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

> Need accurate answers for candidates about your company? Knowledge Base — what it is, where to set it, and how the AI uses it.

# Knowledge Base

It's for **facts and context you want every AI operation to get right** — from how interviews are evaluated to candidate experience preferences and workspace-wide defaults.

### Why it's important (for you and the candidate)

* **Trust**: Confident, consistent answers feel professional; vague or wrong ones hurt your brand.
* **Your time**: Less back-and-forth later when candidates already heard the right story in-screening.
* **Fairness**: Everyone hears the **same** policy facts, not whatever the model improvises.
* **Speed**: Recruiters aren't the bottleneck for the tenth "Is this role hybrid?" of the week.

### Where to set it up

**Interviewers → Tools → Knowledge Base** in the sidebar.

<Frame caption="Knowledge Base list with Filter by scope dropdown showing all available scope options.">
  <img src="https://mintcdn.com/heymilo/Mi5otd3F7xTrauQ6/images/kb-scope-filter.png?fit=max&auto=format&n=Mi5otd3F7xTrauQ6&q=85&s=25ea8d3041b59e292e3dc701164656f2" alt="Knowledge Base list with scope filter dropdown open" className="mx-auto" width="1024" height="432" data-path="images/kb-scope-filter.png" />
</Frame>

### What to add

Create **small, focused** items — one topic per entry when you can.

<Frame caption="Add knowledge form showing Title, Scope, Content, and Active toggle.">
  <img src="https://mintcdn.com/heymilo/738-drGLLDm40WkE/images/kb-add-knowledge-form.png?fit=max&auto=format&n=738-drGLLDm40WkE&q=85&s=a1d32c80402727982caa89069583bc26" alt="Add knowledge form" className="mx-auto" style={{ width:"65%" }} width="800" height="1024" data-path="images/kb-add-knowledge-form.png" />
</Frame>

Each item has:

* **Title**: What you would scan for in a list (e.g. "Remote and hybrid policy," "Overall scoring calibration").
* **Scope**: Controls where this knowledge is applied. Choose the scope that matches what you want the AI to do with it.

| Scope                     | What it does                                                                                            |
| ------------------------- | ------------------------------------------------------------------------------------------------------- |
| **Workspace Context**     | Global context applied across all AI operations: evaluations, interviews, and agent creation.           |
| **Candidate Persona**     | Shapes how the AI approaches and interacts with candidates during the interview.                        |
| **Interviewer Knowledge** | Context specific to how the interviewer should behave or respond.                                       |
| **Interview Evaluation**  | Applies to how the AI scores and evaluates candidate answers. Use this for scoring calibration prompts. |

* **Content**: Plain language, as if you're saying it to a candidate on the phone. If legal or comms approved specific wording, use that. The more precise, the better the AI can apply it.
* **Active**: On = AI can use it; off = hidden without deleting (great for seasonal or paused entries).

**Start here:** The five questions your team answers on every screen. Turn those into your first five items.

### After you've added items

* **Edit** when anything changes: stale answers are worse than missing ones.
* **Deactivate** what you don't need right now; **delete** only what you'll never use again.
* **Filter by scope** to find entries quickly as your list grows.

### How the AI uses this

During **voice/video** (and similar flows), when a candidate asks something that matches what you've written, the AI **pulls from these articles** so answers stay **grounded in your text**. Entries with **Interview Evaluation** scope apply to how the AI scores answers across your workspace.

**Reminder:** Put **accurate, approved** information here. The Knowledge Base is not a substitute for **policy or legal sign-off** on sensitive topics.

### Next steps

* [Adjusting Scoring (Interview Evaluation)](scoring-calibration-guide) — Use Knowledge Base prompts to tune scoring across your workspace
* [Configuring Your AI Interviewer](../../getting-started/log-in/configuring-your-ai-interviewer) — How interviews and context fit together
* [What Candidates Experience](../../getting-started/log-in/what-candidates-experience) — What they see and hear
* [AI Settings](.) — Other workspace AI defaults