AI Agent
What it is
An AI agent is a system built on a language model that can break a goal into steps, call tools such as search, code readers, or APIs, observe the results, and decide what to do next. A chatbotChatbotA conversational interface that answers customer questions, usually from fixed docs. replies to a message; an agent works a problem until it's done or it needs a human.
Why it matters
Most technical tickets can't be answered from a single document. Someone has to check logsLogsTimestamped records of events a system emits as it runs., read the code path, and compare account config. Agents can do that legwork at machine speed, which is where support time actually goes.
Best practices
- Scope the tools – Give the agent read access to the systems a human investigator would open, and nothing it doesn't need.
- Require citations – Every claim should link to the log line, commit, or doc it came from so a reviewer can check it in seconds.
- Keep a human gate on actions – Let the agent investigate freely, but route anything that changes code or customer data through approval.
How Decimal helps
Decimal's agents investigate customer issues across your code, logs, tickets, and docs, then answer the customer, hand engineering a cited root cause, or open a fix for review.