> ## 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.
> Sally is HeyMilo’s autonomous sourcing experience: she searches connected data sources, compares people to your role, scores fit, and surfaces candidate recommendations based on rules you set.
# Sally & data sources (Databricks)
Today: ATS-powered Sally sourcing with Databricks is Avionté only. HeyMilo workspace and external sources may still apply per your workspace setup.
### What Sally does
Sally helps you move from “we need more qualified people in this req” to **ranked candidate recommendations** without manual tab-hopping.
She can:
* Search across the **Databricks you enable** for that sourcing setup
* Compare candidates to the **posting / agent** you configure
* Score **match quality** against your rules
* Respect **deal breakers** (hard requirements that must pass)
* Operate in modes such as **Automatic**, **Manual**, or **Threshold** (exact labels depend on your UI), within your **quota** and **time limit**
Think of Sally as: **one place to turn sourcing rules into a steady stream of candidate recommendations**—with caps and guardrails you control.
### What a Databricks is (Data sources)
A **Databricks** is a **searchable dataset** HeyMilo builds and keeps updated for Sally. It is a **copy** of relevant data from a source (HeyMilo, ATS, etc.), organized so Sally can search and rank candidates quickly and consistently.
**In practice for recruiters:**
* **HeyMilo Databricks**: Built from activity in your workspace (for example interview volume affects how long **provisioning** takes).
* **ATS Databricks**: Built from your connected ATS data (for example scope may depend on how many candidates are in scope for your integration). **Availability of ATS sources depends on which ATS HeyMilo supports for your workspace.**
* **External Databricks**: Optional **external talent data** from approved partner sources. When enabled, it may be available **immediately** or follow a separate enablement flow depending on product setup.
**Provisioning:** When a Databricks is **still building**, the UI may show a **not ready** state (for example a **yellow** indicator). You **cannot select** that source for Sally until provisioning finishes. This is normal for large workspaces or large ATS scopes.
### Create a Sally sourcing config
**Where:** **Source Candidates → Sourcing Engine (ATS)** → **Sourcing Configs** → **Create Config** (or equivalent).
**Fields you will commonly see:**
| Concept | What it means for recruiters |
| --------------------- | -------------------------------------------------------------------------------------------------------------------------------------------- |
| **Config name** | A label your team recognizes (for example “Default sourcing — RN roles”). |
| **Quota** | **Maximum number of candidates** Sally will source for this config (a cap on volume). |
| **Databricks** | **Which data sources** Sally is allowed to search. Select **one or more** when ready. **Do not select** sources that are still provisioning. |
| **Ingestion mode** | **How** recommended candidates enter the pipeline: for example **Automatic** vs **Manual** vs **Threshold** (definitions appear in-product). |
| **Threshold** | A **match bar** Sally uses when that mode applies (for example only surface people above a certain match %). |
| **Time limit (days)** | How long **this sourcing run / window** stays active under the config. |
| **Set as default** | When checked, this config can be preselected for new postings that use Sally (if your workspace supports defaults). |
**Deal breakers (second tab / section):**
**Deal breakers** are **hard constraints**. If you set them, candidates who fail them should not be treated as viable matches (for example minimum education, authorization, or other must-have fields your team configures).
**Important:** You typically need **at least one Databricks selected** before you can configure deal breakers (the UI may block the Deal breakers step until a data source is chosen).
### Use Sally on a posting
**Typical flow:**
1. Create or edit a **posting** (or the flow your workspace uses to attach sourcing to a role).
2. Select a **sourcing config** you already created. Save your agent.
3. Open the **Sally** view for that posting.
4. Expect a short wait the first time results populate (for example **one to two minutes** in some cases).
5. Review **candidate recommendations**, open profiles, and proceed with your normal workflow (invite, shortlist, etc.).
If something looks empty, confirm **Databricks** are **provisioned**, the **sourcing config** is attached, and you are still inside the **time limit** for that configuration.
### Advanced settings
Controls how Sally searches the candidate database. Defaults work for most teams.
| Setting | What it does | Default |
| ------------------------- | ----------------------------------------------------------------------------------- | ------- |
| **Max Search Iterations** | Search passes per data source — higher is more thorough but slower. | 3 |
| **Min Match Quality** | Minimum similarity score to accept a candidate (0.0–1.0) — lower casts a wider net. | 0.30 |
### Quick troubleshooting
| Symptom | What to check |
| -------------------------- | ---------------------------------------------------------------------------------------------------- |
| Cannot select a Databricks | It may still be **provisioning**. Wait until the UI shows **ready** (not yellow / blocked). |
| Deal breakers disabled | Select **at least one** Databricks first. |
| No candidates in Sally | Config not attached to posting, **quota** reached, **time limit** expired, or thresholds too strict. |
| “Create a databrick first” | Complete **Databricks** enablement / provisioning before **Create Config**. |
### Related pages
* [**Sourcing**](/deep-dive/sourcing) (overview: Interviewed Candidates, Sourcing Engine (ATS), Apply Email)
* [**Integrations**](../../integrations/ats-quickstarts) (ATS connection prerequisites)
* [**Creating an interviewer**](../../getting-started/log-in/creating-an-interviewer) **/ postings** (where sourcing configs attach)