AI adoption for mid-sized companies: what HCA, Ardent, and Tractor Supply's rollouts mean for a 200-person Nashville business

Three of Middle Tennessee's largest employers, HCA Healthcare, Ardent Health, and Tractor Supply, have moved AI out of pilots and into daily work. Their public accounts agree on the lessons: start with administrative work where the data is standardized, treat change management as the real project, measure before and after, and set governance before you scale. AI adoption for mid-sized companies follows the same rules with far less machinery.

That last point matters. HCA has 189 hospitals and more than 320,000 colleagues. A 200-person physician group in Franklin, a general contractor in Mt. Juliet, or a freight brokerage in La Vergne has one leadership team and one payroll. What the big systems have said on the record is more useful to that company than any vendor deck, because they have already paid for the mistakes.

What are Nashville's biggest companies doing with AI?

HCA's CFO told the UBS Global Healthcare Conference in November 2025 that the company was "in early innings," that it wants "either a clinical or a financial return" before scaling, and that the hard part was "how hard it is to roll out AI tools and change behavior on the ground." His line that stuck: "We have to be just as good at implementation and change management as we do about building AI tools." The administrative use cases he called the shortest pathway to value span IT, supply chain, HR, revenue cycle, and physician practice management.

In May 2026, HCA's chief transformation officer, Dr. Michael Schlosser, gave a longer account in MIS Quarterly Executive, republished on HCA Healthcare Today. The nurse handoff tool took about a year of iteration with bedside nurses before accuracy and usefulness scores reached the high 90s. His group runs three teams: product, implementation (change management, training, on-the-ground rollout), and data. Every product has a business-unit sponsor with a day job.

Ardent's chief digital and transformation officer, Anika Gardenhire, described the Ambience pilot's success factors this way: "You have to train really well, you have to have at-the-elbow support and the solution itself needs to be usable." Clinicians who were not in the pilot asked, "When am I going to get my turn?"

Tractor Supply built Hey GURA on infrastructure its stores already had (SD-WAN, in-store cameras, handhelds), piloted it in 200 stores, and was chainwide within months. By December 2025 it had named OpenAI its primary AI partner and said it had automated roughly 1,500 processes (Modern Retail). Glenn Allison, its VP of AI platforms: "It's all centered around the team member. We're using AI to save time and reinvest that time back into the customer experience."

Elsewhere in town, AllianceBernstein, Barge Design, Schneider Electric, and Metro joined the AWS Skills to Jobs Tech Alliance in October 2025 to build local AI and cloud talent (Nashville.gov).

What do these rollouts teach about AI adoption for mid-sized companies?

Six lessons show up in every one of these accounts. Each comes with a translation for a company with 200 people, a Nashville example, and one number to track.

Start with administrative workflows where the data is standardized

HCA's CFO put administrative work first because it has structured inputs, repeatable outputs, and no patient at the bedside. Ardent chose documentation and coding for the same reason: every visit produces the same kind of note.

For a mid-sized company, that means accounts payable, intake, scheduling, proposals, onboarding paperwork, and reporting. A freight brokerage on the I-24 corridor in La Vergne processes proofs of delivery and carrier invoices all day; the format barely changes, and the exceptions are where the money leaks. That is the first workflow, not the sales pitch.

Track: hours per week the team spends on that workflow, measured before you change anything.

Treat implementation and change management as the actual project

Schlosser's plainest line: "Change management is often the number one failure point. You can sometimes skip it in a pilot because everyone's excited." HCA staffs a dedicated implementation team from day one. Ardent paired training with at-the-elbow support.

You do not need a department. You need one manager who owns adoption, two hours of hands-on training per person, and someone within shouting distance for the first three weeks. If the Copilot seats you bought last year sit idle, this is why. The software was the easy part.

Track: weekly active users as a share of licensed seats at week four and week twelve.

Pick one consequential workflow, not a tool rollout

The nurse handoff idea came from nurses, and HCA spent a year on that single workflow before expanding. Tractor Supply's assistant answers one kind of question for one kind of employee: what is this product, where is it, and what goes with it.

For a commercial general contractor in Mt. Juliet, the equivalent is RFI and submittal drafting, or daily report generation from field notes. Choose the one that touches every project every week. Giving 200 people a chat tool and hoping is the opposite of this lesson.

Track: turnaround time on that workflow (days from RFI raised to RFI answered, for example).

Measure before and after

HCA attaches a value-tracking team to every idea from the discovery phase, doing what Schlosser calls "napkin math" before any build. Ardent measured provider productivity, patient contact time, clinical accuracy, and provider satisfaction. Tractor Supply tracks sales transactions when the tool was used and service speed on specific tasks.

If you run on EOS or Pinnacle, you already have a scorecard. Pick the number on it that the workflow moves, capture two weeks of baseline, then run the pilot. Schlosser also warns about the failure mode: "we may believe an idea will have a significant impact, but it doesn't show up that way in the data." When that happens, stop.

Track: one scorecard number, baseline versus week eight, plus a satisfaction check with the people doing the work.

Make leaders use the tools visibly

HCA's governance committee, including the CEO and CFO, conducts quarterly business reviews of the AI portfolio. Business-unit sponsors stay with each product through scaling because "they're well-positioned to help us drive the adoption." Tractor Supply's CEO, Hal Lawton, talks about AI on earnings calls.

At 200 people, visibility is cheaper and more powerful. The CEO who drafts the weekly L10 summary with the tool and says so, or the practice administrator at a Cool Springs physician group who reviews the AI-drafted prior authorization letters herself for a month, does more for adoption than any memo.

Track: share of the leadership team that has personally shipped one workflow with the tool in the past 30 days.

Set governance early, then keep it running

HCA created a Responsible AI Committee with its chief legal officer, chief privacy officer, and compliance leaders before the first tool went live. Every application passes governance before it can be tested, and Schlosser calls the follow-up "continuous governance," because "models drift, populations evolve and risk profiles change." Metro Nashville keeps a public AI use registry, 24 tools as of July 2026, and runs a 31-question risk assessment (Nashville Scene).

Your version is a one-page acceptable-use policy, a register of approved tools with a named owner, and a quarterly review on the leadership calendar. A physician group needs the HIPAA and Tennessee Information Protection Act questions answered before the pilot, not after. We publish a free AI acceptable-use policy template for Tennessee companies.

Track: tools in the register versus tools actually in use; the gap is your shadow AI.

What the big systems can do that you cannot, and what you can do that they cannot

Schlosser calls HCA's scale "almost a superpower for innovation." It is. HCA can spend a year refining a model with bedside nurses, run three Innovation Hub hospitals, keep biostatisticians on a value-tracking team, and get the FDA's feedback on a model. Tractor Supply can make OpenAI its primary partner and get its engineers on the phone. You cannot, and you should not try to build what they build.

Here is what you have that they do not.

Speed. Ardent took roughly seven months from announcing the enterprise rollout to completing it across its medical groups, and that was fast for a 30-hospital system. A 200-person company can baseline, pilot, and decide inside one quarter.

Fewer approval layers. HCA needs regional and hospital-level leaders who "know what I need from them." You need one meeting.

One leadership team. HCA's sponsors have day jobs and lend the transformation group "a little bit of their time." Your sponsors are the people in your Monday meeting, and the work is their day job.

Buy, don't build. Ardent bought Ambience. Lifepoint chose a scribe partly because it fit its athenahealth EHR. The same categories of tool now sell at mid-market prices, and the selection criterion is the same: does it fit the system your people already use. Our guide to Nashville AI companies and vendors covers who sells what.

How should a mid-sized company roll out AI next quarter? A 30-60-90 outline

Two things make this go faster. Our half-day leadership team AI working session compresses the day 1 to 30 decisions into one afternoon, and the free 10-minute AI readiness scorecard tells you which of the six lessons your company is weakest on before you start. If the training cost is the sticking point, the Tennessee IWT grant can reimburse up to $25,000 of it.

How we can help

Working Model is an operator-led AI consulting and training firm in Nashville. We run the leadership working session, a four-week AI Fluency Program for teams, and a fractional Chief AI Officer engagement for companies that want someone to own the governance calendar and the scorecard. Prices are on the pricing page, and our note on what an AI consultant costs in Nashville explains the ranges.

The short version

HCA, Ardent, and Tractor Supply did not win on model quality. They won on picking the right first workflow, staffing the rollout, measuring honestly, and governing early. AI adoption for mid-sized companies runs on the same four habits, and a company with one leadership team can build them in a quarter.

Frequently asked questions

How do you roll out AI in a company of 200 people?

Pick one administrative workflow that runs every week, baseline the number it moves, write a one-page acceptable-use policy, and pilot with five to ten people who do the work. Give them training and at-the-elbow support, review weekly in the leadership meeting, and decide at day 90 whether to scale, adjust, or stop. Then repeat with the next workflow.

Should a mid-sized company start with a tool rollout or a single workflow?

A single workflow. HCA spent a year on nurse handoffs before expanding, and Tractor Supply's assistant answers one kind of question for one kind of employee. Buying 200 Copilot or ChatGPT seats and hoping produces idle licenses. Choose the workflow, then choose the tool that fits the system your people already use.

How do you measure whether an AI rollout worked?

Measure before you turn anything on. Ardent tracked documentation time, patient contact time, accuracy, and clinician satisfaction; its pilot showed 45% less documentation time. For a mid-sized company, take one scorecard number, capture two weeks of baseline, and compare at week eight. Add weekly active use and a short satisfaction check from the people doing the work.

Do mid-sized companies need an AI governance committee?

Not a committee, but the function. HCA's Responsible AI Committee reviews every tool before testing and rechecks it over time. Your version is an acceptable-use policy, a register of approved tools with a named owner, and a quarterly review on the leadership calendar. Health services and any company handling personal data should answer the HIPAA and Tennessee Information Protection Act questions before the pilot.