Reconcile CSV rows against a web portal
Match by explicit keys, retain observed values, and review discrepancies without changing the portal.
Before the first run
- A validated input CSV with stable keys
- An authorized read-only portal scope
- Defined comparison rules including units and dates
Define what the agent can do.
Read and compare only. Correcting a portal value or updating the source file’s authoritative record is a separate task.
The implementation sequence
- 01
Validate the file
Check headers, required keys, duplicates, and type parsing before opening a browser. Preserve the raw input alongside normalized values.
- 02
Match one record at a time
Use the approved identifier set to locate the portal record. Return an ambiguous state when more than one plausible record appears.
- 03
Capture comparable values
Record the observed field, unit, currency, date, and source reference. Do not compare a monthly total with a year-to-date total simply because their labels look similar.
- 04
Apply explicit rules
Run deterministic equality or tolerance checks outside the agent. Preserve both values and the rule that produced each discrepancy.
- 05
Review the exception report
Separate matches, mismatches, missing records, and inaccessible records. The report should let a reviewer inspect one discrepancy without repeating all lookups.
Stop for human approval
Authorize the record scope and export destination; a responsible reviewer decides what action to take on discrepancies.
Approval design principles →Implementation options
Playwright
Deterministic browser automation and testing across Chromium, Firefox, and WebKit.
Verified Sep 27, 2026Browser infrastructureBrowserbase
Managed browser sessions with observability, proxies, and computer-use integrations.
Verified Sep 27, 2026Agent & orchestrationBrowser Use
An open-source browser agent library with managed agents and cloud browser sessions.
Verified Sep 27, 2026Agent & orchestrationStagehand
A browser SDK that mixes natural-language actions with deterministic browser code.
Verified Sep 27, 2026Failure & recovery
- Locale-dependent CSV parsing
- Ambiguous record matches
- Incompatible units or periods
- Portal stale data
What drives cost
- Rows and lookup latency
- Pagination and search complexity
- Mismatch review
Related use cases
Sources & verification
Reviewed Sep 27, 2026. Architecture recommendations are editorial analysis; linked vendor documentation supports the underlying capability and safety facts.
- playwright.dev/docs/locators
- playwright.dev/docs/auth
- developers.openai.com/api/docs/guides/tools-computer-use