🚨 New @a16z investment thesis: AI scribes. Imagine if you had a superhuman assistant that listened to your conversations, took perfect notes, and flagged follow-ups. And, it could make you smarter by surfacing insights or trends you missed. More from me & @illscience👇

Let's start with the history of scribes. It's an ancient profession dating back to the invention of written language. The role was critical in facilitating communication of all kinds - business, legal, religious, etc. - as well as recording history.

Today, scribes are most commonly seen in the medical, legal, and govt fields. There are ~100k medical scribes in the US, one for every 10 physicians. But many other roles have scribe responsibilities - translators, educators, journalists, researchers, salespeople, assistants.

Using a scribe is typically quite expensive, as you're hiring another human to listen and transcribe. AI makes scribes accessible + affordable to everyone - and extends the role beyond what a human could do. There's a few layers of the stack that powers AI scribes ⬇️

1️⃣ Speech-to-text This is a model that transcribes speech. Many voice scribes use an open source or API-available model for ASR, but may tune it on vocab or accents. For some use cases, it may be particular important the model is performant offline, in high noise areas, etc.

2️⃣ Structure and summarization A raw transcript of your conversations often isn't the most helpful format to consume information. LLMs can be used to summarize and structure the transcripts, and surface insights or follow-ups - what do you need to know?

3️⃣ Workflow that "actions" these outputs Even better than processing these transcripts? Taking action for you. Imagine a scribe that could push info from a sales call into a CRM. Or, that could use a doctor's conversation with a patient to fill out insurance pre-auth forms.

At @a16z, we're excited about products that use an AI scribe as a wedge into vertical software. This voice data tends to be "upstream" of everything else a business does - so a scribe can become a system of record for a person / company. A few examples and early learnings:

🩻 Freed AI (@freedmedicalai) Freed does SOAP notes for doctors. Per @ErezDruk on X, the co reached $10M ARR in a year! Clinicians save hours per week by using Freed - they no longer have to spend nights and weekends on write-ups for appointments.

🐶 Scribenote (@scribenoteinc) Like Freed for human medicine, Scribenote is an AI assistant for vets. The product is uniquely built for full and convenient voice capture of the many different appointment types vets face (ex. a tooth check 🦷).

📣 Rilla Rilla is a sales coach for in-person meetings. The scribe records, transcribes, and analyzes every appt. This allows sales managers to coach reps in a much more granular way, turning what was a "black box" into systemized evaluations. https://t.co/nmHzLF1gaL

💻 Granola (@meetgranola) Granola is a notetaker for people in back-to-back meetings (recruiters, salespeople, VCs 👀) It's full of focused features - no meeting bot, merge manual + auto notes, full transcript but fast summaries, clean sharing, etc. https://t.co/yW8WNrGA0I

🌊 Aqua (@aquavoice_) Aqua innovates on the model, with the ability to edit text via voice (ex. "put that all in bullet points") This makes it a great fit for longer form or more nuanced writing, where you may go back and forth on what you want to say. https://t.co/qpUW79aIno

If you're building an AI voice scribe, reach out! 👋 We've already made investments here and are planning to make many more in the coming months. My DMs are open, or you can email us at omoore@a16z.com + anish@a16z.com.

None of the above should be taken as investment advice or an advertisement for investment services; Some of the companies mentioned (above/below/here) are portfolio companies of a16z. A list of investments made by a16z is available at https://t.co/ZjOr5fzIIL.

yes