Building For Agents
A practical guide for engineers designing APIs, MCP servers, observability systems, or delegated services for AI agents.
K
Kirti Rathore
13 min read
Posts tagged "observability"
6 articles found
Fixing bugs reduces to having their reproduction test case.
A root cause is only useful once it can be reproduced. The practical debugging loop is to localize the fault, turn the alert into a reproduction, and test hypotheses until the fix is validated.
K
Kirti Rathore
4 min read
What if every bug report came with root cause attached?
Most alert queues are full of duplicates, expired assumptions, and evidence-free mysteries. The useful AI workflow is not just RCA. It is hypothesis validation all the way to a proven fix.
K
Kirti Rathore
5 min read
Oncall isn't supposed to be this hard
Well-configured alerts connect the symptom to traces, logs, deploys, and the suspect commit.
K
Kirti Rathore
6 min read
FixBugs is not a GitHub bot
Most people who land on fixbugs.ai still think we fix GitHub issues. That framing is wrong, and it's our fault. This is where debugging actually happens — the IDE, the issue tracker, the observability pipe — and what a debugging agent has to see to be useful.
K
Kirti Rathore
7 min read
How To Build High Performance AI Agents - Insights from PyDelhi Talk
Key insights and strategies from my PyDelhi talk on building high-performance LLM agents that can process gigabytes of data and deliver results in under a minute, with lessons learned from optimizing FixBugs.
K
Kirti Rathore
2 min read