@ian_dot_so: Great deep dive on @tryramp Inspect Building an in-house coding cloud agent product wasn’t an obvious bet in December 2…
Summary
The tweet highlights Ramp's internal AI coding tool, Inspect, which is praised as highly effective and impactful, outperforming tools from AI frontier labs.
View Cached Full Text
Cached at: 08/27/26, 07:35 AM
Great deep dive on @tryramp Inspect
Building an in-house coding cloud agent product wasn’t an obvious bet in December 2025.
But thank god the team cooked on it because it’s easily the best and most impactful internal tool I’ve ever used in my career
The Pragmatic Engineer (@Pragmatic_Eng): It’s rare to see a non-AI frontier lab build an in-house AI coding tool that works so much better than what frontier labs have to offer. Ramp did this with Inspect: an AI coding harness fully integrated with their systems, running on cloud machines, and being a smash hit inside
Similar Articles
@Redpoint: Why did a spend management platform start its own AI research lab? @karimatiyeh describes @RampLabs as a “collection of…
Ramp, a spend management platform, launched its own AI research lab called Ramp Labs a year ago. The lab has worked on projects like a production-focused coding benchmark 'Ramp SWE-Bench', integrating Claude Code into RollerCoaster Tycoon, and a mechanistic interpretability playground.
@GergelyOrosz: I see so many people asking: why is Ramp building their own coding agent? Spotify an agentic development environment? T…
The tweet discusses why companies like Ramp, Spotify, and Thomson Reuters are building their own AI tools instead of buying them, citing reduced costs and increased profitability due to AI advancements.
Introducing Ramp Applied AI Solutions (4 minute read)
Ramp launches Applied AI Solutions, a service that embeds AI agents into finance teams to improve operational efficiency and deliver measurable ROI, leveraging a model-agnostic approach.
Ramp SWE-Bench (3 minute read)
Ramp built a private benchmark from 80 production backend tasks to evaluate AI coding agents on review-ready patches, exposing trade-offs in accuracy, latency, and cost without public-benchmark contamination.
How Ramp engineers accelerate code review with Codex
Ramp engineers use OpenAI's Codex with GPT-5.5 to accelerate code review and build internal agentic tools, reducing review time from hours to minutes.