Has anyone else noticed that sometimes an Agent runs really fast and other times it’s noticeably slower?
For me, it’s usually the same prompt and the same task. Sometimes I get a response in a few seconds, other times it can take 10–15 seconds. Nothing major, just enough to notice and wonder what’s going on.
I was curious whether this is expected behaviour, or if there are any best practices to help make Agents run more consistently fast. Things like how the task is set up, model choice, or anything else worth tweaking.
Would be good to hear if others have seen the same and what’s worked for you.
A few fairly normal things can cause this, even if the prompt looks the same each time.
Sometimes it’s just the model provider. These models are shared infrastructure, so if a lot of people are using the same one at that moment, replies can take a bit longer.
Other times the Agent ends up doing slightly more work under the hood. Maybe it’s pulling in more context, checking memory, or making a tool call. From the outside it looks identical, but the work isn’t always the same.
If you’ve got multiple Agents or sessions running at once, that can also add a bit of overhead.
If it ever feels consistently slow at a specific moment, you can ask the Agent why it took longer and it may surface something useful. Or feel free to DM me or drop the thread URL to engineering@joinbrim.ai when it happens and we can check what was going on in the background.
Happy to walk through it in more detail if helpful.