Browser automation
Bringing data out of web applications that have no API — complementary RPA module
Some systems a business depends on — legacy CRMs, supplier portals, internal ERP screens — never expose an API. Connectors can't reach them. The Portlane ecosystem is extending to cover this gap with a complementary RPA module: a desktop browser automation agent that captures a person's actions once and runs them again as a repeatable, parameterized scenario.
How it fits the platform
Connectors remain the primary way data enters and leaves Portlane. Browser automation is a complementary entry point for the subset of systems that have no API at all — it produces the same kind of normalized records a connector would, then hands them to Portlane over a secure transmission channel so they can feed existing flux.
| Connectors | Browser automation (RPA) | |
|---|---|---|
| Target | Systems with an API, database, or file interface | Web applications with no API |
| Authoring | Configuration (customer tier) | Record once in a browser — no code |
| Runs on | Your on-prem moteurs (Cœur) | A desktop agent (Mac/Windows) on a user's machine |
| Output | Feeds a flux directly | Normalized records delivered to Portlane |
How it works, at a glance
- Record — A business user walks through the task once in a real browser (clicks, form fills, data reads). No scripting.
- Parameterize — The captured sequence becomes a reusable scenario; the values that should vary between runs become named, fillable variables.
- Rejeu — The scenario runs again on demand, in bulk from a spreadsheet (one run per row), or automatically whenever a new file lands in a watched folder.
- Deliver — Extracted or confirmed data is normalized and sent to Portlane over a signed, timestamped, replay-protected channel, ready to feed existing flux.
Built for unsupervised, business-user operation
- Nothing is lost — Runs queue locally and retry automatically through network drops or platform downtime.
- Sessions persist — Logins and application state carry over between runs, so there's no re-authenticating every time.
- Every run is auditable — A history of past runs, including screenshots at the point of failure, makes it possible to diagnose an issue without touching code.
- Secrets stay secret — Sensitive variables such as passwords are masked in the interface and never stored in plain text.
Why not a general-purpose RPA tool?
Enterprise RPA suites solve a similar problem but bring their own licensing, infrastructure, and operating model on top of an existing integration stack. Portlane's RPA module is designed to plug directly into the Cœur a team already runs, so automated web tasks and API-driven flux share one operating model instead of two.
This capability is under active development as part of the Portlane ecosystem. Availability and scope will be confirmed closer to release — demander un devis to discuss your use case.