The agentic enterprise: rethinking operations through 'Jobs to be Done'

In the race to operationalise AI, many enterprises are making a costly strategic error: applying a new generation of technology to workflows designed during past digitalisations.

As executives authorise heavy investments in agentic AI, a concerning pattern is emerging. Organisations are falling into the “digitalisation trap”, taking legacy processes — often bloated with years of operational debt and systemic inefficiencies from previous IT transformations — and simply bolting on AI to execute those same flawed steps slightly faster.

This is a single-loop transformation that yields marginal productivity bumps but fails to deliver structural business value. If your underlying operating model is constrained by the limitations of past digital rollouts, accelerating it with AI won’t magically produce a modern, forward-thinking enterprise; it often just creates faster bottlenecks.

To unlock the transformative ROI promised by agentic technology, leaders must shift their mandate.

We must stop asking: “How can AI automate our current process?” and start asking, “How can agentic AI help us achieve our core business objectives entirely differently?”

The strategic pivot: returning to ‘Jobs to be Done’

Breaking free from superficial digitalisation requires abandoning rigid process maps and returning to first principles. For leadership, the ‘Jobs to be Done’ (JTBD) framework is not just a design concept; it is a critical strategic lens for evaluating AI investments.

Consider Harvard Business School professor Clayton Christensen’s seminal “milkshake” study. A fast-food chain boosted sales not by changing the milkshake itself, but by understanding the “job” customers “hired” it to do: a satisfying, long-lasting breakfast on a boring commute. They stopped focusing on product features and designed for the desired outcome.

Similarly, in the enterprise, the JTBD framework forces organisations to look past how work is currently done. By clearly defining the core “job” required by the customer or business, leaders can use agentic AI — systems capable of autonomous reasoning and multi-step planning — to architect entirely new, effective operating models. Agentic workflows solve problems autonomously. By framing operations around the Job to be Done, leaders liberate agentic technology from legacy constraints, allowing it to drive impact precisely where it matters most.

The anatomy of a job: unpacking the universal steps

To effectively apply the JTBD lens, it’s crucial to understand the inherent structure of any “job.” Regardless of the specific task, a customer’s journey generally follows this eight-step universal process, forming a ‘Job Map’:

1. Define: Figuring out what needs to be done.
2. Locate: Gathering the necessary inputs, data, or items.
3. Prepare: Setting up the environment or tools to execute the job.
4. Confirm: Verifying that everything is ready to proceed.
5. Execute: Performing the core task.
6. Monitor: Tracking the execution to ensure it is going well.
7. Modify: Making adjustments or troubleshooting if needed.
8. Conclude: Finishing the job and assessing the outcome.

To truly unlock agentic potential, however, leaders must go beyond merely mapping these steps. Each step needs to be dissected using specific analytical primitives (or core components) to uncover the granular areas of opportunity where agentic AI can deliver maximum impact:

• Functional primitives: The explicit, mechanical tasks taking place in each step — where traditional automation often focuses.
• Emotional primitives: How the customer or internal team wants to feel or be perceived while doing the job (e.g., in control, anxious, relieved) — crucial for designing human-centric agentic experiences.
• Consumption primitives: What makes the process of doing the job easy or difficult (e.g., time to complete, physical effort, cost) — identifying areas for agentic efficiency gains.
• Pain points: Metrics or specific bottlenecks that prevent the customer from executing a step successfully — the critical targets for agentic problem-solving.

By understanding these primitives within each step of the job, leaders can pinpoint exactly where current systems fail and where autonomous agents can be designed to eliminate friction, enhance satisfaction, and drive efficiency.

Why leaders must embrace the Job Map for agentic AI:

• Unlocks innovation: By focusing on the “what” instead of the “how,” the job map reveals precisely where current tools and processes fall short, showing leaders exactly where to deploy agentic AI for maximum disruptive impact. It moves beyond incremental improvements to true reinvention.
• Aligns teams: It gives cross-functional product and design teams a shared, unbiased language to understand the exact problem they are solving. This common ground is vital when asking agentic AI to autonomously perform complex, multi-step jobs.
• Measures success: It helps define precisely how customers measure success for each step, enabling organisations to build agentic solutions that solve for those specific outcomes, rather than simply automating existing, suboptimal metrics.

The executive mandate in action: reimagining BFSI claims

To understand the bottom-line impact of this shift, consider the claims function in the Banking, Financial Services, and Insurance (BFSI) sector.

Traditionally, claims processing is a high-friction cost centre. It is plagued by unstructured data across diverse formats, regulatory complexity, and stubbornly low Straight-Through Processing (STP) rates that drive up the cost-to-serve.

If a business leader approaches this through the lens of superficial digitalisation, they might deploy an AI tool to extract data from submitted PDFs faster than a human clerk. This addresses a single tactical bottleneck, but the rigid, multi-stage manual review process remains largely intact. The financial returns are negligible.

However, applying the Jobs to be Done framework redefines the strategic goal. The core job isn’t merely “processing a 10-page claim form”.

• The customer’s job: “Help me recover from an unexpected loss or expense as quickly and seamlessly as possible.”
• The business’s job: “Protect the company’s bottom line while retaining the customer through an efficient, accurate, and trustworthy evaluation.”

When you design for these dual outcomes using agentic technology, the operational paradigm transforms:

1. Intelligent orchestration over linear processing: Instead of a sequential assembly line of document triage, an agentic system acts as an autonomous orchestrator. It instantly ingests and synthesises multi-modal inputs — emails, damage photos, unstructured medical reports — identifying the fastest, most seamless path to resolution for the customer.
2. Real-time contextual intelligence: To protect the bottom line, the agent actively queries core enterprise systems. It autonomously cross-references policy coverage, analyses historical data for fraud signals, and assesses claim validity accurately and instantly.
3. Exponential growth in straight-through processing: By removing the reliance on rigid, predefined steps, agentic workflows can confidently evaluate clear-cut cases end-to-end. This drastically drives up STP rates, triggering automated, trustworthy payouts and significantly reducing operational overhead, fundamentally transforming the ‘Execute’ and ‘Conclude’ steps of the job map by making them highly efficient and automated where appropriate.
4. Elevating the claims processing team: Rather than paying claim experts to manually review routine documents, the system only escalates high-complexity or ambiguous edge cases. The claims processing team is presented with an agent-synthesised briefing of the incident and recommended actions, allowing them to focus strictly on high-value cognitive work, fraud prevention, and human empathy. This shift empowers the team to excel in the ‘Monitor’, ‘Modify’, and critical ‘Conclude’ activities, fulfilling the emotional primitive of being in control and effective, and addressing the pain point of being bogged down by routine tasks.

The bottom line

For enterprise leaders, the mandate is clear. Stop treating AI as a tactical upgrade for legacy processes built during past digitalisations. By coupling agentic technology with the Jobs to be Done framework, organisations can execute a true double-loop transformation. The result is not just a faster enterprise, but a structurally agile, highly cost-effective, and customer-centric business capable of fundamentally outmanoeuvring the competition.