Adaptive Case Management: Structuring Unpredictable Knowledge Work

Adaptive case management (ACM) is an approach to organizing knowledge work that cannot be scripted in advance. Instead of forcing work through a predefined flowchart, adaptive case management gives skilled workers a shared case file, explicit goals, and the discretion to choose the next step based on what each case reveals. It applies structure where structure helps — deadlines, mandatory checks, audit trails — and preserves freedom where professional judgment matters most.

The distinction is not academic. Fraud investigations, legal matters, patient care plans, and contested insurance claims are case-driven: the next action depends on what the previous action uncovered. Business Process Model and Notation (BPMN) diagrams assume the path is known before work begins. Adaptive case management assumes it is not, and builds the operating model around that reality.

This guide explains what adaptive case management is, how it differs from traditional business process management (BPM), and when to choose ACM over BPMN. It also unpacks the case file concept, discretionary versus mandatory tasks, the case lifecycle, dynamic role assignment, rules engines, and the Case Management Model and Notation (CMMN) standard — then shows how organizations apply ACM to investigations, legal work, healthcare coordination, and complex claims.

What Is Adaptive Case Management and Why Does Knowledge Work Need It?

Adaptive case management is a discipline — and a class of software — that coordinates unpredictable, knowledge-intensive work around a case rather than a process. A case is a goal-oriented container that accumulates data, documents, tasks, and decisions as a situation evolves. The worker, not a process engine, decides what happens next within governed boundaries.

The term was popularized by the 2010 book Mastering the Unpredictable, edited by Keith D. Swenson, then Vice President of Research and Development at Fujitsu America and chairman of the Workflow Management Coalition. The book, published by Meghan-Kiffer Press, defined ACM precisely in its published excerpts.

"Systems that are able to support decision making and data capture while providing the freedom for knowledge workers to apply their own understanding and subject matter expertise to respond to unique or changing circumstances within the business environment."

— Definition of adaptive case management from Mastering the Unpredictable, edited by Keith D. Swenson, Meghan-Kiffer Press, 2010

The stakes are enormous because knowledge workers now form the core of enterprise value creation. Peter F. Drucker, the management theorist who coined the term "knowledge worker," framed the challenge decades ago in his 1999 California Management Review article, Knowledge-Worker Productivity: The Biggest Challenge.

"The most valuable asset of a 21st-century institution, whether business or non-business, will be its knowledge workers and their productivity."

— Peter F. Drucker, California Management Review, 1999

Yet that productivity is leaking. Recent research quantifies how badly fragmented tools and improvised coordination hurt knowledge work:

• Roughly 25 percent of the workweek is lost to searching for information, according to Atlassian's State of Teams 2025 report, which surveyed 12,000 knowledge workers across six countries.
• An estimated 2.4 billion hours are wasted annually across Fortune 500 companies on information retrieval alone, per the same Atlassian research.
• 50 percent of knowledge workers report having unknowingly duplicated work a colleague had already done.
• 62 percent of employees said they spend too much of the workday hunting for information, according to Microsoft's Work Trend Index published on May 9, 2023.

The Cost of Forcing Unpredictable Work into Flowcharts

When organizations model case-driven work as rigid flowcharts, workers route around the system. They coordinate in email threads, track evidence in personal spreadsheets, and make decisions that leave no audit trail. Consequently, context fragments and rework multiplies.

Coveo's Employee Experience Relevance Report, based on a survey of 4,000 employees in the United States and United Kingdom, found workers spend about three hours per day searching for information, and that 42 percent of the information they sift through is irrelevant to their role. Adaptive case management attacks this problem directly by making one shared case context the system of record for the work.

How Adaptive Case Management Differs from Traditional BPM

Traditional BPM is procedural. Analysts model a process at design time — typically in BPMN, the Object Management Group (OMG) standard finalized as BPMN 2.0 in January 2011 — and a process engine then drives every instance down the modeled path. This works superbly for routine work such as order processing, employee onboarding, or invoice approval, where repetition justifies detailed modeling.

Adaptive case management inverts the logic. It is declarative rather than procedural: the system defines what can or must be done, while the knowledge worker decides how and when, guided by goals and constraints. Swenson's core argument, from the excerpts of Mastering the Unpredictable, remains the sharpest statement of why this matters.

"Scripting the work process in advance offers little benefit for increasing knowledge-worker productivity, much less the ability to adapt to changes in the business environment."

— Keith D. Swenson, editor of Mastering the Unpredictable, former Vice President of Research and Development at Fujitsu America

In practice, the two approaches differ along four axes:

• Timing of design: BPM fixes the path at design time; ACM plans continuously at runtime as facts emerge.
• Locus of control: BPM engines dictate the next step; ACM workers select it within guardrails.
• Center of gravity: BPM organizes around the flow; ACM organizes around the case file and its data.
• Handling of variance: BPM treats deviation as an exception to be minimized; ACM treats variance as the normal condition of knowledge work.

Importantly, ACM complements rather than replaces BPM. As Flowable's practical guide to adaptive case management notes, mature platforms let structured BPMN sub-processes run inside a flexible case, so routine fragments stay automated while the overall journey stays adaptive.

Adaptive Case Management vs BPMN: When to Use Each Approach

The choice between adaptive case management and BPMN is a choice about predictability. Model what is stable; leave open what is not. The comparison below summarizes how the two approaches diverge across the dimensions that matter for architecture and governance decisions.

Dimension	BPMN (Structured Process)	Adaptive Case Management
Work type	Repeatable, high-volume, predictable	Knowledge-intensive, discovery-driven, variable
Modeling style	Procedural: sequence flows, gateways, events	Declarative: goals, stages, conditions, milestones
Who decides the next step	The process engine, per the model	The case worker, within governed boundaries
Path visibility	Known end to end before execution	Emerges as the case unfolds
Change tolerance	Low: model changes require redeployment	High: plans adjust per case at runtime
Best-fit examples	Order fulfillment, onboarding, invoice approval	Investigations, legal matters, care plans, complex claims
Supporting OMG standard	BPMN 2.0 (January 2011)	CMMN 1.0 (May 2014), CMMN 1.1 (December 2016)

A practical selection rule follows from the table:

• Choose BPMN when at least 80 percent of instances follow the same steps and volume justifies full automation.
• Choose adaptive case management when the next step depends on findings, expert judgment, or external events you cannot enumerate in advance.
• Combine both when a flexible case wraps stable fragments — the pattern the OMG formalizes through its BPM+ family of BPMN, CMMN, and DMN standards.

Moreover, hybrid designs dominate real deployments. A claims case may invoke a BPMN payment sub-process; a BPMN onboarding flow may spawn a case when an anomaly appears. The standards were designed to interoperate, not compete.

The Case File: One Shared Working Context, Not a Predefined Flowchart

The case file is the heart of adaptive case management. A case file is the single, shared container that holds everything about a piece of work: structured data, documents, communications, tasks, deadlines, decisions, and the complete audit history. It gives every participant the same current picture of the case, at any moment, without asking anyone.

This is the decisive architectural difference from workflow systems. In BPMN, data rides along as process variables serving the flow. In ACM, the flow serves the data: the case file persists across every task, participant, and phase, and it outlives any individual activity. Workers coordinate through the shared context instead of through handoffs, which directly attacks the information-search losses documented by Atlassian and Coveo.

A well-designed case file typically contains:

• Parties and roles — the customer, patient, claimant, or subject, plus every worker touching the case.
• Documents and evidence — contracts, medical records, photographs, statements, each versioned and access-controlled.
• The task list — mandatory steps, planned discretionary activities, and ad-hoc tasks added along the way.
• Timeline and audit trail — who did what, when, and why, including decisions not to act.
• Milestones and deadlines — statutory clocks, service-level targets, and review gates.

On modern low-code platforms this container is assembled from data models rather than custom code. Teams building on Informat, an AI-powered low-code development platform, for example, compose a case file from linked data tables, document attachments, and role-based views, so the shared context exists on day one rather than after a year of bespoke development.

Document-Centric Collaboration Inside the Case File

Much knowledge work is really document work: reviewing evidence, annotating drafts, comparing versions, assembling submissions. Adaptive case management treats documents as first-class case citizens with version history, annotations, and chain-of-custody metadata. As a result, a new participant can join a two-year-old case and reconstruct its state in hours, not weeks — and auditors can verify exactly which document version informed which decision.

Ad-Hoc Tasks and Discretionary Activities: The Building Blocks of Case Management

Adaptive case management does not abandon structure; it stratifies it. Every case blends three categories of work, and the discipline lies in classifying them honestly rather than pretending everything is mandatory.

• Mandatory steps are non-negotiable activities required by law, regulation, or policy — a fraud referral above a threshold, a privacy notification, a clinical safety check. The system enforces them regardless of worker preference.
• Discretionary activities are predefined options the case worker may activate when judgment warrants — order an independent medical examination, request a forensic accounting review, schedule a witness interview. They are modeled in advance but selected at runtime.
• Ad-hoc tasks are created on the fly for situations nobody anticipated. The worker defines the task, assigns it, and the case file records it with full auditability.

The CMMN standard formalizes the discretionary layer through a construct called the planning table: a curated menu of activities the worker can pull into the live case plan. Guardrails still apply — entry and exit criteria (called sentries) can make a discretionary task available only after a triggering event, and high-impact selections can require four-eyes approval.

Consequently, discretion becomes governable rather than invisible. When a claims adjuster adds a subrogation review, the choice is logged, timestamped, and attributable — a sharp contrast with the untracked judgment calls that live in email today. ACM does not remove human judgment; it makes judgment auditable.

The Adaptive Case Lifecycle: Open, Plan, Execute, Review, Close

Even when the path through a case is unpredictable, the case itself follows a recognizable lifecycle. Adaptive case management platforms structure this arc explicitly, which is how they deliver governance without dictating sequence. The canonical lifecycle runs through five phases:

1. Open. Intake captures the triggering event — a claim, a referral, a complaint — classifies its type and severity, and instantiates the case file with initial data and mandatory obligations.
2. Plan. The assigned worker reviews the facts, selects relevant discretionary activities from the planning table, sets milestones, and establishes the initial strategy for the case.
3. Execute. Tasks run, evidence accumulates, and the plan is revised continuously. Planning and execution interleave: each finding can add, remove, or resequence work.
4. Review. Quality gates verify completeness and compliance — supervisor sign-off, peer review, or automated checks that every mandatory step closed properly.
5. Close. The outcome is recorded, the parties are notified, and the full case history is archived for audit, appeal, and analytics.

The loop between plan and execute is what makes the model adaptive, and closed cases feed organizational learning: recurring ad-hoc tasks become tomorrow's discretionary templates. Academic evidence supports the approach — a study presented at the Business Process Management 2025 workshops and published by Springer in 2026, comparing BPMN with declarative rule-based ACM methods, found that the ACM approach delivered significantly higher execution flexibility, greater knowledge-worker autonomy, and lower dependence on IT for change requests.

Dynamic Role Assignment and Rules Engines in Adaptive Case Management

Two mechanisms keep an adaptive case moving without collapsing into chaos: dynamic role assignment, which continuously matches cases to people, and rules engines, which inject policy without prescribing sequence. Together they answer the two operational questions every case organization faces — who should work this, and what must never be missed.

Who Is the Best Person for This Case Right Now?

In routine workflow, assignment is static: a queue owns a step. In adaptive case management, assignment is a live optimization. The right owner for a case can change as the case reclassifies — a routine claim becomes suspected fraud, a standard matter becomes cross-border litigation. Platforms therefore route and re-route using signals such as:

• Skills and credentials — licensure, certifications, language, jurisdictional authority.
• Current workload — active caseload and remaining capacity, not just availability.
• History and continuity — prior involvement with the same party or matter.
• Seniority thresholds — case value or risk bands that mandate senior review.
• Conflict screening — automatic exclusion of workers with declared conflicts of interest.

How Do Rules Engines Guide Case Workers Without Constraining Them?

Rules engines in ACM behave like guardrails, not rails. Decision models — often expressed in the OMG's Decision Model and Notation (DMN) — watch the case data and react: a payment above a threshold triggers a mandatory review task; a missing document blocks closure; an approaching statutory deadline escalates priority. The worker still chooses the route; the rules guarantee the boundaries.

Increasingly, AI extends this guidance layer. Camunda's whitepaper on orchestrating case work, summarized in its December 2025 blog on the "messy middle" of case work, describes AI agents that determine candidate tasks at runtime under full human oversight and audit trails. This convergence of cases, rules, and agents is the same trajectory driving hyperautomation and AI-powered workflow automation across the enterprise — automation proposing, humans disposing.

The CMMN Standard: Formal Notation for Adaptive Case Management

Case Management Model and Notation (CMMN) is the OMG's formal standard for modeling cases. The OMG published CMMN 1.0 in May 2014 and revised it as CMMN 1.1 in December 2016. Where BPMN draws a route, CMMN draws a territory: it declares what work exists, under which conditions it becomes available or required, and which milestones mark progress.

The notation's core vocabulary maps directly onto the concepts covered above:

• Case plan model — the overall boundary of the case, drawn as a folder.
• Stages — nestable groupings of related work, activated by conditions rather than sequence.
• Tasks — human, process (calling BPMN), decision (calling DMN), or case tasks (spawning sub-cases).
• Discretionary items — dashed-outline tasks available through the planning table.
• Sentries — entry and exit criteria that arm tasks and stages when events or conditions fire.
• Milestones and event listeners — achieved states and reactions to timers or user actions.

In regulated healthcare, the OMG's BPM+ Health initiative applies BPMN, CMMN, and DMN together to encode clinical pathways — evidence that the layered-standards model works in the most safety-critical knowledge work there is. Vendor support is exemplified by Flowable's production CMMN engine, which runs case models natively alongside BPMN and DMN.

Is CMMN Widely Adopted Compared to BPMN?

No — CMMN adoption remains narrower than BPMN's, and honesty about this helps architects plan. Camunda notably dropped CMMN support from its platform, arguing in a July 2023 engineering analysis that BPMN constructs such as ad-hoc sub-processes can replicate most CMMN patterns while remaining clearer at scale. Meanwhile, Flowable and several regulated-industry platforms continue to invest in native CMMN. The pragmatic verdict: the ideas of CMMN have outlived arguments about its diagrams — case files, discretionary tasks, sentries, and milestones now shape products whether or not the notation itself is used.

Adaptive Case Management in Action: Investigations, Legal, Healthcare, and Claims

Adaptive case management earns its keep in domains where discovery drives the work. Four patterns recur across industries, and each illustrates a different balance of mandatory structure and professional discretion. These are also the domains where enterprise digital transformation strategies most often stall when teams apply flowchart thinking to case-shaped work.

Investigations and Fraud Casework

An investigation is the purest case: each piece of evidence determines the next request. A suspicious transaction may spawn account reviews, interviews, subpoenas, or referral to law enforcement — or exonerate the subject in a day. ACM supplies what investigators need most: chain-of-custody document handling, statutory clocks as mandatory milestones, and an audit trail that survives courtroom scrutiny. Discretionary task libraries encode investigative tradecraft without forcing every case through it.

Legal Case Management and Matter Lifecycles

Legal matters combine externally imposed structure — court deadlines, filing rules — with internally driven strategy. Adaptive case management models the matter as a case file holding pleadings, discovery, correspondence, and privilege logs, while sentries arm tasks when the court acts. Conflict screening runs at assignment, and discretionary research or motion tasks let counsel adapt strategy motion by motion without losing firm-wide visibility.

Patient Care Coordination

A care plan is a living case: test results, specialist input, and patient response continuously reshape it. Mandatory elements — medication interaction checks, consent, discharge criteria — coexist with clinical discretion over referrals and therapies. The BPM+ Health work cited above exists precisely because rigid pathways fail heterogeneous patients; multidisciplinary teams need one shared record of decisions and rationale across settings and months.

Complex Claims Adjudication

Insurers increasingly split claims by predictability: straight-through BPMN processing settles the simple majority in minutes, while contested, high-value, or suspicious claims open as adaptive cases. Adjusters then draw on discretionary activities — independent medical examinations, engineering reports, special investigation unit referrals — as facts warrant. The common thread across all four domains is consistent:

• One shared context replaces scattered email, spreadsheets, and file shares.
• Discovery drives sequencing, so plans are revised, not violated.
• Expert judgment is exercised inside guardrails, not around them.
• Every action and inaction is auditable for regulators, courts, and quality teams.

Building Adaptive Case Management on Modern Low-Code Platforms

The tooling market reflects steady demand. The global BPM-platform-based case management software market was valued at 1,585 million US dollars in 2025 and is projected to reach 2,411 million US dollars by 2032, a 6.2 percent compound annual growth rate, according to Reports and Markets research published for the 2026 cycle. Established vendors include Pegasystems, Appian, Hyland, IBM, Newgen Software, Flowable, and Camunda, as cataloged in SS&C Blue Prism's overview of adaptive case management.

Low-code platforms have become a natural home for ACM because the case file is, at bottom, a data model — and data modeling is what low-code does fastest. On Informat, for instance, a team can stand up linked case tables, role-based task views, document attachments, and escalation rules in days, then iterate as case types evolve — the build-measure-adjust economics examined in our analysis of low-code ROI and enterprise value in 2026. A pragmatic implementation sequence looks like this:

1. Model the case file first. Define parties, data, documents, and milestones before touching any flow.
2. Codify only the truly mandatory steps. Resist the urge to script preferences as requirements.
3. Build a discretionary task library. Interview senior workers and encode their repertoire as optional activities.
4. Add routing and escalation rules. Automate assignment signals and deadline guardrails.
5. Measure outcomes, not step compliance. Track cycle time, quality, and rework — then template what works.

What Is the Difference Between Adaptive Case Management and Workflow Automation?

Workflow automation executes a known sequence of steps with minimal human intervention; adaptive case management coordinates unknown sequences with maximal human judgment. Workflow answers "how do we do this faster?" while ACM answers "how do we handle what we cannot predict?" Most enterprises need both — automation for the routine 80 percent, adaptive cases for the consequential 20 percent where expertise decides outcomes.

Conclusion: Structuring Unpredictable Knowledge Work with Adaptive Case Management

Adaptive case management resolves a false choice that has constrained BPM for two decades: total scripting or total chaos. By organizing work around a shared case file, separating mandatory steps from discretionary and ad-hoc activities, assigning the right person dynamically, and letting rules guard boundaries instead of dictating routes, ACM gives unpredictable knowledge work the structure it can actually live with. The CMMN standard — whether adopted as notation or absorbed as concepts — supplies the formal vocabulary, and hybrid designs pair adaptive cases with BPMN fragments and DMN decisions.

For leaders, the takeaways are concrete:

• Classify work honestly. If the next step depends on what you discover, it is a case, not a process.
• Invest in the case file. Shared context is the highest-leverage cure for the 25 percent of the week lost to searching.
• Govern discretion; do not delete it. Auditable judgment beats both rigid scripts and invisible workarounds.
• Start small on flexible tooling. Low-code platforms such as Informat make the first adaptive case type a weeks-long project, not a multi-year program.

Drucker called knowledge-worker productivity the biggest management challenge of this century. Adaptive case management is, so far, the most credible systems answer to it — structure for the unpredictable, freedom within the frame, and a complete record of how skilled people actually got things done.