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

# AI guides

> Straight answers to the questions teams ask before they trust AI with real work: made-up numbers, audit trails, rollback, data safety, and who signs off.

Each guide answers the question first, then shows the checks to run and how common AI tools compare. Comparisons were checked against vendor documentation in October 2026.

<CardGroup cols={2}>
  <Card title="AI made up numbers" icon="calculator" href="/docs/ai-guides/ai-made-up-numbers-financial-model">
    Catch fabricated figures in a financial model before it goes out.
  </Card>

  <Card title="An agent deleted production data" icon="database" href="/docs/ai-guides/ai-agent-deleted-production-database">
    What to do in the first hour, and the controls that prevent a repeat.
  </Card>

  <Card title="Audit trail and rollback" icon="clock-rotate-left" href="/docs/ai-guides/ai-spreadsheet-tools-audit-trail-rollback">
    Which AI spreadsheet tools record every change and let you undo one.
  </Card>

  <Card title="AI vs human auditability" icon="magnifying-glass" href="/docs/ai-guides/ai-vs-human-auditability">
    Which work is easier to audit, and what evidence each leaves.
  </Card>

  <Card title="Humans and AI editing live" icon="users" href="/docs/ai-guides/humans-and-ai-editing-live">
    Tools that let people and agents edit the same file at once.
  </Card>

  <Card title="Human in the loop for finance" icon="user-check" href="/docs/ai-guides/human-in-the-loop-ai-finance">
    Where the reviewer belongs, and how much autonomy to grant.
  </Card>

  <Card title="AI won't use my template" icon="file-powerpoint" href="/docs/ai-guides/ai-wont-use-my-template">
    Why AI rebuilds your deck, and how to make it edit in place.
  </Card>

  <Card title="Will AI replace accountants?" icon="briefcase" href="/docs/ai-guides/will-ai-replace-accountants">
    Which tasks AI takes on, and what stays with the accountant.
  </Card>

  <Card title="Shadow AI without a ban" icon="shield-halved" href="/docs/ai-guides/shadow-ai-governance">
    Give people a sanctioned tool, then govern it: policy, permissions, audit.
  </Card>

  <Card title="Prompt injection and agents" icon="syringe" href="/docs/ai-guides/prompt-injection-ai-agents">
    How an email or file can steer an agent, and what limits the damage.
  </Card>

  <Card title="Is company data safe in AI?" icon="lock" href="/docs/ai-guides/is-company-data-safe-with-ai">
    Training, retention, who can see it, and where the model runs.
  </Card>

  <Card title="AI agent costs and ROI" icon="coins" href="/docs/ai-guides/ai-agent-cost-control-roi">
    Seat and token pricing, runaway runs, budgets, and measuring return.
  </Card>

  <Card title="Explaining what an agent did" icon="list-check" href="/docs/ai-guides/explainable-ai-agents">
    The instruction, sources, steps and approval behind each output.
  </Card>

  <Card title="Is the scheduled job still right?" icon="calendar-check" href="/docs/ai-guides/monitoring-scheduled-ai-work">
    How recurring AI work drifts, and the checks that catch it.
  </Card>

  <Card title="Switching AI models or vendors" icon="right-left" href="/docs/ai-guides/ai-vendor-lock-in">
    Where lock-in hides, and how to keep your work portable.
  </Card>
</CardGroup>

For connecting MCP servers to Athena, see [MCP servers](/docs/developer-integrations/mcp).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.