It sees everything in your instance. And fixes what it sees.
152 specialist agents. Eight apps. One pattern — scan, score, fix, with a human on the gate.
GeneWorks is an AI workforce for ServiceNow. Every app follows the same move — it scans your instance read-only, scores what it finds, and applies fixes only on your approval. Command Center maps the whole platform; CMDB Assessment, Catalog Center, Normalization and UpgradeAssist each score and remediate a corner of it; Document Builder writes the paperwork; and the delivery loop turns a requirement into a tested, deployed change. Twice the delivery capacity. Full control.
ServiceNow delivery is slow, expensive, and brittle.
Every enterprise running ServiceNow faces the same three walls. GeneWorks eliminates all three.
Every change needs a specialist
Adding an SLA, writing a business rule — each needs a certified developer, a ticket, a review cycle, and days of lead time.
⟶ avg. 3–5 days per small config change
No one knows what will break
Platforms accumulate thousands of interdependencies. A change to one field cascades into SLAs, notifications, and integrations — invisibly.
⟶ "it worked in dev" is a $250K incident
Institutional knowledge walks out
The architect who designed the routing left two years ago. Every new hire starts from zero.
⟶ unrecorded knowledge = compounding risk
Scan. Score. Fix. With you on the gate.
Every app runs the same four-step move. Point it at any instance and see the truth in seconds — then fix it with a governed, reversible change.
Scan
Map the instance, read config — touch nothing.
Score
Health scores, maturity, ranked findings.
Propose
Fixes ranked by impact and effort.
Fix
You click; the change runs the gated loop.
Out of 100 stories, GeneWorks completed 73. Here's a clean handoff for the remaining 27.
No theatre. No "AI did everything." When ACLs, API limits, or judgment calls stop us, we say so — and hand you a runbook for the rest. The two human sign-offs exist exactly because some work belongs to humans.
Week 1 looks good. Month 6 looks unrecognizable.
The longer the agents run on your instance, the more your idioms, ACL patterns, and integration map land in the workspace. By month six, the agents read like the team.
| Phase | Window | Effort reduction | What's happening |
|---|---|---|---|
| Cold start | Week 1–4 | 25–35% | Agents learn the instance, team idioms, infrastructure map. |
| Context maturing | Month 2–3 | 40–50% | Failure log and design-pattern library accumulate in the workspace. |
| Compounding gains | Month 4–6 | 55–60% | Agents propose like the team. Reviewer flags drop. |
| Steady state | Month 6+ | ~65% | Plateau. The remaining work is genuinely hard. |
The longer GeneWorks runs, the harder it is to replace.
Every scan adds context. Every fix records a dependency. Over months, GeneWorks accumulates a unique, instance-specific intelligence that no other vendor, consultant, or off-the-shelf tool can replicate.
Why not just use what already exists?
There are several ways teams try to solve the ServiceNow delivery problem. None of them scan, fix, and compound. GeneWorks does.
vs SI Consultants — faster, cheaper, and it remembers everything
A certified consultant costs $150–400/hr with a 3–5 day turnaround per change, and takes the context with them when they rotate. GeneWorks delivers the same artifact — assessment, design, code, tests, packaged change — in minutes, and every session makes it smarter about your instance.
vs raw LLM tools — an LLM on a platform is not a platform
Point Claude Code or Copilot at ServiceNow and you still lack the spec engine, the governance gates, the test runner, and the memory. The model is the commodity; the wrapper is the product. GeneWorks brings 500+ ServiceNow skills and a living map of your instance.
vs ServiceNow Now Assist — we build platforms; Now Assist answers questions
Now Assist summarises tickets and drafts responses. It doesn't map your instance, score your CMDB, tell you what breaks on upgrade, or build and test a change. GeneWorks operates one layer below — the platform-engineering layer Now Assist sits on top of.
vs generic AI coding assistants — context is everything; generic AI has none of yours
Generic tools generate code in isolation. They don't know your ACLs, business rules, or what breaks when you change a field. GeneWorks carries a permanent map of your platform. Paste-and-pray is not a deployment strategy.
The GeneWorks introduction deck.
The problem, the platform, the app suite, how we compare, the economics, and the ask. Use the arrows or your keyboard (← →); press the ⤢ button to present fullscreen.
Self-contained HTML deck — open it anywhere, present from any browser, no software required.
One platform. Eight ways to use it.
Each app runs the same scan-and-fix loop on a different corner of your instance — every read-only, every one carrying a number that used to be impossible. The examples below are real scan output.
Command Center
Scans your entire instance and renders it as one interactive graph — every application, table, and shared foundation, and the connections between them. Tabs: Map · Collisions · Dead weight · Activity · Upgrade. Ask the map anything in plain English.
Application Explorer
Scans a single application end to end — out-of-the-box vs modified vs net-new — and shows where the customizations concentrate. The foundation for any upgrade review.
CMDB Assessment
A full CMDB maturity assessment — health score, findings, and a sequenced roadmap — in a ~12-second read-only scan, versus four to five weeks by hand. Completeness splits mandatory vs recommended; correctness covers stale (60d), orphan, and duplicate CIs. Every finding is a one-click "Make it a Change".
Catalog Center
Rates every catalog item on Build Quality, Experience, and Governance, and attaches the score each fix recovers. Author new items and update existing ones from inside the app. Developer and Leadership views.
Normalization
Matches your company and vendor data against GeneWorks' own reference database — kept current by 18 dedicated agents — and applies the results only on your approval. Accept, reject, or escalate each variant.
UpgradeAssist
Cross-references every documented change between your current and target release against your live instance — read-only — and classifies each one. Blockers surfaced first, grouped by app.
Document Builder
Generates delivery documents from the real instance state, not a blank template. Assess: Health Check Report, Upgrade Assessment Report. Deliver: As-Built, Data Dictionary, Integration Spec, Runbook, Security & Access Design, SOP, Statement of Work, TDD, Test Plan. PDF export.
The Delivery Loop
Six stages — Requirement → Design → Task Planner → Execute → ATF Tests → Functional Tests — with two human gates. The planner reasons about execution order and edge cases, then the agents build, run ATF on our own runner, and drive a live browser functional test before you verify.
Workforce, scanning, the loop & security.
Section 01 — Agent Workforce
152 specialist agents under one orchestrator. Gene classifies intent, routes to the right specialist, synthesizes results, and holds the authoritative task state. Active locks prevent concurrent conflicting changes on the same resource.
| Group | Agents (examples) |
|---|---|
| Strategy & Design | Functional Consultant, Solution Architect, Platform Architect, Process Designer, Integration Lead |
| Build | ITSM / ITOM / CSM / HRSD / SecOps specialists, GRC Analyst, Developer, Scoped App & Integration engineers |
| Quality | ATF Tester, Selenium Tester, QA, UAT, Performance Engineer |
| Documentation & Compliance | Genewriter, GeneDocs, Compliance Advisor, Release & Change Coordinators |
| Operations & Knowledge | CMDB Steward, Knowledge Curator, Evidence Collector, Self-Healing, Dependency Mapper, Workspace Coordinator |
Every build agent runs the same internal chain: Architect → Functional → Coder → ATF-Tester → Selenium-Tester, so each produces a complete, tested artifact. Managed tier runs the latest Claude Opus; BYO connects your own model. Low-token execution keeps cost at roughly a tenth of raw LLM builds.
Section 02 — Read-Only Scanning
Every app scans read-only. Scans read sys tables, dictionary, audit, and configuration to build maps, scores, and findings — and write nothing back. Command Center builds the dependency graph; CMDB Assessment scores completeness and correctness; UpgradeAssist cross-references release notes against the instance; Application Explorer composes baseline-vs-custom analytics. Findings are proposed; they are never applied without a human action.
Section 03 — The Delivery Loop
- Requirement: input becomes a structured, versioned spec.
- Design (✋): full solution design with an explicit "what I'm assuming" list. Human sign-off before build.
- Task Planner: the approved design is decomposed into an ordered, dependency-aware plan.
- Execute: agents build real ServiceNow code via REST, carrying load-bearing decisions through explicitly.
- ATF Tests: server + client suites run on GeneWorks' own runner — no CloudRunner license.
- Functional Tests (✋): live browser tests with screenshot evidence. Human verification before the change closes.
Section 04 — Impact Analysis (H1–H8)
Before any build, GeneWorks scores blast radius across eight dimensions — the same read-only graph analysis behind Command Center's Collisions tab.
Each dimension scores Green / Yellow / Red; the combined score sets the overall risk level. Low proceeds after design sign-off; Medium pauses mid-build for review; High does not start without explicit authorization.
Section 05 — Native Execution & Security
- REST-native: every artifact is created by calling the ServiceNow API directly — no file generation, no XML imports.
- Credentials stay local: stored in your browser, never sent to GeneWorks. Direct, authenticated REST.
- Read-first: scans never write; remediation is explicit and gated.
- Security fence on every message: injected server-side before the model sees it; not modifiable at runtime.
- Blocked patterns: mass deletes, production truncation, credential modification — blocked at the execution layer.
- Sub-production only: production is blocked at the connection test; promotion is your CAB's decision.
Scanning in under 20 minutes.
Three things. No infrastructure, no install. Connect a ServiceNow dev instance and point an app at it.
A ServiceNow dev or PDI instance
Not production. A developer or personal developer instance is perfect.
A service account
Read-only is enough to scan and assess. Write roles only when you move to remediation.
One thing to scan or build
Or just open Command Center and let it map the whole instance for you.
Time to first value
| When | Milestone | What happens |
|---|---|---|
| Day 0 | Connect & Scan | <20 min. Command Center maps the instance; CMDB Assessment and UpgradeAssist run read-only. |
| Day 1 | First Fix | Accept a finding; the six-stage loop delivers a tested change for your sign-off. |
| Wk 1 | Living Map | The workspace holds your instance's dependencies, idioms, and history. |
| Mo 6 | Reads Like Your Team | Compounding memory pushes effort reduction toward ~65%. |
Three rules. Non-negotiable.
Read-only by default
Every scan touches nothing. Remediation is explicit, reviewed, and reversible.
Lower instances only
Dev, PDI, UAT — never production. Enforced at the execution layer, not just policy.
Two human sign-offs
Design review and verification. Neither can be bypassed.
GeneWorks in the field. Real problems. Real deployments.
Three representative scenarios showing how GeneWorks transforms ServiceNow delivery across different organisation types.
Greenfield ITSM Implementation in 6 Days
The Situation
A consulting team had 30 days to stand up a full ITSM environment for a government agency. Prior estimate: 6 weeks with two developers.
What GeneWorks Did
Created P1–P4 SLA definitions with breach notifications · Built 14 catalog items with approval workflows · Bulk-imported 240 users across 9 assignment groups · Wrote and ran 38 functional and ATF tests — all passed · Produced full documentation for every artifact via Document Builder.
Outcome
6-week scope delivered in 6 days. Every artifact documented, tested, and packaged. The team reviewed and committed to production — GeneWorks never touched production.
Legacy Platform Modernisation — No Developer Queue
The Situation
A mature platform had 3 years of technical debt. Their sole internal ServiceNow developer left, leaving a 60-item backlog.
What GeneWorks Did
Command Center mapped the entire platform — Collisions surfaced 12 conflicting business rules · CMDB Assessment scored the data and sequenced the fixes · Rebuilt SLA definitions with correct breach logic · Cleared the 60-item backlog across 3 months — 47 tasks completed, every change checked before it shipped.
Outcome
3 years of technical debt cleared in 3 months without a new hire. Zero production incidents. GeneWorks now runs continuously as the team's platform-engineering capacity.
First ServiceNow Deployment — Senior Architect on Demand
The Situation
A company scaling past 600 employees had just purchased ServiceNow but had no certified internal developers and couldn't justify a multi-month SI engagement.
What GeneWorks Did
Connected to a PDI in 20 minutes — first scan same day · Gene acted as senior architect, explaining trade-offs · Full ITSM environment live within 2 weeks · The IT team learned ServiceNow patterns through the process — no black box.
Outcome
Full working ITSM environment in 2 weeks at a fraction of SI cost. No certified developer needed. The team now runs GeneWorks as its continuous platform-engineering layer.