SDLC automation

  1. 01Intake
  2. 02Requirement
  3. 03Design
  4. 04Work order
  5. 05Build
  6. 06Testing
  7. 07Sign-off
  8. 08Memory

Builds the approved work order in the instance, then tests it on the live application. Configuration comes back as a scoped update set. ATF and browser tests capture screenshots, video, and logs as they run. A named person signs the evidence before anything is handed over for promotion.

In the workspace

One workspace. Every part of delivery.

Specialized ServiceNow products in the same workspace. Gene reads the instance, builds on it, and keeps every change in a reviewable update set.

What it covers

Build, test, then a name on the evidence.

GeneWorks does not replace the customer’s promotion path. It hands back a scoped update set and the evidence that supports it.

05

Build

Task orders run through the authenticated MCP connection, the only path the platform uses to read from or write to an instance. Work is sequenced in dependency order. The result is a scoped ServiceNow update set, not a plugin installed into the instance.

06

Testing

ATF execution through GeneWorks’ own test runner, then browser-driven functional testing with Selenium and Playwright against the deployed application. Screenshots, video, and logs are captured as the run happens, not assembled afterwards.

Gate 2

Verify & sign-off

A named human examines the evidence package and signs before anything is handed over for promotion. Neither this gate nor Design Authority Review can be bypassed. Both record the reviewer’s identity.

07

Handover

Promotion follows the customer’s normal path. GeneWorks does not deploy to production. Safe deploys merge instance changes since last deploy first, so someone else’s work is not overwritten. Every step is checkpointed and reversible.

What is different

Own runner. Live UI. Partial is partial.

Testing

Own runner. No CloudRunner.

ServiceNow testing via MCP normally requires CloudRunner, a paid ServiceNow product. GeneWorks built its own runner. ATF runs without that license cost.

Testing

The actual user journey

Functional tests drive the live UI. Screenshots are saved into the workspace. This is not only ATF assertions.

Output

Fluent / TypeScript where it matters

Flow Designer automation can be expressed as TypeScript via Fluent SDK 4.3+, so artifacts are version-controllable, reviewable, and diffable across releases.

Honesty

Partial is reported as partial

The UI says “Partial build — some tasks failed” rather than claiming success. Failure is a first-class state.

Validation Center (functional and UAT with screenshot and video evidence), self-healing tests, and the evidence and audit store for SOX-type controls sit on this product.

Evidence

Produced as the work runs.

Conventional delivery produces evidence afterwards, assembled by hand, at the point someone asks for it. GeneWorks produces it at the moment of execution, tied to the named human who approved it, and stored immutably against the run that produced it.

On a regulated account, that is the difference between an audit finding and a clean walkthrough. The automation is auditable by construction.

Handoff

Nothing releases without a name.

Verify & sign-off is the last gate on this loop. Organizational Memory files what this pass learned, so the next request starts from here.