Descript meters AI usage with credits

Descript is turning AI from a bundled feature into a meter, which lets it keep entry pricing low while charging more when customers ask the product to do expensive work. A creator who mainly cuts transcripts and exports clips can stay on a normal seat plan, while a team using Underlord, dubbing, voice cloning, avatars, or generative video burns AI credits and buys top ups. That separates lightweight editing demand from GPU heavy automation demand.

• The product already splits usage into two currencies, media minutes and AI credits. Credits are consumed by specific actions like text to speech, dubbing, avatar generation, video regenerate, and lip sync, so pricing rises with the actual costliest workflows instead of forcing every customer onto a high flat subscription.
  2 descript
• This matters because Descript serves very different jobs in one app. Some users upload a Zoom recording, delete filler words, and export. Others use Underlord to reformat videos, generate new scenes, clone voices, and repurpose content at scale. One seat price would undercharge the second group or overprice the first.
  1 sacra 3 descript
• The broader market is moving the same way. AI video and voice companies such as Synthesia, HeyGen, and ElevenLabs monetize high velocity generation with consumption based pricing because every extra minute or render has real inference cost. Descript is applying that logic inside an editing workflow, not just a generation tool.
  3 descript 4 sacra

Going forward, more of Descript's expansion should come from customers shifting from manual editing into automated creation. As the product moves from transcript cleanup into full co editing and generative production, AI credits become the clearest path to higher revenue per seat without making the base editor too expensive for mainstream users.