VQV Signal

AI AT WORK WATCH TECHNICAL

Ramp says its Inspect agent writes and ships about 75% of its code

According to a post on X by @aiDotEngineer, Ramp's Zach Bruggeman said at AI Engineer's Code Summit that the company's Inspect agent writes and ships about 75% of Ramp's code. The company uses one agent setup across the whole company, with shared sessions, and screenshots and browser recordings serve as evidence. The full talk is on YouTube. The post describes this as a concrete account of how far one company has pushed coding agents into production.

Source: Scobleizer News Pulse (X) · x.com Found via Robert Scoble’s Scobleizer News Pulse · a post on X, not independently verified Published 2026-11-20T20:45:00+00:00 Detected 2026-11-21T04:03:18+00:00
View original source

According to a post on X by @aiDotEngineer, Ramp's Zach Bruggeman said at AI Engineer's Code Summit that the company's Inspect agent writes and ships about 75% of Ramp's code. The company uses one agent setup across the whole company, with shared sessions, and screenshots and browser recordings serve as evidence. The full talk is on YouTube. The post describes this as a concrete account of how far one company has pushed coding agents into production.

AI-assisted summary based on the listed source.

At AI Engineer’s Code Summit, Ramp’s Zach Bruggeman said the company’s Inspect agent writes and ships about 75% of Ramp’s code, using one agent setup for the whole company with shared sessions and screenshots and browser recordings as evidence. The full talk is on YouTube. Why it matters: A concrete account of how...

The claim gives builders a company-level example of coding agents producing a large share of shipped code, rather than only experimental output. Teams weighing similar setups may want to look at how shared sessions and recorded evidence are used to review agent work. The 75% figure is a company statement relayed through a social post, so the full talk is the place to check the details.

Teams using AI at work may want to compare this against current productivity and review workflows.

Signal Strength 81% Technical label WATCH Public Interest 29 Category AI AT WORK Reader Depth TECHNICAL

Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.

Public Interest components
Recognizable Entity Score 0 Practical Impact Score 28 Novelty Interest Score 70 Consequence Score 22 Curiosity Score 16 Shareability Score 46

VQV surfaced this signal because it is recent, relevant to AI Coding Tools, connected to Scobleizer News Pulse (X).