“We started looking at this from a two-part problem of how do you use traditional computer vision to preserve structure, and how do you use VLMs as like a correction loop”
The a16z Show · The Missing Infrastructure for AI Agents: Documents | Reducto’s Adit Abraham
Play from 5:50
5:50
Audio streams from the publisher’s own feed. Verbatim does not host or copy it.
What was described
pivoted the company focus to document processing after YC and built a unified model (R1) that handles parsing, structure preservation, and VLM correction end-to-end
Said nearby on the same subject
“We can train an AI model to do what we know you need. Because, again, a lot of times this is coming after we've been engaged as the contact center business process outsourcing company.”
“We built them a platform… we built this really cool email agent because they're getting a lot of their orders through email… it will read the email, parse, add to ordering system, send invoice, update them.”
♫
Said on SaaS That App - Building Tech-Enabled Businesses