Before You Can Be AI-Ready, You Have to Be Operationally Ready
When you have to “make the organization AI-ready,” you’re probably thinking about training programs and tool rollouts. But those alone aren’t going to get the job done.
You need to lead with infrastructure, not technology. You need operational readiness: ensuring your people and processes are aligned so a new system works properly on day 1.
You might know Ann Piccirillo from her Forbes column on human intelligence (HI) + AI. At JDA TSG, she’s our Chief Human Resources Officer, and she’s led the design of AI-augmented recruiting and HR operations at scale.
In her career, she’s scaled workforces from 300 to 10,000+ employees across hypergrowth companies, led HR through M&A, stood up recruiting functions from scratch, and built the operational infrastructure required to sustain those pivots.
Through all of that growth, she’s seen one mistake repeat itself: organizations will invest in AI tools before they have the operating model to support them.
She shares some thoughts about why this keeps happening, as well as the path to implementation success: setting up clean processes, clear accountability structures, and the right talent in the right roles, long before you bring in the AI tool.
Key takeaways:
|
Training initiatives won’t lead to AI adoption
When a board or CEO tells you to “get ready for AI,” usually the default response is to launch a training program: certifications, tool tutorials, workshops on prompt engineering.
But that’s not going to be effective.
What usually happens is people get so fixated on the tools that they don’t think about how the tools really fit into their workflow.
“Where it tends to break down is that ‘upskilling for AI’ is often treated as a training initiative instead of an operational redesign,” Ann says. “Companies roll out courses or tools without first defining how AI actually changes the work.”
Organizations won’t be able to redesign roles, retrain people, or set new performance expectations until they actually map out what the workflow looks like, and where AI fits in. It’s an important step they’re skipping.
They skip it because AI feels like a solution in itself. They focus on the tool, rather than the people using it or the workflow it’s supposed to support. Leaders assume their people will catch up, or that the technology is intuitive enough that a formal redesign isn’t necessary.
What they miss is the connection between your infrastructure and your revenue: a poorly integrated tool doesn’t just slow people down. It costs you.
“Infrastructure is revenue,” Ann says. “You don’t just drop in an AI tool and let it run by itself. It needs infrastructure, and that infrastructure feeds your revenue.”
The tool needs to be connected to your systems, built to work with them, integrated with them, and implemented throughout your organization.
When organizations skip the operational design step, they end up with inconsistent adoption, systems that aren’t part of your centralized information systems department, and passive-aggressive resistance from employees. Sometimes, those expensive AI tools get left in the dust, and people keep working the old way.
Being trained on how the tool works isn’t the same as understanding how it works in the context of your company.
The stakes are especially high in regulated industries, like financial services, healthcare, and tax. If the wrong output, unreviewed, reaches a client, that’s a huge issue. If a compliance gap goes uncaught, that’s dangerous.
And worse: When something goes wrong, no one is clear on who owns it.
How to create operational readiness to pave the way for AI readiness
If an organization says they’re ready for AI, Ann starts by looking at the work before the roles: What decisions and tasks are being done by the system, and where is human judgment still necessary?
Map the workflow end-to-end
It’s important to think about the new tool in the context of your normal workflow so it can support it.
- Where is the technology generating outputs?
- Where does a human need to review, intervene, or make a call?
- Where does accountability sit?
Then, you can shift or create roles based around the workflow, not the tool. Once you have a defined workflow, roles, reporting structures, and skills should follow naturally.
When Ann onboarded an AI-powered sourcing and first-interview platform as part of JDA TSG’s recruiting operation, that’s what she did.
She mapped out the workflow. That helped her understand what roles were necessary and where they fit.
What that looked like: She didn’t eliminate sourcers; she changed their roles, training them as associate recruiters, and built out two distinct candidate pathways (bot interview or human interview) based on generational feedback from candidates.
“This change in workflow is a change in your operational design,” Ann says. “It’s not just upskilling, it’s how I’m designing recruiting to meet the people we want to hire where they are. And at the same time, infuse efficiencies through AI tools.”
Create a three-level accountability structure
When AI gives you an answer, you need to know who owns the result.
Ann’s structure has three levels:
- Individual employees are accountable for output quality
- Managers are accountable for team accuracy rates
- The business unit is accountable for watching trends and making corrections
“Meaningful human oversight requires more than just inserting a person into the workflow. It requires authority, context, and time to exercise real judgment,” Ann says. “Too often, ‘human in the loop’ becomes a compliance checkbox.”
That’s why putting a “human in the lead” is so important.
When you integrate more accountability and oversight into existing roles, you’re giving people the bandwidth, context, and decision rights to catch problems, not just review an output at the same speed the system produces it.
Decentralize day-to-day decisions, and centralize compliance and governance
In JDA TSG’s seasonal tax program, Ann found that when decisions had to be made centrally, that led to bottlenecks that slowed down operations in important moments.
When you have those bottlenecks with AI that’s operating at scale, they can compound very quickly. One slowdown can ripple out into an entire shutdown.
The fix: pushing scheduling, time-off approvals, and team-level problem-solving authority down to employees, while keeping payroll compliance, state tax law questions, and governance questions all centrally standardized.
Operating models need to outlast AI tools
Ann knows that the tools organizations are deploying today will become outdated faster than most leaders expect. AI tools evolve on a quarterly cycle, which means your operational infrastructure has to be built for flexibility first. That’s the only way to keep humans in the lead and oversight always running, regardless of what tools come and go.
But flexibility isn’t just about swapping tools in and out. True flexibility comes from building an operation where the surrounding structure — roles, performance metrics, and incentives — can shift as the work shifts.
When AI handles volume, the meaningful signals change: accuracy rates, client impact, judgment under ambiguity, and accountability for AI-assisted decisions matter more than output. If your incentive structure doesn’t move with that, you’ve redesigned the work but left the old reward system in place, and the two will pull your people in opposite directions. That’s where mistakes get made and tools get left behind.
The structure has to be redesigned all the way down, or the tool won’t survive contact with the people using it.
FAQs
Why do training programs and tool rollouts alone fail to drive AI adoption?
When organizations treat AI adoption as a training initiative, they don’t define how AI actually changes the work, which means roles, workflows, and accountability structures stay the same while a new tool gets layered on top. People will actually use the tool when they see a company-wide alignment around it.
How do we know if our organization is actually operationally ready for AI?
Can you clearly describe what successful AI adoption looks like in your organization? If your answer is vague — or defaults to “people will be trained on the tool” — you’re not ready.
Operational readiness means you can map the end-to-end workflow, identify exactly where human judgment is required, and point to who is accountable at every level.
How should performance expectations change when AI is part of the workflow?
The metrics need to shift from output volume to output quality. Rather than measuring how much someone produces, you should be evaluating accuracy rates, judgment under ambiguity, and accountability for AI-assisted decisions.