Improving Agents is a Data Mining Problem — Vivek Trivedy, LangChain
Source
AI Engineer
Author
AI Engineer
Published
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
context compaction — Summarizing an agent's earlier conversation to free room in the context window so a long session can keep going.
fine-tuning — Taking a trained model and training it a bit more on your own examples so it gets better at one specific job.
Why it matters
It gives a decision rule for the prompt-tuning versus fine-tuningTaking a trained model and training it a bit more on your own examples so it gets better at one specific job.Full definition → question and shows traces are the substrate that makes AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → improvement measurable rather than anecdotal.
