AI in established work processes: A critical perspective
A pattern emerging in organizations
A pattern keeps repeating itself in organizations when discussing AI use cases. The range of possibilities is extensive: AI for summaries, report generation, customer communication, research, presentation creation, knowledge management, and forecasting. The list of identified applications keeps growing.
But look closer at this expansion and a more fundamental question surfaces:
Are these use cases actually solving the relevant problems?
Many of them are demonstrably useful. Yet a systemic problem exists beneath the surface. AI is being deployed to optimize individual activities within work processes that themselves go unquestioned. The result is acceleration of existing ways of working. Not redesign.
That distinction matters more than it might appear.
The research paradox
Available research data points to a significant gap. McKinsey's "State of AI 2025" examined organizational practices and found something striking: workflow redesign showed the strongest correlation with organizations reporting measurable business impact from generative AI.
Consider that.
Yet only about 21 percent of surveyed organizations using generative AI reported that they had fundamentally redesigned their work processes. A follow-up survey in July 2026 confirmed the pattern persisted. Around 79 percent of organizations continued to skip workflow redesign entirely.
The contradiction is remarkable. It speaks for itself: Organizations can execute individual tasks faster. But the underlying work practice remains untouched. And the research says this is precisely where the impact lives.
The customer proposal: A concrete example
A typical workflow for creating a significant customer proposal follows an established sequence:
- Information gathering and research
- Review of past projects
- Team briefing
- Developing a storyline
- Draft creation
- Document circulation
- Numbers and formatting review
- Leadership comments
- Revisions and sign-offs
- Final delivery
Enter AI systems and here is what happens: individual steps accelerate. Research becomes faster. Meeting notes are automated. Text generation gets support. Images are generated. Formatting errors are corrected. The number of AI use cases multiplies.
But the workflow itself remains structurally unchanged. Same people. Same handover points. Same approval structure. Same output. Same sequence.
It is like installing a high-performance engine in a system originally designed for far lower capacity.
Redesigning from the goal backwards
A different approach would be to start from what actually matters: developing the best possible solution to the customer's problem. Working backwards from that result, processes could be fundamentally restructured:
- Internal knowledge could be continuously available rather than manually researched
- External research could be conducted in parallel, compared, and critically questioned
- AI could identify gaps in reasoning during the thinking process itself
- Different hypotheses could be tested and validated before any commitment
- Experts could focus more intensely on the customer's actual problem and challenge assumptions
- AI could adapt content as new insights emerge
- Quality assurance could happen continuously, not only at the end
- Leaders could spend less time on editorial details and more on whether the recommendation is sound
- The final presentation becomes a byproduct of a better thinking and working process
This is fundamentally different from simply adding AI to existing processes. This is a workflow designed around the complementary strengths of AI and human work.
Workflow redesign across funtions
This principle extends across organizational functions. The pattern is always the same: stop asking whether AI can accelerate the current step. Start asking whether the step should exist at all.
Customer Service: Instead of asking whether AI agents can answer faster, consider the deeper question: Why did the customer need to contact you in the first place? Can AI anticipate problems? Can it partially solve them before the customer notices? Which tasks should an AI system handle entirely, and when is human intervention actually valuable?
Management Reporting: Instead of asking whether AI can generate monthly reports faster, ask this: Why wait until month-end to create static reports about processes an AI system could monitor continuously, interpret in real time, and escalate when needed?
Talent Acquisition: Instead of asking whether AI can write better job descriptions, ask: If AI is changing the work itself, are you actually hiring for the right roles?
These questions are harder. They also reveal where the real opportunity lies.
The harder questions that come first
Workflow redesign is fundamentally more disruptive than isolated use-case optimization. It forces organizations to confront questions they often prefer to avoid:
- Which activities should disappear entirely? Not everything should be automated. Some activities should simply stop existing.
- Which decisions can AI make without repeated approval? If every execution requires sign-off, most of the benefit evaporates.
- Where is human judgment genuinely valuable? In ambiguity. Empathy. Responsibility. Problem definition. Ethics. Trade-offs. Creativity. Context understanding.
- Where should humans actually be in the process? This requires intentional design, not default inclusion.
These questions have no comfortable answers. But they are where real transformation begins.
Human AI collaboration: A design problem, not an assumption
The concept of "human involvement" is often treated as automatically safe and optimal. The available research draws a more nuanced picture. A meta-analysis in Nature Human Behaviour by Vaccaro, Almaatouq, and Malone examined 106 experiments with 370 effect sizes. Their finding: on average, human-AI combinations outperformed humans alone. But they underperformed the better performer between human or AI working independently. Results differed substantially by task type. Collaboration benefited creative work. For decision-making tasks, it offered less advantage.
This is a critical finding. Human-plus-AI is not a strategy. It is a design problem. Depending on context, these divisions make sense:
- Human decides. AI provides information.
- I decides. Human checks exceptions.
- Both work together.
- AI executes the task alone.
- Humans deliberately do the work themselves to build expertise.
The goal is not maximum automation and not maximum human involvement. The goal is the best overall system.
Redefining Sucess Metrics
This leads to a complete reset of how success is measured:
Not: Does AI save eight minutes writing an email?
But: Did the overall five-day process compress to two days?
Not: How many AI-generated documents were created?
But: Did a decision that took three weeks become a three-day decision?
Not: How many prompts were used?
But: How did rework change?
Not: How many licenses were purchased?
But: How did customer outcomes improve?
Not: How many hours were saved?
But: What became possible with the freed capacity?
AI as a strategic tool
The strategic value of AI does not emerge from marginal acceleration of existing work. It emerges from one fundamental question: Why does this work take this form at all?
That question has far-reaching consequences. Once enough workflows have been fundamentally redesigned, the actual work content changes for employees. Decision authorities shift. Required skills evolve. Some handover processes vanish entirely. Leaders direct different functions. Agents assume tasks that once defined a role.
At that point, this is no longer workflow redesign alone.
This is organizational transformation.
Author’s note: The observations in this article draw on recurring patterns from our work with organizations navigating AI transformation, combined with publicly available research. Client situations have been generalized and anonymized; no individual organization or engagement is being described.
Sources: McKinsey State of AI 2025 and Operating Model Advantage 2026; Vaccaro, Almaatouq & Malone, Nature Human Behaviour, “When combinations of humans and AI are useful.”
