# Ramp Labs Research

> Long-form engineering and research writeups from the Ramp Labs team.

Technical blog covering agentic systems, interpretability, post-training, and production ML.

## Articles

- [PorTAL: Portable Task Adaptation for LoRA](https://labs.ramp.com/research/portal-portable-task-adaptation/)
  - Tag: Research
  - Date: 2026-07-01
  - Summary: Learn a task adaptation once in a base-agnostic form, then port it to new frozen models by refitting only a thin per-base alignment — recovering ~98% of per-task LoRA's lift on an unseen model within the same family and ~94% across families.
  - Markdown: https://labs.ramp.com/research/portal-portable-task-adaptation/index.md
  - JSON: https://labs.ramp.com/research/portal-portable-task-adaptation/article.json

- [Building Fast & Accurate Agents with Prime-RL Post Training](https://labs.ramp.com/research/prime-rl-post-training/)
  - Tag: Research
  - Date: 2026-05-07
  - Summary: How Prime-RL post training improves agent speed and accuracy for production workflows.
  - Markdown: https://labs.ramp.com/research/prime-rl-post-training/index.md
  - JSON: https://labs.ramp.com/research/prime-rl-post-training/article.json

- [Coding agents ignore their own budgets](https://labs.ramp.com/research/coding-agents-ignore-spend/)
  - Tag: Research
  - Date: 2026-04-21
  - Summary: Agents can't be trusted to manage their own token budgets. Spend control has to live in a separate, evidence-grounded system outside the agent doing the spending.
  - Markdown: https://labs.ramp.com/research/coding-agents-ignore-spend/index.md
  - JSON: https://labs.ramp.com/research/coding-agents-ignore-spend/article.json

- [Latent Briefing: Efficient Memory Sharing for Multi-Agent Systems via KV Cache Compaction](https://labs.ramp.com/research/latent-briefing-kv-cache/)
  - Tag: Research
  - Date: 2026-04-10
  - Summary: A Ramp Labs writeup on using KV cache compaction to share memory efficiently across multi-agent systems.
  - Markdown: https://labs.ramp.com/research/latent-briefing-kv-cache/index.md
  - JSON: https://labs.ramp.com/research/latent-briefing-kv-cache/article.json

- [How we built Steer, our interpretability playground](https://labs.ramp.com/research/how-we-built-steer/)
  - Tag: Interpretability
  - Date: 2026-04-02
  - Summary: A deep dive into building Steer - an interactive tool for exploring and understanding how language models process information internally.
  - Markdown: https://labs.ramp.com/research/how-we-built-steer/index.md
  - JSON: https://labs.ramp.com/research/how-we-built-steer/article.json

- [How we made Ramp Sheets self-maintaining](https://labs.ramp.com/research/ramp-sheets-self-maintaining/)
  - Tag: Engineering
  - Date: 2026-03-23
  - Summary: How we built a system that lets Ramp Sheets automatically detect and fix its own issues - reducing manual maintenance and improving reliability.
  - Markdown: https://labs.ramp.com/research/ramp-sheets-self-maintaining/index.md
  - JSON: https://labs.ramp.com/research/ramp-sheets-self-maintaining/article.json

- [Post Training Ensemble vs. Singular Model Approaches with Tinker](https://labs.ramp.com/research/post-training-ensemble-tinker/)
  - Tag: Research
  - Date: 2025-11-03
  - Summary: Comparing ensemble and singular model strategies for post-training optimization, and what we learned building Tinker to explore these tradeoffs.
  - Markdown: https://labs.ramp.com/research/post-training-ensemble-tinker/index.md
  - JSON: https://labs.ramp.com/research/post-training-ensemble-tinker/article.json

- [We built an agent to prompt our internal finance agent](https://labs.ramp.com/research/agent-to-prompt-finance-agent/)
  - Tag: Agents
  - Date: 2025-10-02
  - Summary: What happens when you build an AI agent whose job is to figure out how to prompt another AI agent? Lessons from recursive agent architectures.
  - Markdown: https://labs.ramp.com/research/agent-to-prompt-finance-agent/index.md
  - JSON: https://labs.ramp.com/research/agent-to-prompt-finance-agent/article.json

- [How we built Agent Fill](https://labs.ramp.com/research/how-we-built-agent-fill/)
  - Tag: Agents
  - Date: 2025-08-27
  - Summary: The story behind Agent Fill - an AI agent that automatically fills out forms by understanding context, extracting data, and navigating complex workflows.
  - Markdown: https://labs.ramp.com/research/how-we-built-agent-fill/index.md
  - JSON: https://labs.ramp.com/research/how-we-built-agent-fill/article.json

## Optional

- [Full manifest (https://labs.ramp.com/research/manifest.json)](https://labs.ramp.com/research/manifest.json)
- [Sitemap (https://labs.ramp.com/research/sitemap.xml)](https://labs.ramp.com/research/sitemap.xml)