# Source Onboarding

Connect a source and start syncing without writing ingestion code. Amfelt recommends a table type for each table, and flags schema drift when you run discovery again.

## Solutions · Source onboarding — Onboard a source without a project plan.

Point Amfelt at a source to discover its schemas, then have it recommend a table type for each object, which decides how it is historized. Accept the recommendations you agree with and the job is ready to run. During a run, the advisor can apply the answers it is most sure of. No connector code, no pipeline scaffolding.

Links: [Join the waitlist](/waitlist/)

## Metadata discovery

Run discovery and the source's schemas, types, and relationships are read for you, so you see what a source contains before you load it.

## Smart recommendations

LLM-powered classification recommends a table type for each object, which sets how it is historized. You accept the ones you agree with, and the advisor can apply those it is most sure of.

## Drift flagged

Run discovery again and each new column, changed type, or likely rename is graded against the contract. A rename is flagged as a candidate, never applied.

## How it works — Three steps from credentials to flowing data.

- **Connect and discover**: Point an agent at your source and run discovery. Amfelt reads its schemas, types, and relationships.
- **Review the recommendations**: Amfelt recommends a table type for each object, profiling the data first on selected sources. You accept the ones you agree with, and the advisor can apply those it is most sure of.
- **Start the sync**: The agent moves the data and checks each sync against the contract.

## See it in action — Data in motion, without pipeline code.

Movement from source to data platform, set up by reviewing recommendations instead of writing pipeline code.

Links: [All integrations](/integrations/)

## From the blog — Data tools are getting better. Data teams are still stuck.

The modern data stack keeps getting more sophisticated. Yet practitioners still spend much of their day maintaining pipelines. We are solving the wrong problems.

Links: [Read the article](/blog/data-tools-solving-wrong-problems/)

## FAQ — Common questions.

### My source is not in the catalog yet. What now?

The catalog grows demand-first: early customers set the build order. Tell us what you need and it moves up the queue.

### Does my data leave my infrastructure?

Agents run inside your network and move datasets directly from source to target, without passing them through the control plane. The control plane does receive some facts about your data: configuration, contracts, run results, and the property statistics the advisor uses, including the earliest and latest value of date and time properties. [Security & Trust](/platform/security/) lists each.

### How long does onboarding a source actually take?

Once the agent is deployed, you start discovery from the app and it runs on the agent without you watching. Recommendations follow, so most of the onboarding time goes into reviewing what Amfelt recommends.

### What happens when a source schema changes?

Run discovery again and Amfelt compares the source with its contract, grading each change as a patch, minor, or major change. A likely rename is flagged as a candidate rename, never applied as one. The app lists drift for the development environment today, and applying changes automatically under a policy you set is on the roadmap.

### What about incremental loads and change data capture?

Full loads ship today. Incremental loads, which read only the records that changed since the last run, are on the roadmap, and so is change data capture.

## Bring new sources into your platform.

Join the waitlist to hear when Amfelt opens to more teams.

Links: [Join the waitlist](/waitlist/)