> ## Documentation Index
> Fetch the complete documentation index at: https://docs.rovax.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Effort (Reasoning Level)

> How the effort parameter controls the depth of the agent's reasoning in Timely.ai, its impact on cost and latency, and when to use each level.

Effort defines how many reasoning steps the agent can use to formulate a response. In Timely.ai, it is controlled by a slider from 10 to 100 steps (5-step increments), divided into four labeled levels:

* **Fast** (10–20 steps) — simple questions, FAQs, confirmations, and greetings. Near-instant response, lower credit cost.
* **Balanced** (21–40 steps) — general support, lead qualification, scheduling. A good starting point for any new agent.
* **Detailed** (41–70 steps) — technical analyses, multi-cause diagnostics, responses requiring chained logic.
* **Maximum** (71–100 steps) — critical cases where accuracy matters more than speed or cost.

The trade-off is straightforward: more effort means more elaborate responses, higher latency, and higher credit consumption. Do not use maximum effort as a default — the correct level is the minimum that delivers sufficient quality for the use case.

<Note>
  Start with Balanced for any new agent and adjust upward only if tests show shallow or incorrect responses on questions that represent the actual complexity of the use case.
</Note>