The data collection agent: describe the sync, get the pipeline
Tell Bridge what to move — source, target, tables, filters and cadence — in a sentence. The data collection agent reads the request, profiles the source, drafts the job configuration and recommends the sync strategy, so a pipeline that used to take days to configure is ready to review in minutes.
The data collection agent turns a plain-language request into a review-ready Bridge pipeline: it drafts the source and target, maps fields, proposes filters and a schedule, and recommends whether full load, log-based CDC or batch fits your data.
What it does
Natural-language sync drafting
Describe the movement you need and the agent parses the intent into a concrete configuration: source and target types, table-to-table mapping, field filters, partition strategy and scheduling parameters.
Source profiling
The agent reads the structure and characteristics of the source tables — column types, keys and distributions — so the draft is grounded in the real schema, not a guess.
Sync-strategy recommendation
Based on data volume, latency requirement and source type, it recommends the best fit: a one-off full load, log-based CDC, or a scheduled batch — and explains why.
Field mapping & filters
Column mapping and filter conditions are drafted for you to confirm, instead of being hand-built line by line. Edits happen in the visual canvas.
Schedule drafting
A schedule is proposed to match your freshness need — continuous CDC, or an interval batch — with the parameters ready to adjust.
How it works
- Describe the need in plain language. For example: "Sync the orders and order_items tables from the production Oracle database to Snowflake every five minutes, keeping only the last 90 days."
- The agent profiles the source. It reads structure and characteristics of the tables you named, so the draft reflects the real schema.
- The agent drafts the job. Source and target, field mapping, filters and a schedule are produced as a single review-ready configuration.
- You review and confirm. The draft opens in the visual low-code canvas — or exports as a template for bulk reuse — and you approve before anything runs.
- It runs as a normal Bridge pipeline. Log-based capture where supported, in-flight transforms, and a load into the target, exactly as a hand-built job would behave.
The agent drafts; you decide. Nothing is created in a target until you confirm, which keeps the human in control of every pipeline.
Why it matters
Most integration effort is not the runtime — it is the configuration: choosing sources and targets, mapping fields, setting filters and schedules, and repeating it across hundreds of tables. The data collection agent compresses that front-loaded work.
- Lower barrier. A new engineer can stand up a pipeline from a sentence, not a training course.
- Consistency. Drafts follow the same mapping and naming rules as your reviewed templates, so the estate stays uniform.
- Speed. Configuration that used to take days drops to minutes of review, and the draft can be promoted to a template for the rest of the estate.
- Control preserved. The agent drafts; the visual canvas and your approval gate remain the place where a pipeline actually goes live.
Questions about the data collection agent
Is the data collection agent a separate product?
No. It is a capability of Bridge that drafts sync jobs for you to review and run inside your own environment. There is no separate licence and no data sent to an external service to produce the draft.
Does it replace the visual canvas?
No. It produces a draft; you confirm it in the visual low-code canvas (or export it as a template). The canvas and the bulk-creation templates remain the place where pipelines are reviewed and promoted.
What data does the agent see?
It operates on the metadata and structure of the tables you point it at, inside your deployment. Rows stay in your environment — the agent reads schema, not your data, to build the draft.
Which connectors does it support?
Any of the 100+ connectors Bridge supports today, across databases, warehouses, data lakes, files, messaging, time-series and IoT. The recommendation adapts to the source type.
How is this different from bulk pipeline creation?
Bulk creation generates many pipelines from a selected set of tables and a reviewed template. The data collection agent is the front door: it turns a plain-language description into that first draft, which you can then promote to a template and roll out in bulk.
Try the draft, keep the control
Download the build, describe a sync, and confirm the pipeline in the visual canvas before it runs.