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

# How to increase determinism in your AI agents

> You can't make an LLM deterministic, but you can push variance out of everything around it — retrieval, control flow, and grounding — so behavior becomes repeatable.

**Determinism** is: same input, same output. LLMs are probabilistic, so you'll never
get bit-for-bit determinism from the model itself — and that's fine. The goal isn't a
deterministic *model*, it's a **repeatable system**: given the same situation, the
agent reliably does the same reasonable thing. Most of the variance you actually feel
comes from *around* the model, and that part you can control.

## Where non-determinism actually comes from

<AccordionGroup>
  <Accordion title="Retrieval variance (the big one)" icon="magnifying-glass">
    If the agent sees different context on different runs, it behaves differently —
    even at temperature 0. Fuzzy, wide, or ranked-by-vibes retrieval is the largest
    source of "why did it do something different this time?"
  </Accordion>

  <Accordion title="Control-flow variance" icon="code-branch">
    When the harness lets the model decide *everything* — including steps that should
    be fixed — you get a different path each run. Some decisions belong in code, not
    in the prompt.
  </Accordion>

  <Accordion title="Grounding variance" icon="anchor">
    Ungrounded reasoning fills gaps by guessing, and guesses vary. Grounded reasoning
    over retrieved facts converges.
  </Accordion>

  <Accordion title="Sampling variance" icon="dice">
    Temperature and sampling. Real, but usually the *smallest* contributor — and the
    one people reach for first while ignoring the larger three.
  </Accordion>
</AccordionGroup>

## Push variance out of the loop

<Steps>
  <Step title="Make retrieval stable and precise">
    Deterministic ranking (explicit [field boosts](/guides/search-configuration), not
    opaque relevance), narrow result sets, and addressed records mean the agent tends
    to see the *same* context for the same query. Stable inputs → stable behavior. →
    [Context search accuracy](/guide/context-search-accuracy)
  </Step>

  <Step title="Fix what should be fixed in the harness">
    Encode the parts of the workflow that are genuinely fixed as **control flow**, and
    let the model decide only the genuinely open parts. Determinism where you can
    afford it buys you reliability where you can't. → [The harness](/guide/the-harness)
  </Step>

  <Step title="Ground every decision in retrieved facts">
    An agent reasoning over specific, cited entities converges far more than one
    improvising from a vague prompt. Grounding is a determinism lever, not just an
    accuracy one. → [Provenance](/guide/provenance)
  </Step>

  <Step title="Verify with a critic">
    A critic sub-agent that rejects outputs unsupported by context collapses the tail
    of weird one-off behaviors into a consistent "checked" band.
  </Step>
</Steps>

## A useful reframing

> You don't make the model deterministic. You shrink the space it's allowed to be
> non-deterministic in — by stabilizing its inputs (retrieval), fixing its scaffolding
> (harness), and grounding its reasoning (context).

The more of the surrounding system is structured and repeatable, the more the model's
residual randomness stops mattering.

<Card title="Next: Self-learning loops" icon="rotate" href="/guide/self-learning-loops">
  Agents that improve with use without drifting.
</Card>