Internal analysis · Zions Bancorporation

Fluxnova vs. our existing automation stack

How Fluxnova — a BPMN engine forked from Camunda 7, governed by — stacks up against IBM BAW, UiPath, and Power Automate. Includes a deep dive on the engine capabilities Fluxnova brings, and how they compare.

Fluxnova
IBM BAW
UiPath
Power Automate
Section 01

High-level review

Fluxnova is wired up differently from the other three — it's an engine with APIs, you bring your own UI. BAW, UiPath, and Power Automate all ship the UI with the product. Strengths and watch-outs below reflect that.

FINOS / Linux Foundation

Fluxnova

Under evaluation

A BPMN engine you build on top of — not a closed suite. Engine is forked from Camunda 7, which has been running in production for about a decade.

Model
Engine-first. Bring your own UI. Platform team runs it as an internal product.
License
Apache 2.0 (OSI-approved)
Deployment
On-prem or cloud. Docker / Kubernetes. Per-BU clusters with built-in tenancy inside each.
Context
v1.0 shipped Nov 2025. Releases roughly quarterly. Engine itself is Camunda 7 — ~10 years in production.

Strengths

  • Engine + full API - custom build the UI based on business needs
  • You control the engine. Size clusters and pods per BU via IaC (infrastructure as code).
  • Two isolation layers: per-BU infrastructure, plus built-in tenancy inside each
  • BPMN modeler for authors. Delegates + modeler templates play the role of BAW toolkits or custom packages/components, but with greater control.
  • Bank-led governance — Fidelity, First National, and others rotate as primary contributors under FINOS
  • No per-user, per-bot, or per-PVU meter. Apache 2.0 — fork it if you need to.

Watch-outs

  • Shipped UI is basic on purpose — real UX is on you
  • Needs a platform team: Java, APIs, IaC, Kubernetes
  • Smaller prebuilt connector catalog than Power Automate or UiPath — expect to build delegates
  • No built-in RPA bots — keep UiPath or Power Automate for UI automation of legacy apps
IBM

IBM Business Automation Workflow

Enterprise BPM suite. Engine, UI, case, and tooling all bundled together.

Model
Closed suite. The UI ships with the product.
License
Proprietary (IBM Program License)
Deployment
On-prem or IBM Cloud Pak for Business Automation (OpenShift)
Context
Mature. Lineage: Lombardi → IBM BPM → BAW.

Strengths

  • Enterprise mature, proven at large banks
  • Process Portal, case UI, document list — all out of the box
  • Toolkits for reusable integration assets (same idea as Fluxnova delegates)
  • Strong audit, case management, and regulated-industry tooling

Watch-outs

  • The shipped UI is what you get. Extending it is a lot of work.
  • High TCO — licensing + IBM middleware + infrastructure
  • One shared instance. BU isolation is logical, not hard.
  • Proprietary extensions lock you to the IBM stack
UiPath Inc.

UiPath

Best-in-class RPA. Authoring and orchestrator UX come with the product.

Model
Commercial SaaS or hybrid. Authoring happens in Studio / Orchestrator.
License
Per-bot + per-user. Automation Cloud or on-prem.
Deployment
SaaS (Automation Cloud), hybrid, or self-hosted
Context
Market leader in RPA. Expanding into BPM and agentic AI.

Strengths

  • Industry-leading attended and unattended bots
  • Great at automating legacy desktop apps
  • Mature AI Center, Document Understanding, Task Mining
  • Big marketplace of prebuilt activities and connectors

Watch-outs

  • Per-bot licensing gets expensive fast
  • Opinionated UX — you can't really reshape it per business unit
  • Many tenants = real operational overhead
  • XAML workflows aren't portable. Teams end up writing custom APIs to work around it.
Microsoft

Microsoft Power Automate

Low-code workflow + RPA, fused into M365 / Azure.

Model
Commercial SaaS. Authoring happens in the Power Platform.
License
Per-user / per-flow, plus hosted-bot minutes
Deployment
SaaS (Azure). On-prem data gateway for hybrid reach.
Context
Rapidly iterating. Copilot-powered low-code suite.

Strengths

  • Deep M365, Dynamics, Teams, SharePoint hooks
  • Low barrier — business users can build cloud flows
  • AI Builder + Copilot for generative automation
  • Dataverse as the governed data backbone

Watch-outs

  • Opaque licensing — per-user, per-flow, AI credits, all at once
  • Strong inside Microsoft; weaker outside
  • Proprietary flow model — limited BPMN/DMN support
  • Authoring UX is fixed. Zions teams already build side-APIs to patch it.
Section 02

Fundamental capabilities

Cells are colored by tool. Darker = stronger support. Hover any cell for the specifics.

Capability
Fluxnova
IBM Business Automation Workflow
UiPath
Microsoft Power Automate
Architecture model
Engine-first vs. closed suite
Can you swap the UI, or are you locked to the vendor's experience?
Full
Weak
Weak
Weak
Architecture model
API-first / bring-your-own UI
Coverage and completeness of APIs for building custom frontends
Full
Partial
Partial
Partial
Architecture model
Operational control of the engine
Can the platform team pull levers on scaling, persistence, and execution?
Full
Partial
Partial
None
Platform
Reusable integration assets
How a central team delivers connectors / templates to business units
Strong
Full
Partial
Partial
Platform
Per-business-unit isolation
Hard boundary between BU workloads — separate DBs, clusters, even infrastructure
Full
Partial
Partial
Partial
Platform
Multi-tenancy inside a BU instance
Department-level separation inside the same cluster
Full
Partial
Partial
Partial
Citizen developer
Authoring surface for non-developers
Can a business user build or modify a working automation without engineering?
Strong
Partial
Strong
Full
Citizen developer
Governance & guardrails
Can a central team contain blast radius without blocking authoring?
Strong
Partial
Partial
Strong
Citizen developer
Hand-off path to platform engineering
When a citizen-built automation outgrows its author, can engineering take over without a rewrite?
Strong
Partial
Weak
Partial
Standards & portability
BPMN 2.0 process execution
Native, standards-compliant process modeling and execution
Full
Strong
Partial
Weak
Standards & portability
BPMN artifact portability
Can your BPMN files leave the vendor? Different question from whether the engine runs BPMN.
Full
Weak
None
None
Standards & portability
DMN 1.3 decision engine
Portable decision tables and FEEL expressions
Full
Strong
Weak
Weak
Standards & portability
Case management (CMMN)
Unstructured, knowledge-worker-driven case handling
Strong
Full
None
Weak
Automation style
Long-running orchestration
Durable, multi-day / multi-system workflows with state
Full
Full
Partial
Partial
Engine capabilities
External task pattern (pull workers)
Workers poll the engine for work; can run anywhere, in any language
Full
None
Partial
None
Engine capabilities
Live in-flight process migration
Move running instances between process versions without cancel/restart
Full
Partial
None
None
Engine capabilities
BPMN compensation handlers
Standards-based undo semantics for multi-step, multi-system work
Full
Partial
None
None
Engine capabilities
Batch operations at scale
Bulk migration, cancel, retry, modify across thousands of instances
Full
Partial
Partial
Weak
Engine capabilities
Job prioritization + incident model
Per-job priority values, retry cycles, incidents as first-class entity
Full
Partial
Partial
Weak
Engine capabilities
Tunable history / audit levels
Per-deployment audit volume; decoupled from runtime
Full
Partial
Partial
Partial
Engine capabilities
Process unit-testing framework
Assert against a BPMN process the way you assert against code
Full
Partial
Partial
None
Automation style
UI / attended RPA bots
Screen-scraping and desktop automation of legacy apps
None
Partial
Full
Strong
Automation style
API / system-to-system integration
REST, SOAP, messaging connectors between back-office systems
Strong
Strong
Strong
Full
Data Control
On-prem + cloud parity
Can you run the same workload behind the firewall or in cloud?
Full
Full
Strong
Weak
Commercial
Licensing cost model
How spend scales with users / bots / processes
Full
Weak
Partial
Partial
Commercial
Avoiding vendor lock-in risk
Portability of processes, data, skills
Full
Weak
Weak
Weak
Governance & maturity
Underlying engine maturity
How battle-tested is the actual code running production workloads?
Full
Full
Strong
Strong
Governance & maturity
Governance model & community
Who actually decides what ships, and on what cadence?
Full
Strong
Partial
Partial
Governance & maturity
Financial-services alignment
Is the roadmap shaped by financial institutions?
Full
Strong
Partial
Partial
Support level:FullStrongPartialWeakNone

Can Fluxnova orchestrate the rest of our automation stack?

Yes — as an orchestration engine, not a turnkey dashboard. A master BPMN process in Fluxnova drives UiPath bots, Power Automate flows, BAW processes, and VBA-wrapped services through their existing REST APIs. Industry shorthand: orchestration of orchestrators.

Master BPMN process
Fluxnova engine
drives
UiPath botsIBM BAW processesPower Automate flowsVBA / Office ScriptsAny REST service
What Fluxnova owns

Cross-tool state, audit trail, retries, compensation, and the readable BPMN model itself. One source of truth for how work moves end-to-end.

What it doesn't do automatically

A single-pane-of-glass UI is a separate build — a Zions Automation Console on top of Fluxnova's API plus the others'. Real project (~1–2 quarters), not a config switch.

Section 03

Detailed deep dive

Nine dimensions that matter for bank-scale automation. Click any row to open the four-column side-by-side.

Section 04 · Under the hood

Camunda 7 Lineage

Fluxnova's engine is forked from Camunda 7 — with 10 years of production-hardened BPM runtime. These eight capabilities are the ones where the lineage matters for a regulated bank, with the real analogs (or gaps) for IBM BAW, UiPath, and Power Automate called out on each card.

DistinctiveCamunda 7 lineage

External Task pattern

The engine publishes work to a topic. Workers poll, lock, and complete it. Workers can run in any language, on any host, on any network — even outside the engine's cluster. The Camunda docs put it plainly: an external worker doesn't need to run on the same machine, in the same cluster, or even on the same continent.

Why it matters for Zions

Run a worker in a different security zone from the engine. Engine in the shared platform cluster, worker inside the core-banking perimeter calling CICS. And pick the right language per worker — Python for ML scoring, Node for a webhook handler, Java for a Db2 writer.

How the others compare
Fluxnova
Match

Pattern comes straight from Camunda 7 — topics, polling, lock/complete are first-class APIs.

IBM Business Automation Workflow
Gap

No direct analog. Service tasks run inside BAW; polyglot workers need a separate integration bus.

UiPath
Partial

Orchestrator queues are conceptually similar — bots pull work — but the worker must be a UiPath bot, not an arbitrary service.

Microsoft Power Automate
Gap

Flows are push-driven; no equivalent pull-from-topic pattern.

DistinctiveCamunda 7 lineage

Live process instance migration

A migration plan maps activities from one process version to another. Running instances move forward in-flight. Task properties (assignee, variables) are preserved. Runs sync or async-batched, with validation at plan time and execution time.

Why it matters for Zions

Ship a new version of the dispute or loan process while thousands of instances are already running. Move them forward — no cancel, no restart, no manual reassignment. Removes one of the main reasons teams fork processes or freeze deployments.

How the others compare
Fluxnova
Match

Full migration API; batch-async for volume; validation at both plan and execution time.

IBM Business Automation Workflow
Partial

In-flight migration is possible but brittle and largely manual — teams typically carry old + new versions in parallel.

UiPath
Gap

Workflow versions exist; there is no concept of migrating in-flight instances.

Microsoft Power Automate
Gap

Flow versions exist; no in-flight migration.

DistinctiveCamunda 7 lineage

Batch operations at scale

Bulk migration, cancel, modify, retry across thousands or millions of instances. Three-job architecture (seed → execute → monitor) keeps transactions small, so failures don't rewind the whole batch. Monitored in the Monitoring Tool.

Why it matters for Zions

A regulator changes a rule. A vendor outage strands 40,000 instances in the same state. You need to retry, cancel, or modify them surgically without knocking the engine over. This is exactly what batch operations are for.

How the others compare
Fluxnova
Match

Same batch framework; seed/execute/monitor pattern; tunable invocations per job.

IBM Business Automation Workflow
Partial

Bulk admin operations exist but aren't as structured; large batches often done with custom scripts.

UiPath
Partial

Queue-level bulk operations exist, but process modification at scale isn't a native pattern.

Microsoft Power Automate
Gap

No equivalent bulk runtime operations.

Bank-relevantCamunda 7 lineage

BPMN compensation handlers

BPMN compensation events. If a later step fails, earlier compensatable activities run their compensation handlers to undo work — refund, reverse entry, release collateral. Long-running transaction semantics without distributed ACID.

Why it matters for Zions

Loan origination fails at funding. You need to release the credit reservation, reverse the hold on collateral, and notify the customer. Compensation handlers model that in BPMN instead of smearing undo logic across integration code.

How the others compare
Fluxnova
Match

Native BPMN compensation events + handlers; works with subprocesses and transactions.

IBM Business Automation Workflow
Partial

Supports the BPMN construct but less commonly used — most teams implement rollback in integration code.

UiPath
Gap

No formal compensation pattern; rollback is ad-hoc inside workflow logic.

Microsoft Power Automate
Gap

No formal compensation pattern; rollback modeled as branching or Try/Catch scopes.

Bank-relevantCamunda 7 lineage

Job prioritization with incident management

Every job gets a priority (long int, higher = more important), settable on the BPMN or at runtime via API. Failed jobs run retry cycles (e.g. 5 retries, 5 min apart). When retries exhaust, an Incident is created and lands in the Monitoring Tool.

Why it matters for Zions

Under load, the engine can service VIP / high-value transactions before routine work on the same runtime. When something breaks, the incident model gives operators a clean queue of "things a human needs to look at" — not a pile of half-logged exceptions.

How the others compare
Fluxnova
Match

Full priority + retry + incident model inherited from Camunda 7.

IBM Business Automation Workflow
Partial

Retry and error handling exist; priority model is less elegant, incidents surfaced through Business Performance Center instead of a dedicated construct.

UiPath
Partial

Queue priorities for bots; incident-equivalent is "jobs in faulted state" in Orchestrator.

Microsoft Power Automate
Partial

Basic retry policies; no formal incident-management primitive.

Bank-relevantCamunda 7 lineage

Tunable history / audit without runtime coupling

History levels (NONE → ACTIVITY → AUDIT → FULL) are fully configurable and decoupled from the runtime. The runtime never reads the history DB. Custom handlers let you stream events to any backend — S3, Splunk, a data lake, etc.

Why it matters for Zions

Audit requirements differ by workload. A marketing-campaign process doesn't need full variable-change history; a sanctions-screening process does. Tunable levels + pluggable handlers mean regulatory audit where required, without paying the write cost everywhere else.

How the others compare
Fluxnova
Match

Same decoupled history architecture; custom DbHistoryEventHandler + user operation log.

IBM Business Automation Workflow
Partial

Extensive audit, but tightly coupled to runtime database; harder to tune per workload.

UiPath
Partial

Orchestrator logs + Elasticsearch-style export; single level of verbosity.

Microsoft Power Automate
Partial

Run history is SaaS-managed; Purview overlays governance but the engine history model is fixed.

Bank-relevantCamunda 7 lineage

Optimistic locking + exclusive-per-instance jobs

Optimistic concurrency control — no pessimistic row locks slowing things down under load. At the same time, jobs from the same process instance run exclusively by default, so parallel gateways inside one instance don't race each other.

Why it matters for Zions

Under heavy load (thousands of dispute claims, wire confirmations, whatever), the engine scales out horizontally without lock contention. And a single instance with two parallel branches stays safe without the developer having to think about it.

How the others compare
Fluxnova
Match

Inherited Camunda 7 concurrency model.

IBM Business Automation Workflow
Partial

Scales well historically, but concurrency tuning is opaque and uses pessimistic locking in places.

UiPath
Partial

Orchestrator handles concurrency at the queue level; engine internals not exposed.

Microsoft Power Automate
Partial

Runtime-managed; concurrency is a Microsoft-operated concern.

Bank-relevantCamunda 7 lineage

Process testing framework

camunda-bpm-assert + JUnit + an in-memory H2 deployment. Unit-test a BPMN process like you'd unit-test a method — assert the process is waiting at a specific task, complete it, assert the next task is user-assigned, and so on.

Why it matters for Zions

Automation processes behave like code and regress like code. Real unit tests mean process changes go through normal CI/CD with actual coverage — not "smoke-test in lower env and hope."

How the others compare
Fluxnova
Match

Full camunda-bpm-assert lineage; tests run in-memory or against a real engine.

IBM Business Automation Workflow
Partial

Unit testing possible but heavy; most teams rely on integration testing in dev environments.

UiPath
Partial

UiPath Test Suite exists (separately licensed), focused on UI/robot flows.

Microsoft Power Automate
Gap

No formal unit-testing story; validation is runtime-only.

Bottom line on the lineage

The Fluxnova project is about a year old. The engine underneath it has been running in production at banks and regulated enterprises for about a decade. The features above aren't new to production — just new to the FINOS label.

Section 05 · Interactive

Weighted fit explorer

Slide the dimensions you care about. Bars rescore in real time so you can see how the four stack up for a specific Zions scenario — not a generic leaderboard.

Weight each dimension (0–100)

Engine control & operational levers
IaC-driven scaling, source access, tunable runtime
90
API-first / bring-your-own UI
Build a custom front-end per business unit on top of APIs
95
Per-BU isolation + built-in tenancy
Hard separation across business units, soft separation inside
85
Licensing cost efficiency
Lower cost per process / per user = higher score
70
BPMN / DMN standards fidelity
Portability of processes away from any single vendor
75
Attended + unattended RPA depth
Surface-level desktop and UI automation of legacy apps
35
Long-running orchestration
Durable processes that span days, weeks, multiple systems
80
Prebuilt connector breadth
Out-of-box integration surface area (as opposed to self-built delegates)
50
FSI-aligned governance
Consortium model vs. single-vendor roadmap
80
Underlying engine maturity
Years of production hardening in the actual runtime
70

Composite fit score

#1Fluxnova
87.6
#2IBM Business Automation Workflow
61.2
#3UiPath
59.6
#4Microsoft Power Automate
51.5

Weighted average of dimension scores per tool. Higher = better fit for the current weights.

Weighted profile (radar)

FluxnovaIBM Business Automation WorkflowUiPathMicrosoft Power Automate
Methodology & dimension sources

Per-tool dimension scores (0–100) are estimates drawn from the Fluxnova docs (fluxnova.finos.org, docs.fluxnova.finos.org, GitHub discussions), Camunda 7 docs, IBM BAW product pages, UiPath docs, and Microsoft Learn. They reflect a US regional bank context — not a universal ranking.

Weights are applied as a normalized linear combination, so the total scale stays consistent regardless of slider range.

Section 06 · Summary

What Fluxnova can do for Zions Bancorporation

Fluxnova isn't a drop-in replacement for any one tool we run today, but in many instances it can do the same job better. It also provides functionality that we are currently lacking in several areas.

Opportunity

Stop writing side-APIs to patch closed suites

We already write custom APIs to bridge BAW, UiPath, and Power Automate to the systems BUs actually need. Fluxnova exposes that surface out of the box — any operation you'd want against the engine is already in the API. Less glue code to own.

Opportunity

Real isolation per business unit, not just a tenant ID

Different BUs can run on their own databases, their own clusters, even their own infrastructure — sized via IaC — while sharing tooling. Built-in multi-tenancy handles department-level separation inside each BU. Addresses the UiPath tenant-sprawl overhead we've already hit.

Opportunity

A seat at bank-led governance

FINOS governs Fluxnova, with a rotating primary-contributor role (Fidelity through June, then others). Zions can be involved in several ways even if we're not a primary contributor — board involvement, roadmap influence, or upstreaming our customizations as configurable options. Same pattern Fidelity and others already use.

Opportunity

Consolidate, don't collect

Today it's BAW + UiPath + Power Automate + custom API glue. Fluxnova is a credible orchestration backbone that lets us retire redundant capability (especially BAW) while keeping UiPath for UI-level RPA and Power Automate for M365-adjacent flows.

Opportunity

Workers in the right security zones

The external task pattern lets workers run outside the engine cluster — different VPC, different security zone, different language. Sanctions screening inside the restricted perimeter, webhooks in the DMZ, engine orchestrates both. BAW can't do this. Power Automate can't do this.

Opportunity

Ship process changes without freezing the business

Live in-flight migration moves running instances forward onto a new process version — no cancelling, no reassigning, no running old + new in parallel. For case workflows that stay open for weeks, that's monthly releases vs. quarterly ones.

Risks to manage

You have to own a platform team

Fluxnova trades vendor licensing for in-house capability. We need a small team with Java / API / IaC / Kubernetes skills, treating the platform as an internal product — shipping delegates, modeler templates, and reusable UI components. Without that, "build your own UI" becomes "build everything from scratch."

The included tasklist interface is pretty basic

That's the philosophy, not a bug — you're expected to build your own UX on top. But it means you can't evaluate Fluxnova on its out-of-box UX. The bar for Zions: can our platform team ship the UI patterns we need? Pilots should prove it.

Smaller prebuilt connector catalog

Power Automate and UiPath both have bigger connector libraries. For systems we already expose as internal services this is a non-issue; for niche SaaS, budget for custom delegates and treat them as platform-team deliverables.

Zions and open source

The bank hasn't historically leaned on open source. Making Fluxnova work needs a governance story — FINOS participation, a liaison role, a clear contribution policy — not just a technology decision. Sell the consortium model, not the licensing savings.

Recommended next steps

  1. 1Stand up a Fluxnova sandbox (Docker/Kubernetes) with IaC templates for per-BU sizing — proves the operating pattern, not just the tool.
  2. 2Build one reusable delegate + modeler template as a reference implementation — pick an integration we already care about.
  3. 3Pilot 2 processes: one BAW candidate, one UiPath-heavy candidate, each with a purpose-built UI on top of Fluxnova APIs.
  4. 4Prove what the others can't match — run a live in-flight migration and a batch cancel/retry on the pilot workloads.
  5. 5Benchmark the pilots on developer hours, run cost, BU-specific UX fit, and BAW renewal exposure.
  6. 6Formalize Zions' FINOS participation — board seat or contributor role — and open channels to the current primary contributors.

Sources

Prepared for Zions Bancorporation · Internal analysis. Incumbent notes also informed by direct operator experience with the Zions automation stack.