At a glance: In six months, the product organization of a Fortune 100 financial services company grew its self-described advanced AI users from 19% to 52% and weekly use of its core AI ticketing tool from 27% to 74%. Most people reported saving an hour or more per week, roughly 12 full-time employees' worth of capacity, and 78% said AI also improved the quality of their work. Prodify Senior Fractional PM Jen Wulf led the engagement by treating AI adoption as a product problem.
Most large companies have rolled out AI tools by now. Not all have product teams that actually use them effectively. When a Fortune 100 financial services company brought Prodify in to drive AI adoption across the product organization behind its consumer app, leadership had the mandate and the tools, but not the results. Six months later, weekly use of the team's core AI ticketing tool had jumped from 27% to 74%, the share of product owners who considered themselves advanced AI users had nearly tripled from 19% to 52%, and teams were reporting that their product deliverables were getting both faster and higher quality. Here's how Prodify Senior Fractional PM Jen Wulf got them there, and what other large organizations can take from it.
Why Do Enterprise AI Rollouts Stall? A Top-Down Mandate Meets Bottom-Up Resistance
Executive leadership had a clear sense of urgency. Peers across the industry were moving quickly on AI, and the company wanted to operate like a modern, tech-forward business. There were already promising pockets of success: teams were using AI to extract requirements from decades-old legacy systems, and to prototype and feed user feedback into AI agents so they could iterate faster in a heavily regulated environment with many layers of approval.
Inside the product team, though, adoption was concentrated among the top ~5% of AI early adopters. Several things were getting in the way for the rest of the organization:
- No time to learn. Teams were already stretched thin delivering against existing OKRs and said they didn't have time to learn new skills, even ones that would ultimately save them time. This was the largest, most common hurdle to clear.
- Skepticism from early bad experiences. Many people had tried AI tools, didn't get the output they wanted, and wrote them off. Others worried that leaning on AI would make their product skills rusty.
- Tools rolled out without enablement. New capabilities were being launched, including AI agents built into the team's software development ticketing tool, but people didn't know how to access them, how they worked, or why they should use them. The lack of enablement paired with an explosion of new tooling created widespread overwhelm.
- Confusing tooling. The company had built its own internal AI chat platform that offered several models. Many users didn't realize they could switch models or did not select the appropriate model for their use case. They defaulted to a lightweight one, got weak answers, and concluded that "AI doesn't work." The platform also didn't retain chat history reliably, which made it hard to build on prior work and context.
- No clear "why" or "how." There was urgency to adopt AI, but no shared vision for what adoption should achieve and no real understanding of what was blocking product owners.
- Enablement running on volunteer time. Like many companies right now, much of the AI evangelism was happening through volunteers squeezing it into already-full jobs. This led to inconsistent programming, little to no coordination with enterprise-level AI adoption efforts, and a lack of overall strategy.
Complicating all of this, different departments operated very differently, from product process to tooling. A single, company-wide mandate was never going to land the same way everywhere.
How Do You Drive AI Adoption on a Product Team? Treat It Like a Product Problem
"There's so much urgency and buzz around AI that people lose sight of the fact that it's still a product problem. It's understanding your users, their pains and their goals, and then making a plan based on that." - Jen Wulf, Senior Fractional PM, Prodify
The work started where any good product work starts: understanding users, their pains, and their goals.
1. Research first: a survey of ~200 people and 15 in-depth interviews
Jen surveyed roughly two-thirds of the product organization (about 200 people) and followed up with interviews of 15 product team members and cross-functional collaborators. That research surfaced hidden blockers. The biggest: many people were spending seven or more hours a day in meetings. They simply had no time to learn AI, a frustrating chicken-and-egg problem, since AI was one of the best ways to win that time back.
2. Free up time first: meeting transcripts as fuel for product work
Because meetings were the biggest drain on time and energy, that's where we started. The organization didn't have a transcription service, and an internal tool was still being built. Rather than wait for that internal tool, Jen pushed to get everyone upgraded access to Microsoft Teams Premium so meetings could be transcribed automatically.
Those transcripts became the context product managers fed into their AI tools, forming the foundation for product deliverables like requirements, user stories, and follow-ups. This also allowed product team members to have their AI "attend" meetings for them when they were unavailable or optional, synthesize action items and begin executing on next steps. Jen paired the rollout with case studies on using meeting-support tools well, plus examples from early adopters. That small win opened up just enough space for people to start learning: how to prompt, which tool fits which task, and what good output actually looks like. This triggered a massive shift towards AI adoption.
3. Let peers teach peers: bi-weekly AI demos
Generic "AI for PMs" advice online rarely survives contact with a large enterprise's specific processes, systems, and regulatory requirements. So Jen recruited a group of 10–15 AI early adopters, identified through the survey, interviews, and referrals, who were already doing impressive things. One had built an integration across monitoring, code, and ticketing tools that caught bugs before users reported them and proposed a fix.
These early adopters led bi-weekly demo sessions showing colleagues exactly how they used AI within the company's own tools and guardrails. Along the way, Jen documented a library of company-specific use cases, including some that aren't classic PM use cases but mattered enormously here. Examples included writing new stories and epics according to detailed organizational guidelines, creating agents to support self-service internal tools, and creating user persona agents.
4. Enable the tools people already had
Jen also closed the gap between tools being launched and tools being used:
- Partnered with the teams building AI agents in the software development ticketing platform that check epics and user stories for consistency and completeness, which is increasingly important as development moves toward AI-assisted and agentic workflows, then rolled them out to product owners with training.
- Taught product managers and product owners to build their own custom agents.
- Supported the rollout of a leading commercial AI assistant and the new internal meeting-support tool with practical enablement, so people could get value quickly rather than figure it out alone.
What Results Can AI Enablement Deliver? Adoption, Speed, Quality, and Confidence All Went Up
Measured between the baseline and follow-up surveys (Q1 to Q3), the results were significant:
| Metric |
Q1 baseline |
Q3 follow-up |
| Weekly use of the AI ticketing tool |
27% |
74% |
| Product owners who rate themselves advanced AI users |
19% |
52% |
| Respondents saving an hour or more per week |
Not measured |
Majority (~12 FTEs of capacity weekly) |
| Respondents reporting better-quality work |
Not measured |
78% |
- Adoption rose across every core AI tool, with the priority ticketing tool AI usage growing nearly 3x, from 27% weekly use to 74%.
- Confidence nearly tripled. At the start, most product owners described themselves as beginners or unsure, with only 19% self-reporting as "advanced AI users." By the follow-up, advanced users were the single largest group, at 52%.
- Real time savings. The majority of respondents reported saving an hour or more per week by using AI. Across the organization, this added up to roughly 12 full-time employees' worth of capacity every week.
- Better work, not just faster work. 78% of respondents reported that AI also improved the quality of their work.
Because self-reported metrics can be optimistic, we validated them against actual tool usage data. Usage trends moved in the same direction as the survey results over the same period.
Two other patterns stood out. People who engaged consistently with the program's resources saw significantly stronger results, which held true across seniority levels and internal teams. Ultimately, the conversation inside the organization shifted from "Should I use AI?" to "How do I get even more effective with it?"
"I was so shocked how impactful the enablement was. We were doing demos bi-weekly and building a library of information, but I would have never imagined it would have had this kind of impact." - Jen Wulf, Senior Fractional PM, Prodify
Key Takeaways for Large Organizations Rolling Out AI
If you lead a product organization at a large, complex company, here's what we'd want you to know before your next AI push:
- Treat AI adoption as a product problem. Start by talking to your people. Understand where they spend their time, what they've tried, and why it didn't stick. The blockers are often not what you assume.
- Take pushback seriously, because it's often valid. When people resist a tool, sometimes it's because the tool doesn't meet their needs. Alleviate the real pain and adoption follows. As Jen put it: "Some of the reasons people weren't adopting these tools were really valid. Talk to your people and understand whether their pushback is real, or whether you're giving them the tools they actually need."
- Consider the implications of buy vs build decisions. Frontier models are improving rapidly, protecting enterprise data, and providing a dedicated app experience. Consider whether building internally is still worth it, especially if the timeline to launch is going to delay AI adoption.
- Enablement isn't optional. Launching tools and hoping people figure them out isn't a strategy. Structured, ongoing enablement was the single most surprising driver of results in this engagement.
- Let peers lead. Examples from colleagues working within the same systems and regulatory constraints are far more persuasive than generic best practices. Find your early adopters and give them a stage.
- Measure what the business cares about. "We need to adopt AI" isn't a goal, and adoption alone is only a leading indicator. Tie your metrics to outcomes like time saved, product quality, and development velocity.
- Don't run enablement on volunteer time. Enthusiastic champions doing this on top of full-time jobs will burn out or move on. Give AI enablement dedicated ownership, whether that's an internal hire or an experienced partner.
- Plan for the operating model, not just the tools. AI is reshaping how product managers, engineers, and delivery teams work together across the full product and software development lifecycle. Consider how AI can help with prioritization, roadmapping, design, delivery, and post-release validation.
Frequently Asked Questions
How long does it take to drive AI adoption in a product organization?
In this engagement, measurable change took about six months, from a Q1 baseline survey to a Q3 follow-up. The first win, freeing up time with automatic meeting transcription, created room for people to start learning before deeper enablement began.
How do you measure whether AI adoption is working?
Run a baseline survey covering tool usage, self-rated AI skill, time saved and work quality, repeat it a few months later, and check the self-reported trends against actual tool usage data. Treat adoption as a leading indicator and tie your metrics to outcomes the business cares about, such as time saved, quality and development velocity.
Should a large company build its own internal AI chat tool or buy one?
Building in-house can protect enterprise data, but it can slow adoption if the tool lags behind commercial assistants. Here, an internal platform with confusing model selection and unreliable chat history led many people to conclude that "AI doesn't work," and adoption picked up once a leading commercial AI assistant was rolled out with proper enablement.
Who should own AI enablement for product teams?
Give it dedicated ownership, whether that's an internal hire or an experienced partner like Prodify, rather than volunteers fitting it around full-time jobs. Volunteer-run efforts tend to produce inconsistent programming and little coordination with company-wide AI initiatives.
What's the fastest way to get skeptical product managers to use AI?
Show them peers doing it inside the same systems and guardrails. Bi-weekly demos led by 10 to 15 internal early adopters, plus a library of company-specific use cases, were far more persuasive than generic "AI for PMs" advice.
Ready to Turn Your AI Mandate Into Results?
If your product organization has AI tools but not yet AI results, Prodify can help. Our coaches have led AI enablement and product transformations inside large, highly regulated enterprises, as well as with scale-ups. Learn more about our AI advising, or request a call to talk about where your teams are today and what it would take to move faster and build better.
Request a Call
Related reading: Client Case Study: Laying an AI Foundation to Drive Growth via Personalization and Incorporating AI Into Your Product Strategy.
