AI BriefingRamp Builds Laminar, an In-House Managed Agents Platform
·2026.10.21 09:00
Key point
Ramp developed Laminar to unify its agentic products on a model-agnostic, cloud-native stack, replacing fragmented vendor solutions and open-source frameworks.
Details
- Ramp built Laminar, an in-house agent harness, to replace a dozen different platforms and vendor SDKs with a cohesive, model-agnostic infrastructure for its agentic products.
- The v1 of Laminar was built in 6 weeks using coding agents, and five months later, every agentic product at Ramp runs on the platform.
- Laminar supports diverse agent archetypes, including interactive support chat, headless data processing, coding agents, and long-horizon knowledge work agents like Ramp Stack and Glass.
- The platform is built on "boring technology" (Postgres, Temporal, S3) and uses a "sandbox-as-a-tool" paradigm to decouple agent session lifetime from sandbox infrastructure.
- Key design choices include non-blocking async agents as the default, full cloud deployment on disposable workers, and a bespoke observability stack for debugging agent traces.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Laminar releases flow-1, an RL-trained model for detecting silent agent failuresWhy harnesses matter for energy (and token) efficiency