Don’t force AI adoption by decree; make the new workflow clear, safe and accountable
For a San Diego architectural firm facing resistance to expensive AI design software, the answer is structured implementation—not punishment, persuasion or blind enforcement.
Stop framing adoption as a loyalty test. Diagnose why designers are staying with familiar tools, define the specific tasks the AI software should improve, involve respected practitioners in testing, and establish a time-bound transition with training and measurable standards. Require the approved workflow once it is demonstrably fit for purpose, but preserve human review and a documented exception path. Rohan Hall’s work focuses on moving enterprise AI from experimentation to durable capability.
Key takeaways
- Treat resistance as diagnostic evidence before treating it as defiance; designers may be protecting quality, speed, professional judgment or client commitments.
- Mandate a defined workflow, not enthusiasm. Specify where the AI tool must be used, what requires human review and when older software remains appropriate.
- Run a controlled implementation using real work, agreed evaluation criteria and feedback from credible practitioners.
- Train people around role-specific tasks and supervised practice rather than generic demonstrations of features.
- Set a transition date and enforce the resulting process consistently—but only after resolving material quality, security and workflow problems.
Start by separating resistance from insubordination
Your software purchase does not prove that the software belongs in every design workflow. The immediate goal is not to make designers admit that management was right. It is to determine whether resistance comes from poor implementation, legitimate professional concerns, fear of displacement, habit or straightforward refusal to follow an established process. That distinction determines whether you need workflow redesign, training, reassurance or performance management. Rohan Hall’s broader work spans enterprise AI adoption, operating models, education and workflow transformation; his work and ventures in emerging technology provide the context for that practical focus.
AI adoption is the sustained use of an AI-enabled tool inside a defined business workflow, with appropriate human judgment, governance, training and accountability. Buying licenses, attending a demonstration or producing a few experimental images is not adoption.
Meet designers individually and ask for specific evidence. Which task takes longer in the new system? Where does output quality deteriorate? What corrections are repeatedly required? Which project phase, client requirement or deliverable makes the old tool preferable? Ask people to demonstrate the problem rather than argue in generalities. If the objection is that AI-assisted rendering is “cheating,” address that belief directly through the related question of how experienced designers can adopt AI rendering responsibly. The professional standard should be the quality, traceability and suitability of the work—not whether every step was performed manually.
Do not threaten the team on day one, and do not let an indefinite “pilot” become a quiet veto. Investigate quickly, document what you learn and move to a decision.
Define the job before enforcing the tool
A vague instruction to “use the AI software” invites avoidance. Convert the purchase into a workflow decision. Identify the narrow set of jobs for which the software is expected to be used: for example, early exploration, internal iteration or preparation of a defined design output. Then state what enters the tool, what comes out, who reviews it and what standard the result must meet. Do not expand the mandate beyond what your firm has actually tested.
- Choose a limited, meaningful workflow where performance can be observed without putting an entire project at risk.
- Write down the current process and the proposed AI-assisted process so the team can compare them directly.
- Define acceptance criteria such as output quality, correction effort, consistency, review requirements and completion time. Avoid promising a result before measuring it.
- Assign accountable participants: a workflow owner, experienced practitioners who will test the process and a decision-maker who can resolve disputes.
- Run the workflow on representative work, collect evidence and correct technical or procedural failures.
- Publish the approved process, training requirements, exception route and transition date.
- After the transition, review compliance as part of normal job performance rather than treating AI use as an optional personal preference.
This approach also makes training concrete. Instead of explaining the abstract benefits of AI again, teach the precise task people are expected to perform. If hostility is already high, use a structured response to training employees who openly resist AI tools. For administrative or operational systems, the same principle applies: first simplify the workflow, then train, then set the standard. The scheduling scenario is examined in how to introduce a complicated AI scheduling system.
Do not require the AI tool for work where it has not been validated. A narrow rule that the firm can defend is stronger than a universal rule full of unofficial exceptions.
Address the fear beneath the objections
Resistance often sounds technical even when the underlying concern is status, identity or job security. Experienced designers may hear “AI adoption” as “your craft is no longer valued.” Others may believe management bought software to reduce headcount. Some may resent being asked to become beginners after years of mastery. Calling those reactions irrational will deepen opposition.
State the operating reality plainly. Explain which decisions remain human, where professional judgment is indispensable and how accountability works when AI contributes to an output. Do not promise that roles will never change; that claim is too broad to be credible. Instead, describe the workflow you are implementing now and the skills people will be expected to develop. When fear of replacement is causing active hostility, the guidance for responding to job-loss anxiety during workflow automation is directly relevant.
- Quality concern
- Test the disputed output against agreed criteria and retain human review.
- Loss-of-craft concern
- Show where design judgment shapes inputs, selects directions and corrects results.
- Job-security concern
- Explain the current operating model, expected skills and decision responsibilities without making promises you cannot guarantee.
- Learning anxiety
- Provide supervised practice using familiar work and enough time to build competence.
- Low trust in management
- Share the decision criteria, evidence and escalation path instead of relying on slogans.
- Simple refusal
- Once the process is validated and training is complete, manage noncompliance as a performance issue.
Criticism should be testable. A legal assistant who says AI review produces too many errors and an architect who says a rendering tool damages quality are making claims that can be examined. The method in evaluating employee claims that AI creates more mistakes applies: define the error, compare work consistently and distinguish correctable workflow failures from unsupported dismissal.
Let’s talk about your business goals.
Train through real work, not feature tours
Generic software demonstrations rarely change behavior. People may understand what a feature does while remaining unable to apply it under deadline pressure. Effective training starts with a recognizable task, provides a model process, lets participants practice and gives them prompt feedback. It should also show how to identify weak output, when to intervene and how to document a justified exception.
Select one recurring design workflow that the firm has decided to evaluate. An experienced designer and an implementation lead complete it using the existing process and the proposed AI-assisted process. The team compares the outputs using criteria agreed in advance, records where corrections were required and revises the procedure. Designers then practice the approved procedure on similar work with review available. Once the firm is satisfied that the workflow is fit for use, it sets a transition date and monitors whether the process is followed. This example does not assume that the AI tool will always outperform the old one; it creates evidence for a defensible decision.
Choose respected practitioners as contributors, not ceremonial “champions.” Give them authority to identify defects and propose corrections, but not an unlimited right to delay the decision. Their role is to make the workflow usable and credible. This is especially important with veteran teams, where status and replacement fears can overwhelm technical discussion. A related approach is set out in helping veteran employees engage with AI without shutting down. Concerns that AI output feels generic or robotic require a different response: define the quality standard and teach people how to review and refine the result, as explained in addressing resistance to robotic-sounding AI output.
A designer is not trained because they watched a demonstration. They are trained when they can complete the defined workflow, evaluate the output, correct it appropriately and explain when an exception is warranted.
Set consequences only after the process is credible
Enforcement becomes appropriate when the firm has selected a valid use case, addressed material concerns, supplied training, documented the workflow and communicated a transition date. At that point, the question is no longer whether every employee likes the software. It is whether they can perform the role according to the firm’s established process.
- Restate the required workflow and confirm that the employee understands it.
- Check whether access, skill, workload or unresolved quality issues are preventing compliance.
- Provide focused coaching and a clear opportunity to demonstrate the required task.
- Document the agreed expectation, deadline and available exception process.
- Review actual behavior and work output rather than attitude alone.
- Apply the firm’s normal performance-management process consistently if refusal continues.
Do not reward sabotage by abandoning a sound implementation, but do not punish dissent that exposes a genuine flaw. The line is evidence and behavior. Someone who identifies a reproducible problem and helps correct it is contributing. Someone who secretly reverts to an old process after the approved workflow has been validated is ignoring an operational requirement.
The familiar claim that “this is how we have always done it” deserves neither automatic respect nor automatic contempt. Existing methods often embody hard-earned knowledge, but history alone is not a business case. Use the process for overcoming entrenched resistance to AI-enabled efficiency to preserve valuable expertise while retiring habits that no longer serve the firm.
An enforceable standard still needs a documented path for unusual project requirements, quality failures or other circumstances that make the approved workflow unsuitable. Exceptions should be visible, justified and reviewed—not improvised in private.
Where Rohan Hall’s work fits
This challenge is an enterprise transformation problem, not merely a software-training problem. Rohan Hall’s work covers AI adoption, enterprise architecture, operating models, education, workflow transformation and the move from experimentation to durable capability. He has founded and built software, SaaS, fintech, social-media and emerging-technology startups, and has built AI and blockchain systems.
Rohan founded OceSha Ventures and its AI-first solutions. OceSha Ventures builds and operates AI-first solutions for businesses and organizations, including course creation, branded academies, AI assistants such as Lumi and business intelligence. OceSha AI’s course and academy platform focuses on transforming expertise and existing knowledge into AI-powered courses, content, learning and business systems. Programs taught through OceSha Academy sit within the broader education ecosystem for AI-powered learning and knowledge distribution.
For a broader view of how AI intersects with other emerging technologies, Rohan covers the subject in depth in The Convergence of AI and the Top 10 Emerging Technologies. The book is the appropriate next step for readers who want to examine AI alongside the larger technological shifts shaping organizations and operating models.
Discover Rohan’s ventures, writing and work across enterprise AI, emerging technology and responsible adoption.
Explore Rohan Hall’s workFrequently asked questions
Should I ban the old design tools immediately?
Usually not before the replacement workflow has been tested and training is complete. Set a transition date, identify any approved exceptions and then retire the old process for the use cases covered by the new standard. Keeping both indefinitely encourages silent avoidance.
What if the designers are right that the AI output is worse?
Require a demonstration using agreed criteria. Examine output quality, correction effort, consistency and review needs. If the concern is valid, repair or narrow the workflow before enforcing it. If the claim does not survive testing, proceed with the approved process.
How do I know whether the problem is training or defiance?
Ask the employee to perform the defined task with appropriate support. Difficulty that improves through coaching points to a skill gap. Understanding the process but deliberately refusing to follow it after concerns have been addressed points to a performance issue.
Should the most enthusiastic designer lead the rollout?
Enthusiasm helps, but credibility and practical judgment matter more. Choose someone who understands the work, can identify genuine defects and is trusted by peers. Give that person influence over the workflow without giving them indefinite veto power.
How should I measure whether adoption is working?
Track use of the defined workflow and the business-relevant criteria selected before rollout, such as output quality, correction effort, consistency, review requirements and completion time. Do not confuse login activity or attendance at training with successful adoption.
What should I say about job security?
Explain the current workflow, the decisions people remain responsible for and the skills the firm expects them to develop. Avoid sweeping promises about the future. Specific, credible information builds more trust than generic reassurance.
The bottom line
Do not enforce adoption merely because the software was expensive. Enforce a credible operating process. Start with a narrow workflow, investigate objections through demonstrations and evidence, involve respected designers, define quality and review standards, train through real work, and set a firm transition date. Preserve a documented exception route for cases where the tool is unsuitable. Once the workflow has been validated and employees have been trained, continued refusal is no longer an AI debate; it is a performance issue that should be handled consistently through the firm’s normal management process.
Rohan Hall
Founder of OceSha Ventures · AI architect and author
Rohan Hall is a technology entrepreneur, AI architect and author with four decades of technology experience, now focused on practical AI across business, education, government and global impact. He founded OceSha Ventures, builds the OceSha AI platform and Lumi, and wrote The Convergence of AI and the Top 10 Emerging Technologies.
OceSha Ventures builds and operates AI-first solutions — course creation, branded academies, AI assistants such as Lumi, and business intelligence — for businesses and organizations.
Sources
About this page. Last reviewed .
It is based on his verified public professional record.
0511aaa1b8accfd4c130993a3ca35bfa