Application · Built on BergFlow and BergAuto

Crawl evidence

Crawled data, with the evidence attached.

BergCrawler captures crawl evidence for every record: the screenshot, the timestamp and the source, so a reviewer can see exactly what a page said when it was collected.

BergCrawlerCrawl · Capture · Record · Review
  1. BergAutoExecutes the crawlAgreed sources and scheduleBergAuto
  2. BergCrawlerCaptures the evidenceScreenshot, time, source
  3. BergFlowRecords and versionsLinked to the evidenceBergFlow
  4. ReviewChanges checkedEvidence attached
Evidence E-5521Captured
Captured page thumbnailA wireframe of a pricing page with three plan columns, the middle plan highlighted, as captured by BergCrawler.
Source
example.com/pricing
Captured
14 Oct 2026, 09:42 UTC
Locale
en-IN
Record
R-3310, v2
Illustrative evidence card. example.com stands in for a real source; captured fields are the defaults on every record.
  • Capture

    Screenshot, timestamp, source and locale.

  • Preserve

    Evidence linked to the record it supports.

  • Review

    Changes checked against what the page showed.

  • 3Evidence fields on every capture: screenshot, timestamp, source
  • 1Governed record per capture in BergFlow
  • EveryChange versioned, never overwritten

Design figures: what every BergCrawler capture carries by default. Locale is captured where a source serves more than one.

How it runs

Where the crawl ends and the evidence begins.

BergAuto executes the crawl. BergCrawler captures and preserves the evidence. BergFlow records the result, routes the review, versions the outcome and keeps the audit trail.

  1. 01Crawl

    BergAuto runs the crawl on an agreed list of sources and an agreed schedule.

  2. 02Capture

    BergCrawler stores the screenshot, the timestamp, the source URL and the locale with each capture.

  3. 03Record

    BergFlow versions the extracted result and links it to the evidence that supports it.

  4. 04Review

    Changes and anomalies go to a reviewer with the evidence attached, and the decision is logged.

Capabilities

What BergCrawler keeps on every capture.

Six capabilities, deliberately narrow: evidence, not opinions.

  1. 01

    Screenshot capture

    A rendered image of the page as it was when the data was collected.

  2. 02

    Timestamp and locale

    When the capture happened and which regional version of the page was served.

  3. 03

    Source provenance

    The exact URL and source behind every extracted value.

  4. 04

    Evidence-linked records

    Each BergFlow record points to the capture that supports it.

  5. 05

    Review routing

    Changes and anomalies routed to a reviewer with the evidence attached.

  6. 06

    Version history

    Every new capture adds a version; nothing is silently overwritten.

Key facts

BergCrawler at a glance

The essentials, as a structured record.

Role
Evidence layer for crawled data
Crawl executed by
BergAuto
Record, review and audit
BergFlow
Evidence captured
Screenshot, timestamp, source and locale
Pairs with
Atlass audits, Radarr signals and data sourcing programmes
Availability
Inside managed engagements

Questions

Questions about BergCrawler

What it captures and where it fits.

BergCrawler

What is BergCrawler?

BergCrawler is the evidence layer for crawled data. It preserves the screenshot, timestamp and source of every capture, so each record can be checked against what the page showed.

Does BergCrawler run the crawl?

No. BergAuto executes the crawl; BergCrawler captures and preserves the evidence; BergFlow records the result, routes the review and keeps the audit trail.

Why keep a screenshot?

Pages change. A screenshot with a timestamp shows what a source said at the moment it was collected, which matters for audits, disputes and training-data provenance.

Where is it used?

Wherever crawled data feeds a decision: AI search audits in Atlass, market signals in Radarr, and data sourcing programmes.

Put provenance on your crawled data.

Tell us the sources and the decisions they feed. We show you what an evidence-linked record looks like on your data.