A no-code platform for enterprises to build custom AI agents, starting with support teams and security analysts.

That what holds enterprise AI adoption back is not model security but the difficulty of building systems that solve real problems — so business users need no-code agents

$9M seed round led by Felicis (partner Jake Storm)

Vinay Wagh was a product director at Databricks, preparing the company to sell to large enterprises and launching Databricks SQL Serverless. Seeing ChatGPT in early 2023, his takeaway was that enterprises needed the missing pieces around the models, not the models. He met Ram Swaminathan — nine years at Bell Labs, 17 at HP Labs, then head of LinkedIn's AI trust team — through a mutual friend.

The company began in security: models detecting PII and an intent-based system to catch prompt-injection attacks, trained on collected alignment-breaking prompts. It then realized security was not the real blocker — enterprises 'don't actually quite know what they are really worried about beyond these leaks' — and widened to no-code agent building for business users.

The founders' insight came from inside the enterprise motion: Wagh had just spent years making Databricks sellable to large enterprises and saw the same gaps forming around genAI.

It started as a security company and pivoted on evidence — prompt injection and PII leaks were real, but customers' deeper problem was building useful agents at all.

Felicis making its largest-ever seed check in the company is a data point on how fast AI-application bets were outgrowing AI-infrastructure bets.

Enterprises say their blocker is security when it is actually usability: the fastest wedge into AI budgets is the person with the problem, not the person with the risk memo.

As of the March 2024 launch, Brevian was running an early release program and planned to accelerate product development and expand the team; no customer counts or revenue figures were disclosed.

FOLLOW THE EVIDENCE

The sources

  1. Brevian is a no-code enterprise platform for building AI agents techcrunch.com