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

# How DataVox Works

> From connecting your data systems to getting answers in plain English. Learn how DataVox processes questions, queries your databases, and returns actionable insights.

DataVox is designed to go from setup to real answers quickly. Here's what the process looks like.

## The flow

<Steps>
  <Step title="We connect to your existing systems">
    DataVox plugs into the systems your organization already uses — your CRM, ERP, finance tools, support platforms, spreadsheets, and more.

    No data needs to be moved or copied. DataVox reads from your systems where they are.

    A typical starting point is 2–3 systems. You add more over time.
  </Step>

  <Step title="DataVox learns your business">
    Before anyone starts asking questions, DataVox is trained on your specific data. It learns:

    * What your teams mean by terms like "active customer," "qualified lead," or "net revenue"
    * How your systems relate to each other (e.g., which CRM accounts map to which ERP records)
    * Your business rules and how you calculate key metrics

    After training, DataVox understands your data the way your best analyst does. This is why it gives accurate answers instead of generic guesses.
  </Step>

  <Step title="Your team asks questions in plain English">
    Anyone on your team types a question the way they'd ask a colleague:

    * "What were our top products in the Southeast last quarter?"
    * "Show me customers with overdue invoices who haven't been contacted in 30 days."
    * "How does this quarter's pipeline compare to last year?"

    DataVox figures out which systems have the relevant data, pulls from all of them at once, and returns a clear answer — typically in under 10 seconds.
  </Step>

  <Step title="You get answers and make decisions">
    DataVox returns results that are ready to act on:

    * **Clear data** — tables, numbers, and breakdowns from your actual systems
    * **Visual charts** — graphs and charts where they help tell the story
    * **Source transparency** — which systems the data came from
    * **Forward-looking signals** — forecasts and risk indicators when relevant

    No waiting for an analyst to clean up the output. The answer is ready when you are.
  </Step>
</Steps>

## Data security

Everything runs on your own servers:

| What                   | Where it runs                             |
| ---------------------- | ----------------------------------------- |
| **AI models**          | On your infrastructure — not in the cloud |
| **Trained knowledge**  | Stored on your servers                    |
| **Data processing**    | Within your network                       |
| **System connections** | Through your existing security controls   |

No data leaves your environment. No external services process your information. Your security team reviews the architecture during the proof of concept.

## How you get started

DataVox starts with a focused proof of concept so you can see results before making any commitment:

```
Proof of concept (about 2 weeks)
    Connect 2–3 of your data sources
    Train DataVox on your data
    Test it with questions your team already knows the answers to
    ↓  you see it works on your data
Full rollout (1–2 months)
    Connect more systems
    Train on all your connected data
    Roll out to business users across teams
    ↓  value grows as more people use it
Expand over time
    Add more systems and users as needed
    New connections take days, not months
```

The proof of concept is designed so you see real, accurate results on your own data before you commit to anything.

## Next steps

<Columns cols={2}>
  <Card title="Key Concepts" icon="book" href="/datavox/introduction/key-concepts">
    Understand the building blocks.
  </Card>

  <Card title="What is DataVox?" icon="lightbulb" href="/datavox/introduction/what-is-datavox">
    Back to the overview.
  </Card>
</Columns>