One agent or many?
Read · 1 of 2
More agents, more tokens
Every agent re-reads its own prompt and tools, so a team of five can easily use several times the tokens of one agent doing the same job.
Teams pay off when the work splits into independent pieces that can run in parallel (researching ten companies), or when the pieces need very different tools. When each step depends on the last, one agent with good context usually wins.
Read this lesson as text
Multi-agent systems cost more. Use them when the work truly splits.
More agents, more tokens
Every agent re-reads its own prompt and tools, so a team of five can easily use several times the tokens of one agent doing the same job.
Teams pay off when the work splits into independent pieces that can run in parallel (researching ten companies), or when the pieces need very different tools. When each step depends on the last, one agent with good context usually wins.
Team or solo?
Task A: Compare 12 laptops: find specs, prices and reviews for each. Task B: Fix a bug: read the error, find the file, change one line, run the tests.
Where does a team of agents help?
- A: twelve independent lookups can run in parallel
- B: each step needs its own agent
- Neither
Show the answer
A: twelve independent lookups can run in parallel. A splits cleanly: each laptop is a separate job, and workers can do them at the same time. B is one chain of dependent steps; splitting it would just pass context back and forth.