Research, insights, and ideas.
Junior against Otter.ai, Fireflies and Granola on public earnings-call audio, across word, number, entity and key point error rates.
by Junior
What building Junior’s PowerPoint extension taught us about how LLMs read slides.
by Eoin Coulter
Pausing Athena was easy. The real work was letting a human enter a conversation already in progress without losing the thread.
by Shahrizod Bobojonov
Leading ASR systems scored 50–75% on entity accuracy across our benchmark of real CDD calls. We raised it to 97%.
by Lucas Descause
For any engineer at a start-up, on-call support is an accepted cost.
by Noah Mack
Ephemeral Agent Sandboxes as an Organizational Force Multiplier
by Richard Zhang
