A video call where each laptop writes its own notes
Each person’s words are written down and translated on their own laptop. The call itself stays on servers.
A video meeting platform with billions of meeting hours a yearArchetype: video meetings, billions of meeting hours a year
Saves up to £344 million a year
in meeting work moved onto the devices in each call.of speech and text work moved onto participants’ devices.
At public list prices, with 9 in 10 of the devices in each meeting doing the work.
- 5jobs can run on participants’ devices
- 8jobs stay on servers
- This visit: nothing yet. Run the jobs below.This visit: nothing measured yet. Run the jobs below.
We assume 10 million people have AI notes on for 8 hours of meetings a month, each meeting counted once; that 1 in 100 of those hours is translated live and 1 in 50 gets translated captions and chat; and that each meeting hour gets one set of notes. The size starts from published usage figures for platforms like this, rounded down; the rest are our assumptions. Each job is priced at the public list price for that kind of cloud job, checked October 2026. Each laptop or phone only handles its own person’s microphone, or what its own user reads. At the lowest public prices: £125 million. With 3 in 4 devices able to do the work: £287 million.
A company that runs its own AI pays less than list price for this work. The speech job was tested in English, Spanish, French, German and Portuguese. Converting recordings for playback also stays on servers, not priced here because its size is not public. Carrying the call, recordings, scheduling, mail, sign-in, dial-in, chat delivery and search across meetings stay on servers and are not counted.
- People with AI notes on
- 10,000,000 (our assumption; published figures say millions)
- Meeting hours
- 8 a month each, each meeting counted once (our assumption)
- Live translation
- 1 in 100 hours (our assumption)
- Captions and chat
- 2 in 100 hours x 54,000 characters an hour x 1.5 languages (our assumption)
- Notes after the call
- one per meeting hour, each reading about 9,000 words and writing about 375 (our assumption)
- Device share
- 9 in 10 (our assumption)
- Live speech to text
- £0.33 an audio hour (public list price, middle of the public range)
- Speech translation
- £1.88 an audio hour (public list price)
- Text translation
- £11.32 a million characters (public list price)
- Notes model
- £1.50 a million tokens read, £7.54 a million written (mid-size model class, public list price)
- Lowest public prices
- £125 million (every line at its lowest public price, the notes at the smallest model class)
- Languages
- Speech lines count every language: the speech job was tested in English, Spanish, French, German and Portuguese.
- Prices
- Checked 15 October 2026, re-checked by 15 January 2027
- Pounds to the dollar
- 0.75491 (reference rate, 15 October 2026)
- Formula
- Units a year x public list price x device share, at 0.75491 pounds to the dollar, rounded down to 3 figures.
- This visit
- The work done in this tab x the same prices x 0.75491, only for jobs that ran on this device and matched their check. Each meeting’s audio is counted once.