“Write ATF tests for the new catalog item and run them.”
TestProven before it promotes
She doesn’t just build it. She runs it.
Phyllis tests the way your team would — she submits catalog orders, runs flows and AI Agents, drives a real browser as a real user, and writes and runs the ATF tests that prove it. A test isn’t done until it has run.
ISO 27001 certified
Per-user OAuth
No training on your data
Approval-gated changes
Lands in your update sets
5,000+ ServiceNow tools
230+ product skills
Hundreds of health checks
Jira & Agile 2.0 two-way sync
Context graph on every build
ATF where it applies
Every write read back
Tables and tools per instance
CMA-grade builds
01One request
Ask once. She runs all of it.
An order, the flows behind it and the tests — run against your instance, with the evidence back in the chat.
Laptop refreshacmedev606 credits
- Ordered Laptop refresh for Abel Tuter: Request RITM0010442 · awaiting approval
- Manager approval and IT provisioning: 2 flows started
- Ran the warranty claim tests: 12 of 12 passed
- Checked the form in the portal: Screenshot attached to the story
@ for records · / for commands
Phyllis is AI and can make mistakes. Please double-check responses.
02In your own words
Test it like a user. In your own words.
Every test runs against your instance and comes back with the evidence — results, transcripts and screenshots.
“Order the laptop item for Abel Tuter and tell me which flows fired.”
Catalog orders
“Test the onboarding subflow with these inputs.”
Flows
“Test my triage AI Agent and show me the transcript.”
AI Agents
“Open the portal as an ITIL user and screenshot the form.”
Browser
“Compare the regression suite before and after the upgrade.”
Upgrade regression
03What she runs
Four ways to prove it. All of them real.
Submits catalog orders, end to end.
She orders the item through the Cart API, resolves every answer to the right record, then reads the RITM back and tells you which flows started.
Runs flows remotely.
The equivalent of Flow Designer’s Test button — run a flow or subflow with real inputs and see how it ended — state, outputs and any error.
Tests AI Agents and Now Assist skills.
Send an AI Agent a real request and get the full transcript back — the tools it called, the records it touched, whether it behaved as instructed.
Uses a browser, like your users do.
Opens pages, reads the form after UI policies run, fills and clicks, views as another user and screenshots the result — watch it live. Available on selected plans.
04By the numbers
Testing, built in.
ATF tools
Tests, steps and suites — including workspace page steps.
Test Management tools
Manual tests in Test Management 2.0.
05How it works
Written, run, and fixed at the cause.
Generate the tests
ATF · Test GeneratorTests, steps and suites written for what changed — or generated with ServiceNow’s own ATF Test Generator.
Run them for real
ATF · Cloud RunnerRun tests and suites, and run the user journeys too — orders, flows, AI Agents and the browser.
Fix the cause, then re-run
Never disabledA failure means a fix — to the build, not the test. Failures are fixed at the cause and re-run before you promote.
06Safe on your instance
Nothing ships without your approval.
Controls that hold at the connection itself — whichever tool, chat or workflow makes the call.
Signs in as you
Per-user OAuth — she inherits your roles and ACLs, and every change is logged under your name.
Approval-gated changes
Approval cards before changes are written — one at a time, or the whole approved build. Nothing reaches production from chat.
Per-instance controls
Build, Read-only or Backlog-only instances. Blocked tables and switched-off tools are enforced on every request.
Audit log
Every approval, write and access change recorded — exportable to CSV.
07Questions
Test, answered.
Still curious? Talk to the team — or book a demo and ask Phyllis in your own instance.
01Does Phyllis write ATF tests or just run them?
Both. She writes tests, steps and suites, can use ServiceNow’s ATF Test Generator, and runs them — a test is never called done until it has run.
02Can she test a flow without ordering anything?
Yes. She can run a flow or subflow directly with the inputs you give her, like Flow Designer’s Test button.
03How does she test an AI Agent?
She sends it a request and returns the full transcript — what it said, which tools it called and what it changed.
04Does browser testing need anything installed?
No plugins in your instance — the browser runs on Phyllis’s side. It’s available on selected plans.
08The whole lifecycle
One teammate. The rest of the lifecycle.
Analyze
Ask ServiceNow anything — your instance, the docs, best practice.
See AnalyzePlan
A sentence, a workshop or a meeting — to a sprint-ready backlog.
See PlanDevelop
The whole change, built in your instance — every write read back.
See DevelopDeploy
Batch and promote update sets — just ask.
See DeployGovern
Your standards and controls — followed on every build.
See GovernAdopt
Capability maps, adoption roadmaps and user guides — get value from what you own.
See Adopt09Get started
Ready to 10× the team you already have?
Your instances, your backlog and your meetings on one side; the Phyllis Knowledge Graph on the other. Phyllis does the work in between.