSK Hynix expands HBM4 output: betting on Cheongju P&T6 and 20K wafers a monthNext →An H100 Killer at 1/10th the Price? Hands-On With Huawei's Atlas 300I Duo AI Accelerator (96 GB) and a Linux Setup Guide

Source: www.aitimes.com · post in Telegram