← All notes

Answer: With agents as tools, a manager agent keeps control and calls specialists; with handoffs, a triage agent routes the conversation and the specialist becomes the active agent for the rest of the turn.

Orchestration choices

Agents can be orchestrated by letting the LLM decide the steps or by determining the flow in code; the two can be mixed.

Agents as tools

A manager agent calls specialist agents through Agent.as_tool(). Pick it when one agent should own the final answer, combine outputs from multiple specialists, or enforce shared guardrails in one place.

Handoffs

• Handoffs are represented as tools to the LLM: a handoff to Refund Agent becomes a tool named transfer_to_refund_agent.
• handoff_description hints when the model should pick that handoff.
• on_handoff is a callback executed when the handoff is invoked.
• The receiving agent sees the conversation history unless you change it with an input_filter.

See also

• ao-handoff
• ao-handoff-orchestrator-agents
• ao-handoff-role-based-multi-agent

Claims

• id: C1 claim: Agents as tools keep a manager in control. source: S1 evidence: "A manager agent keeps control of the conversation and calls specialist agents through"

• id: C2 claim: With handoffs the specialist becomes the active agent. source: S1 evidence: "that specialist becomes the active agent for the rest of the turn."

• id: C3 claim: Handoffs are tools to the LLM. source: S2 evidence: "Handoffs are represented as tools to the LLM."

• id: C4 claim: input_filter filters what the next agent receives. source: S2 evidence: "This lets you filter the input received by the next agent."

Sources

• [S1] Agent orchestration. OpenAI Agents SDK (Python) docs (GitHub). https://github.com/openai/openai-agents-python/blob/main/docs/multi_agent.md . License: MIT (openai/openai-agents-python LICENSE). Accessed 2026-11-09.

• [S2] Handoffs. OpenAI Agents SDK (Python) docs (GitHub). https://github.com/openai/openai-agents-python/blob/main/docs/handoffs.md . License: MIT (openai/openai-agents-python LICENSE). Accessed 2026-11-09.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Summary by iinventedlight (AI-assisted). Sources: GitHub repository docs under Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0) or MIT License (https://opensource.org/licenses/MIT), as listed per source; copyright notices remain with the original authors. Text was condensed and changed from the originals.

@misc{iil_note_6gyskh8j,
  author = {{iinventedlight}},
  title = {{Agents as tools vs handoffs: two multi-agent patterns}},
  howpublished = {iinventedlight Library},
  year = {2026},
  date = {2026-11-09},
  url = {https://iinventedlight.com/library/notes/ao-handoff-agents-as-tools-vs-handoffs/},
  note = {Permanent ID iil:note:6gyskh8j. License: CC-BY-4.0.},
}

Markdown

Spotted a mistake? Use the MCP feedback tool or email hello@iinventedlight.com.

For language models and agents: start at /llms.txt | /agents.md | /.well-known/agent-library.json. This page in Markdown: /library/notes/ao-handoff-agents-as-tools-vs-handoffs.md.