Built by Eclipse - Parallax
Know what depends on what. Prove how you know.
Every impact assessment, change decision and resilience report your organisation makes runs on one question: what depends on what? In most enterprises nobody fully knows, and the regulators have noticed. Parallax reads the evidence your estate already produces and turns it into confidence-scored, human-approved dependency maps you can hand to an auditor.
185
Critical applications mapped in a live proof of concept at a global financial services firm
27% to 68%
Confidence uplift on one relationship when independent telemetry corroborated it
No agents
Works from exports you already produce: nothing installed, no credentials to your hosts
The problem
The question every outage asks first, and nobody can answer
When a shared service degrades, the first question is what else just broke? When a change is proposed, it is what could this touch? When a regulator reviews your resilience, it is show us how you know. All three run on the same dependency picture, and in most enterprises that picture is tribal knowledge, stale spreadsheets and a CMDB layer that was never populated.
The cost is not hypothetical. Industry surveys put an hour of downtime above $300,000 for over 90% of mid-size and large enterprises, and around 80% of unplanned outages are self-inflicted through change: changes assessed against a dependency map that was wrong.
And assertion no longer passes. EU DORA Article 8(4) requires financial entities to map the links and interdependencies between their information and ICT assets. The UK's PRA SS1/21 and FCA PS21/3 required important business services mapped in enough detail to find vulnerabilities, refreshed at least annually, by March 2025. That deadline has passed, and supervisors are now challenging firms to evidence how their maps were built.
- Blast radius is invisible: outage impact is discovered from the phones, not the map.
- Change collisions happen on dependencies nobody documented.
- Resilience regimes demand evidence for every dependency claim, not recollection.
- Manual mapping exercises take months and are stale before they finish.
Why it is still unsolved
Every tool you own sees a slice. None sees the dependency.
Discovery walks the vertical stack: application on host on VM on hyperscaler, because that can be read from the operating system. What an application relies on is lateral, and lateral is where every tool category stops short.
Point-in-time snapshots
Dependency probes capture the TCP connections that happen to be open at scan time. The month-end batch feed, the failover path and the overnight settlement job simply do not exist in the map.
Depth for a few, nothing for the rest
Per-service entry points, per-host credentials and pattern upkeep price top-down mapping at dozens of services. The estate has thousands of applications, and the maps rot silently between releases.
Bounded by the agent perimeter
Application performance tools see real traffic, but only where agents are paid for, and they map processes and technical services, not the business applications your governance runs on.
Connections are not dependencies
A traffic flow shows who opened a socket, not who relies on whom. No direction, no criticality, and no chain resolving an IP address to a named business application.
This is the gap Parallax closes: it corroborates across the slices each tool does see, resolves the result up to business-application level, and puts a human decision on every edge that matters.
What you get
A living dependency map, with the evidence attached
-
Blast radius you can see
Click any application and see what genuinely depends on it, in both directions, with criticality carried. Incident impact and change conflict stop being guesswork.
-
Evidence a regulator can follow
Every relationship is traceable to the signals behind it: which sources, which paths, over what period, and who approved it. The artefact you hand to audit, not a diagram you defend from memory.
-
Relationships written into ServiceNow
Approved edges publish into your CMDB in CSDM vocabulary, extendable to the semantics that matter, such as critical versus operational dependency. Value lands in the platform you already own.
-
A map that stays current
Scheduled re-runs produce a drift report: what appeared, what vanished, what weakened. Every reviewed decision recalibrates the scoring to your estate, so trusted coverage compounds.
Where it earns its keep
One evidence engine, several hard problems
Operational resilience and compliance
Evidence the dependencies of your important business services for DORA, PRA and internal resilience regimes, with the audit trail supervisors now ask for, not assertions from memory.
Impact assessment that works
Changes and incidents assessed against real dependency data. See which applications are affected when a shared service degrades, before the phones start.
CMDB and CSDM completion
Fill the relationship gaps discovery tools leave behind, propose application service candidates where the evidence supports them, and keep the service model honest over time.
Migration and estate change
You cannot migrate what you cannot map. Wave planning, datacentre exit and right-sizing decisions made against evidenced dependencies rather than interviews and hope.
How it works
From raw signals to relationships you can trust
Parallax is an evidence pipeline. It starts deterministic, grounded in structured facts, direction, persistence and multi-source agreement, and applies AI only where it adds judgement: classifying, explaining, and learning from your experts' decisions.
-
The data you already have
ServiceNow discovery and cloud connector data, incidents and changes, integrations, network telemetry, cloud account metadata. No new agents, no new instrumentation.
-
Deterministic triples
Every signal becomes a structured statement: which system, which direction, how persistent, seen in how many independent sources. Facts first, inference later.
-
Confidence, corroborated
Each candidate relationship gets a confidence score. Independent sources must agree, and every corroborating source raises it. Conflicts are surfaced, not hidden.
-
Experts approve, Parallax learns
AI classifies each candidate and explains it in plain English. Your experts accept, correct or reject with the evidence in front of them. Every decision recalibrates the scoring model to your estate.
-
Approved records publish
Only what your experts accepted is written to ServiceNow, with the evidence trail attached to the record.
Nothing is written to ServiceNow without a human decision behind it.
How we work
For estates where "we think" is no longer good enough
- Regulated enterprises - operational resilience programmes
- CMDB and CSDM owners - Enterprise Architects
- Cloud platform teams - large ServiceNow estates
Evidence, not opinion
Relationships are built from signals in your data, never from surveys or workshop recollection. No single source proves a dependency: independent sources must agree, and where they do not, the gap is flagged rather than papered over.
AI explains, humans decide
AI does the classification and the explaining; your experts make the call. Nothing is written to ServiceNow unreviewed unless you set the threshold that allows it.
Semantics that matter
A dependency is not automatically a critical one. Parallax distinguishes the relationships that would trigger resilience action from day-to-day operational links, so your maps inform decisions instead of inflating them.
Built to work together
Part of the Eclipse AI toolchain
Parallax maps the relationships between systems. SOLO reads and remediates the estate beneath them. Together they turn platform data into decisions.
SOLO
Our CSDM automation agent: it processes the estate data Parallax draws on, and accelerates the remediation its findings recommend.
Explore SOLO DiagnosePlatform Health and Value
150+ read-only checks against your live ServiceNow estate: the assessment that shows where the gaps are before you map them.
Explore the assessmentReady to fix your foundation? Let's talk.
We start with discovery to understand your challenges and goals. Let's discuss what's possible.