Strands and LangGraph looked similar in one ops agent test
In an SRE and AI agent project, Strands was used first and was very easy to start with on AWS. The project later moved to because more detailed control was needed. The agent’s job was to read logs and investigate issues.
After one month of testing, the results were almost the same when the same models were used. The main difference was the surrounding and , not the model output itself.
Key points
- Strands was easy to start with on AWS.
- was chosen later for more detailed control.
- The agent looked at logs and investigated SRE and issues.
- One month of tests found almost the same results with the same models.
- For cost savings, measure calls, token use, and input size instead of judging by name alone.