Every UX research team faces the same challenge: reach.

No matter how good the research team is, they can only support a fraction of the decisions that would benefit from user insight.

Most organizations accept that constraint.

Kim, the VP of Design at a publicly traded cybersecurity company, didn’t.

Her team had spent years building a deep understanding of users through interviews, personas, workflow studies, and jobs-to-be-done research. Yet most of that knowledge only influenced decisions when a researcher was directly involved.

She saw an opportunity to change that.

The question wasn’t whether AI could make research faster.

It was whether AI could help research reach further.

Kim refused to accept this constraint, she saw a bigger opportunity

The natural instinct in most organizations would be to add more headcount.

But that is a temporary fix. The additional capacity only extends research’s reach incrementally.

Kim wanted to solve the underlying problem.

How to bring deep user insight to anyone in the organization, whenever they needed it? Without adding more researchers.

That question led her team to explore synthetic users: AI-generated personas built from years of interviews, personas, workflow studies, and jobs-to-be-done research.

Synthetic users, if done right, could help her achieve that vision

The idea was compelling. But compelling wasn’t the same as proven.

The team had experimented with synthetic users on their own and hit a wall. Responses were inconsistent. Quality varied in ways that were hard to predict or control. They couldn’t tell what was working and what wasn’t, or why.

That’s where Echo Point came in.

Not to advocate for synthetic users, but to determine under what conditions they were actually trustworthy. And build a system the team could rely on.

The team discovered three things, that changed everything

Over the next few weeks, the joint team tested synthetic users against real workflows: technical documents, Figma prototypes, screenshots, and recorded walkthrough videos.

Three findings changed how they approached the problem:

1. Trust came from curation, not volume
The most reliable synthetic users weren’t built from the largest amount of research. They were built from carefully selected research artifacts relevant to the task at hand. Too little context produced generic responses. Too much introduced noise. The best synthetic users were purpose-built.

2. Reliability depended on the workflow
Some use cases produced stronger signals than others. Synthetic users responding to text and images generated genuinely valuable insights. Video was more challenging and required additional care.

3. The real breakthrough was the process
What started as an experiment became a repeatable system for building, testing, and validating synthetic users. The team wasn’t just learning whether synthetic users worked. They were building internal protocols on when they worked, why, and how to scale.

What started off as an experiment quickly became a new capability

By the end of the engagement, the team had built two synthetic users grounded in real user research: a SOC Analyst and a Security Engineer.

But the personas themselves weren’t the most valuable outcome.

The bigger achievement was the system behind them.

The team now had a repeatable process for building, testing, and validating synthetic users. They understood how much context to include, which workflows produced reliable signals, and how to evaluate responses before trusting them.

What began as an experiment owned by a small group became a capability that could be scaled more broadly.

Kim’s vision was finally within reach

The rollout ahead was intentionally phased: a small internal team first, limited deployment of the two existing agents, enablement before expansion.

That sequencing matters.

Synthetic users deployed without a protocol aren’t a capability, they’re a liability.

The technical question has largely been answered. The organizational one is just beginning.

What this team built was a foundation rigorous enough to scale.

And for Kim, she was finally able to extend the reach of research.

so that more of the organization could make better decisions.

This case is anonymized due to client confidentiality. If you'd like to speak with Kim directly, I'm happy to make that introduction.

Echo Point works with product leaders who are trying to get more out of the product, design, and research teams. Happy to compare notes if useful.

The expectation was straightforward: buy AI tools, drive productivity, and enjoy the ROI.

So Mike, the GM of Corporate Banking at a top-10 U.S. bank, greenlit the Microsoft Copilot investment.

His product team was the first to get the licenses.

And to help with the roll out, he asked CK, his group CIO, to lead the charge. CK had even arranged for Microsoft to provide the team with some basic training.

Both were expecting to see the productivity gains follow.

Six months in, they were still waiting.

The numbers weren’t encouraging. Usage was low, time savings weren’t showing up, and the way the team worked hadn’t changed.

CK knew the question was coming: if we’re spending millions of dollars on AI tools, what exactly are we getting?

He didn’t have a good answer.

When CK looked closely, the real issue became clear

Product managers at the bank were using Copilot, but not in any way that was going to show up in the numbers.

They were using AI for surface level work - to write emails, summarize documents, and search the web. Tasks that saved minutes, not hours.

But none of them were using AI for core product work like writing detailed specs that took days, synthesizing customer feedback that backed up for weeks, or building market research reports that tied up the team for months at a time.

Copilot could have done all of it. The team just wasn’t using it that way.

And until that changed, the productivity gains Mike was waiting for weren’t coming.

The investment was only going to pay off if the team’s behavior changed

CK knew he had to fundamentally change how the product team interacted with AI.

To help him do that, he brought in Mustafa from Echo Point, a firm that helps CPOs build AI-native ways of working.

Together, they identified three shifts the team needed to make.

• Use AI daily, not occasionally

• Apply it to core product work across the product lifecycle

• Make it part of how the team operated every day

Anything less than that, and the investment was never going to pay off.

So the bank put together a program to change the way PMs operated

The intervention focused on three things:

• Build practical AI-enabled product skills. Teach PMs how to apply AI to real product work like requirements, discovery, customer research, and planning.

• Embed AI into daily workflows. Help them build new AI enabled workflows for repetitive and time-intensive tasks.

• Create accountability. Have every PM demonstrate measurable productivity gains using real work and actual time saved.

The bank did not need another tool rollout. It needed a different way of working.

Eight weeks later, everything had shifted

The results spoke for themselves.

90% of the team was using AI multiple times a day.

70% were saving 10 or more hours a week.

AI confidence had flipped. The majority who had rated themselves at the bottom of the scale were now at the top.

The team identified more than 165 use cases across requirements writing, planning, customer research, and market analysis.

25 workflows were built and demonstrated.

Initial projected annual productivity savings: $3.3 million. From tools the organization had already purchased.

Three workflows. Thousands of hours recovered

The numbers tell you what changed. The workflows show you where the returns actually came from.

Daniel built an API content generator that turned dense technical documentation into client-ready sales content. He reduced effort by 80%, saved 8 hours per task, and identified potential savings of 7,500 hours a month if rolled out to 100 sales team members.

Minette built a blueprint analyzer that interpreted complex technical diagrams and produced structured service blueprints. She cut analysis time from 240 hours to 20 hours per project. Across 14 projects, that is more than 3,000 hours recovered.

Robert built an AI-powered OKR workflow that drafted objectives aligned to business goals, added measurement strategies and risk analysis, and cut OKR creation time by 75%. Just across his team, that would recover over 5,000 hours.

These were not people who came into the program as AI experts. They were the same team that had been using Copilot to write emails eight weeks earlier.

What changed was their confidence and ability to use AI for their core daily work.

CK finally had his answer for Mike

The real unlock was not just giving the team AI tools.

It was showing them what good AI enabled product work looked like, giving them space to learn it, and helping them integrate it into how they worked every day.

Do that, and the returns follow.

Echo Point works with product leaders who have made the AI investment and are focused on turning it into measurable productivity gains. Happy to compare notes if useful.

Brian walked into a capability problem.

He had just stepped in as Chief Product Officer at a public-sector software company serving public safety and administration agencies across North America.

With revenue over $450M and on path to reach $1B, the company had already made aggressive product commitments to its board and investors.

On paper, the team looked experienced. Many PMs had come up through support and implementation. They knew the customers. They knew the systems.

But knowing the product is not the same as building it at scale.

The gap showed up immediately. At its current level of operation, the product organization would not meet what the next stage required.

If he was going to meet the commitments made to the board, he would have to raise the capability of the team.

And he would have to do it fast.

Brian had two choices: rebuild slowly or leapfrog

The conventional path.

Send the product team (PMs, POs and designers) through formal product management training. Invest in fundamentals, build product craft, and raise the bar on product judgment.

Then introduce AI as a multiplier.

It’s a rational approach. But it would take 8 - 12 months.

Brian didn’t have that kind of time.

The alternative was riskier.

Reverse the sequence entirely. Embed AI into the team’s day-to-day work immediately, and build product judgment inside live delivery.

If it worked, the team could operate at a higher level within weeks. If it failed, they would move faster in the wrong direction.

He chose to leapfrog

But choosing the path was the easy part.

If this bet was going to change outcomes, AI could not sit on the sidelines. It had to reshape how product work happened daily.

Brian didn’t need a training event. He needed a behavioral shift. Fast.

Here’s how he operationalized behavior change

Brian knew the shift couldn’t happen through AI tools alone.

So he partnered with Echo Point to close both the product craft and AI capability gap. It unfolded in three stages.

#1. Identify the gaps

They began by looking directly at where the work was breaking down.

The diagnosis was clear. The team needed to strengthen requirements quality, prioritization, and rapid prototyping. And AI existed in the organization but at the edges. It wasn’t touching real product work.

Both had to change.

#2. Leveraged AI to build product craft skills

The conventional path trains product craft first, then layers in AI.

Brian reversed it. He taught his team how to use AI for real product work.

The team worked through structured workshops, applied exercises, and live critiques, all centered on real product problems. All in a safe and guided environment.

The goal was to build AI capability in the work. And improve product craft through it.

#3. Apply it to live work

Then came the moment of truth.

PMs took what they had learned and applied it to live work. No workshops. No guided environment. Real commitments, with real consequences.

If the leapfrog was going to work, it would show up here.

Within weeks, the shift showed up in real work

Tim (product owner) compressed a 30-day, 400-page specification cycle into a single day without sacrificing accuracy, freeing weeks of capacity.

Trym (product manager) reduced backlog refinement time by 80% by building agents that strengthened acceptance criteria and surfaced edge cases.

Victory (user research) built a competitive intelligence engine that clarified market gaps and reshaped roadmap tradeoffs across five competitors.

William (product manager) created synthetic users to pressure-test assumptions early, validating ideas before they consumed roadmap capacity.

This was not a few high performers pulling ahead of the pack.

In under two weeks, more than 60 AI-enabled workflow improvements were built and demonstrated across the team. These were not side experiments. They were changes applied directly to live delivery.

The leapfrog was working.

One example made the shift unmistakable

Chris (product manager) did not present a sharper PRD. He shared a fully working new feature.

A new County ERP workflow engine that handled record inserts, updates, validations, and automated notifications. Not a mockup. But a production ready module.

The estimated engineering effort dropped from 1,500 hours down to roughly 200.

The signal was not just productivity gains. It was a step change in capability.

The numbers told the same story

Six weeks in, Brian had his answer

AI adoption
Before, 60% of PMs used AI once a week for surface-level tasks. Six weeks later, 80% were using it multiple times daily, across the full product lifecycle.

Productivity
Time saved per PM went from less than 5 hours a week to 12 or more. Forty percent of the team was saving 15 to 20 hours weekly. Projected annual impact: $3.6 million. ROI above 1,900%.

PM effectiveness
The team entered with gaps in requirements quality, prioritization, and prototyping. All three improved. Requirements were stronger, decisions moved faster, and prototyping accelerated.

PM engagement
Before the program, 56% of PMs reported feeling less engaged. After, 60% reported being significantly more engaged.

The product organization was ready for its next stage

Six weeks later, the capability ceiling had shifted.

Product work was sharper across the board. Better requirements, pressure tested decisions, and stronger judgment.

Product ops was set up to ensure the team would not slide back.

The commitments to the board did not change.

The team’s ability to meet them did.

This case is anonymized due to client confidentiality. The operating changes and results are exact.

Echo Point works with product leaders to close the gap between where their teams operate today and what the next stage requires. AI capability building, product craft, and organizational design. Let's talk.