When AI drafts the reports, plans or recommendations a team used to produce, the team can deliver more in less time. The same change can remove the parts of the work where people took initiative, built skill or learned from customers. A role can stay useful to the company while giving the person in it little they consider worth doing, and productivity figures will not show that.
The Octalysis Group treats this as a behavioral design problem. We identify the actions a company needs from its people, examine what motivates different employees to carry them out, and design the choices, challenges, feedback, progression and social interactions that support those actions. For some people automation removes only tedium. For others it removes the reason they came to work.
Usefulness to the company and work worth doing are separate questions
A role is useful when its output matters to the business. A role is worth doing, for a particular person, when the activity itself gives them something they care about. The two often coincide, which is why the difference is easy to miss until automation separates them.
Consider a market analyst whose reports are now drafted by an AI system from the same data sources. The company still needs someone to check the reports, route them to stakeholders and answer questions. Whether that work is worth doing depends on what the analyst liked about the old job. If most of the old job was cleaning inconsistent spreadsheets, the analyst may be relieved to see it go, and there is no reason to pretend the manual effort had value it never had. If the part they cared about was the moment an unexpected pattern showed up in the data, removing the report-building may also have removed their route to that moment. Promoting them to coordinate AI output across three teams would make them more useful to the organization. For this analyst, it could also take away the part of the job they came to work for.
As we have argued about AI user engagement, capability alone does not give people a reason to care about the work around it. Inside a team, that problem shows up one person at a time.
Self-Determination Theory helps explain why. Its overview of the theory describes autonomy, competence and relatedness as basic psychological needs, and argues that social contexts supporting them produce more volitional, higher-quality motivation. For role redesign, the practical implication is that a new role should be checked against those needs as well as against business requirements. If the coordination role above leaves the analyst little say over how the work is done and little chance to use the skill they are proud of, the theory gives reason to expect weaker motivation, even though the role is useful.
What employees valued in the work AI took over
Asking what someone valued in a task usually produces a concrete answer. The analyst above valued finding patterns nobody had asked about. An account manager might say they valued being the person a customer called when something broke. Another person might say they valued nothing in particular and were glad to be rid of the work. These answers matter because they point to different redesigns. A role that restores contact with customers does little for someone whose satisfaction came from investigation, and a role full of open-ended investigation does little for someone who wants to be relied on by a specific person.
In Octalysis terms, the difference often shows up in which Core Drives the old work engaged. When a person tries an investigative approach, sees what it produces and adjusts, that loop runs on Core Drive 3, Empowerment of Creativity & Feedback. Curiosity alone does not create that loop. It needs a strategy the person chooses and feedback that responds to the choice. When someone overcomes a real challenge and can see their competence increasing, that is Core Drive 2, Development & Accomplishment. When the work centers on being relied on by a customer or a colleague, Core Drive 5, Social Influence & Relatedness, is doing much of the work. A redesign that restores one of these while the person valued another is likely to miss.
Automation does not reliably take the routine work and leave humans the judgment. In many roles AI now drafts the recommendation, weighs the options and proposes the decision. The design question is specific to each role: which decisions still depend on this person’s knowledge, relationships or accountability, and how do we make those visible?
Ownership grows through what a person puts into the work
One response to automation is to expand responsibility: you no longer write the plans, so now you own the account. Assigning ownership does not by itself create the feeling of ownership. Core Drive 4, Ownership & Possession, includes the feeling that a process or project is yours, and in work that feeling usually comes from investment. People come to own an approach they shaped, an idea they argued for, and a contribution others can still see is theirs.
Not everyone wants more responsibility. Some people want depth, and an investigator may find a management track a demotion in everything but pay. Asking what people valued is only useful if the answer can change the role they end up with.
A hypothetical example: account plans generated by AI
Suppose a company uses AI to generate account plans from CRM data, public information and past correspondence. The plans used to take each account manager several days of research. Now they take minutes, and they are competent.
Tom is one of the account managers. When his manager asks what he valued in the old plan work, his answer is about one relationship: the operations lead at a customer who depended on him. The research days had been where he learned how that customer actually worked.
The customer runs order processing through a system Tom’s company supplies. The generated plan for the account recommends a training program for the customer’s order-entry staff, based on support tickets that show frequent errors in submitted orders. The recommendation is reasonable on the data available to the system. Tom calls the operations lead before proposing it. In that conversation he learns that the errors cluster around orders above a certain value, which require a second approval the software does not support. Staff export those orders to a spreadsheet, collect the approval by email and re-enter the order by hand. The errors come from re-entry. Training would not change the approval step that causes them.
Tom proposes a limited pilot: configure an approval step for high-value orders at one of the customer’s sites and leave the other sites unchanged for comparison. Before the pilot, he writes down what he expects: the spreadsheet workaround should disappear at the pilot site and re-entry errors should fall there. Six weeks later the operations lead tells him the workaround is gone at that site, while a smaller category of errors he had not predicted remains, from orders split across two delivery dates. That feedback is specific. It tells Tom which part of his diagnosis was right, which part was incomplete, and what to investigate next. It also gives his manager something concrete to recognize: Tom found a cause the generated plan missed and tested a fix the customer could evaluate. Praising the account plan itself would have credited the system for the part of the work Tom did not do, and it would have told the team that management could not tell the difference.
Tom’s redesigned role builds on this. It moves the time he used to spend on research into direct contact with fewer customers, with authority to propose configuration pilots and to bring in a product specialist without escalation. His progress is measured by the complexity of the customer problems he resolves and how well his predictions hold up.
He did not get that role simply by asking for it. His manager had to see evidence that he could diagnose problems like this one. The customers in his portfolio had to need that kind of attention, and the team had to have capacity to cover the accounts he gave up. The role was negotiated between what motivates Tom and what the business needs.
Two colleagues who lost the same task show why the answer to “what did you value?” matters. Priya valued depth. The research was where she noticed a customer’s new regulatory exposure or a changed buying committee. Her role is built around the accounts where generated plans are least reliable, such as customers going through mergers, and around explaining to the team configuring the system where and why the plans fail. She can inspect the changes that team makes in response to her recommendations and discuss them with them, and her feedback comes from whether later plans improve on the flaws she flagged. Sam valued very little in the research and says so. For Sam, the automation removed tedium, and Sam wants broader scope and eventually a coaching role. Each of those routes was also negotiated against evidence of ability, customer needs and the team’s capacity, and the combined roles still had to cover every account.
Job crafting explains why negotiated scope matters
Amy Wrzesniewski and Jane Dutton’s conceptual paper on job crafting in the Academy of Management Review proposes that employees shape their jobs by changing task boundaries, relational boundaries and the cognitive boundaries of how they think about the work, and that these changes alter the meaning of the work and the person’s work identity.
In the example, Priya’s change is mainly on the task boundary. Tom’s is largely relational, since it changes whom he works with and how closely. Cognitive crafting is a different kind of change: a person reinterprets what their existing work means, without the tasks or relationships themselves changing. Nothing in the example shows that kind of shift, and a manager should not assume one just because a role has been redesigned. When automation removes a task, it can shift any of these boundaries without anyone deciding to. A deliberate redesign gives people a chance to rebuild them in a way that still produces what the organization needs, and the negotiation is where both sides check that it does.
Relationships at work need occasions
Automation can also remove occasions when colleagues used to rely on each other. If account managers once compared research notes before renewals, generated plans may take that exchange away. Our own team faced a related problem when working remotely. In our REMOTE Framework we describe how The Octalysis Group wrote responsiveness into its shared norms: people are encouraged to call a colleague directly instead of trading messages for a week, and career development is assessed on communication and empathetic output alongside performance output. The context was distance, not AI, but the implication carries over. If the occasions for mutual reliance disappear, they have to be designed back in and rewarded. Tom bringing a newer colleague onto a call with the operations lead, or Priya reviewing a colleague’s approach before a difficult renewal, puts that reliance into the work itself. This is employee engagement design concerned with what people do together.
What motivation design cannot fix
Pay, job security and workload decide whether any of this is credible. If automation is followed by job cuts, or if the remaining people are asked to absorb an unmanageable workload, employees will reasonably read talk of meaningful work as cover. Motivation design works inside a fair deal and cannot substitute for one.
Designing AI-enabled work people want to do
A company that invests in AI capability without designing the human behavior needed to turn it into business results takes a commercial risk. A sales team can generate account plans in minutes and still lose the customer diagnosis, initiative and learning that made those plans worth acting on.
We help companies design how people work with AI so that greater output does not come at the expense of initiative, learning or customer relationships. Using Octalysis, we identify the employee behaviors the business needs, examine whether the new workflow gives people their own reasons to carry them out, and design the choices, challenges, feedback, progression and social interactions that support those behaviors. For a sales team working with generated account plans, that could include a defined role in investigating customer problems, room to test approaches with customers, direct customer feedback on the results, and recognition for solutions colleagues can reuse.
If your teams are adopting AI and you want to know whether the new workflow still produces the behaviors your business depends on, Contact Us with a description of the roles involved and the actions you need from the people in them.
