Lodestar is Pulsar's AI-powered delivery platform for ServiceNow — a constellation of specialist agents, each expert in one stage of delivery, every output gated by human review. Built for regulated clients: UK data residency (AWS Bedrock London), audit trails, and ServiceNow-native knowledge from day one.
The market has already moved
AI is rewriting how ServiceNow work gets delivered. The only question left is how well it’s governed.
AI is automating delivery work
The build and configuration work at the heart of ServiceNow delivery is precisely what modern AI does well. The question for delivery organisations is no longer whether to use AI — it is whether the AI they use is governed well enough to trust.
Buyers now expect it
86% of buyers now demand AI-enabled partners. AI fluency has moved from a differentiator to table stakes — clients want the speed, and they want to know exactly how it is controlled.
Real gains, real risks
Used well, AI delivers +25% speed and +40% quality. Trusted blindly, it ships confident mistakes into production. Lodestar is built for the first outcome: every agent output passes a human review gate before it goes anywhere near a client instance.
Opinionated where it counts
Six constraints shape every Lodestar agent — set on day one, not retrofitted.
ServiceNow Focused
Built with the knowledge of the platform our clients already trust — not a general-purpose tool retrofitted to it.
Claude as the model
Best-in-class reasoning via AWS Bedrock London, with UK data residency as standard.
Multi-agent by design
A constellation of specialists, not one AI asked to do everything.
Human in the loop — always
Every output is a starting point, never a final answer.
Built for regulated clients
Data residency, audit trails and client trust are day-one constraints, not afterthoughts.
Efficiency top of mind
Engineered to reduce token wastage and the time to the right answer.
One pipeline, five specialists
From first scope to go-live, each stage is owned by a specialist agent — and closed by a human review gate.
Understand & Define
Kepler reads scope, stakeholder inputs and documentation, and turns discovery into a structured, prioritised backlog.
A practitioner validates acceptance criteria and scope against the statement of work before design begins.
Design & Build
Polaris turns stories into buildable architecture; Forge turns architecture into ServiceNow components.
Designs are reviewed line by line, and no generated code reaches a client instance without human code review.
Assure & Ship
Lyra builds the test pack; Falcon assembles the release runbook, rollback plan and go-live checklist.
QA coverage and release sequencing are validated on the actual client instance and signed off by the delivery lead.
Specialists, not one AI doing everything
Each agent is expert in one stage of delivery. Pick a star to see what it does — and what you must review.
Kepler
Turns discovery into a backlog the team can build from.
Understand & DefineWhat it does
Kepler reads scope, stakeholder inputs and project documentation, then generates structured user stories, acceptance criteria and a prioritised backlog — in minutes, not workshops.
What it produces
- User stories in As a… I want… So that… format
- Acceptance criteria per story
- Epic and story hierarchy
- Prioritised backlog structure
- Blocked reasons and clarifications required
What you must review
- Check acceptance criteria are testable and complete
- Validate scope against the statement of work
Typical Pulsar use case
IRM programme at a financial services client — a three-sprint backlog structured from discovery outputs in under 30 minutes.
Polaris
Points every story at a buildable design.
Design & BuildWhat it does
Polaris takes user stories, retrieval-augmented knowledge and target-instance access, and produces technical designs a team can actually build — with testing steps, exported to Excel or directly into ServiceNow Agile Development.
What it produces
- Per-story implementation and test scrum tasks
- Story point estimates
- Dependency-aware build orders
- Architecture documentation aware of the target instance's data model
What you must review
- Validate story point estimates — AI tends to be optimistic
- Check table design fits the client's data model
Typical Pulsar use case
Secure by Design Launchpad — scoped app structure and module architecture as a solid high-level design starting point, produced in-session.
Forge
Builds ServiceNow components from approved designs.
Design & BuildWhat it does
Forge generates ServiceNow code from the architecture and stories — Business Rules, Script Includes, Flow Designer components, UI Policies, ACLs and REST integrations.
What it produces
- Production-ready ServiceNow JavaScript
- Flow Designer JSON
- REST integration scripts
- ACL definitions
- Update Set-ready components
What you must review
- Review all code before deployment — check hardcoded values, instance-specific references and logic drift
- No production deploy without human code review
Typical Pulsar use case
FSI Operational Readiness App — assessment form logic, scoring scripts and approval workflows generated from the Polaris architecture.
Lyra
Proves the build before the client does.
Assure & ShipWhat it does
Lyra generates test scripts, test data and QA checklists from stories and acceptance criteria — identifying edge cases, negative scenarios and integration tests along the way.
What it produces
- Test scripts mapped to acceptance criteria
- Synthetic test data sets
- Edge-case catalogue
- Regression checklist
- Defect reporting template
What you must review
- Validate scripts cover the scenarios the client will actually use
- Add client-specific edge cases
- Confirm test data is synthetic
Typical Pulsar use case
IRM Phase 1 Implementation — a full test pack across positive, negative and integration scenarios for all 14 epics.
Falcon
Carries the release safely over the line.
Assure & ShipWhat it does
Falcon produces release documentation, Update Set sequencing, deployment runbooks, rollback plans and go-live checklists from build artefacts and test results.
What it produces
- Deployment runbook
- Update Set sequence plan
- Promotion checklist
- Rollback procedure
- Go-live comms template
- Monitoring checklist
What you must review
- Validate Update Set dependencies and sequencing on the actual client instance
- Confirm rollback steps are tested
- Reflect client-specific approval gates
Typical Pulsar use case
DR deployment — a full runbook and rollback plan across dev, test and production, including integration steps.
Every output is a first draft — a practitioner decides what's right.
Human review is required at every stage — backlog, design, code, test and release. Lodestar accelerates practitioners; it never replaces their judgement.
See Lodestar on your delivery
Bring us a slice of a real programme — we’ll show you what the constellation does with it, and where the review gates sit.