When our client’s CTO began benchmarking his engineering org against top SaaS companies, he found that despite AI tooling access across the team, delivery velocity wasn’t lifting.
Our embedded team leaned into his ambition to set a new bar. For our own engineers, we built clear visibility, peer coaching and better measurement around the tools. Quality never dropped.
AT A GLANCE
- Industry: Online real estate marketplace (U.S.)
- Challenge: A new CTO measured engineering against top SaaS companies. The teams had AI tools, but the first numbers showed a clear gap.
- Wizeline offering: An embedded team that ran a human-focused AI adoption program: great data monitoring, extreme transparency, and peers lifting one another through continuous learning.
- Highlights:
- Doubled, then doubled again: our team's performance numbers over six months
- 98% AI adoption: across our team, up from low and medium use, with most engineers reaching high use
- No drop in quality: at any point, even as delivery sped up
A NEW CTO BROUGHT A NEW WAY TO MEASURE ENGINEERING
Our client runs one of the leading online real estate marketplaces in the U.S.
At heart, it’s a software company and in a changing software space, velocity of innovation on that software is business-critical.
In late 2025, a new CTO joined with one question in mind tied to velocity: how do the organization’s engineers currently compare with the best SaaS companies out there? He came from a company that measured cross-team engineering productivity as a matter of course and wanted similar here. To answer the question, new systems were introduced to track DORA-based metrics such as PR throughput and escaped defects. Every team was measured the same way in the new model, ours included.
Against an industry standard of 3.5 pull requests per week, the organization’s actual baseline came in at one, and the CTO wanted to know why. Was this cultural? Were there operational changes needed? Was AI actually changing delivery velocity?
On paper, the company’s AI rollout was already a success.
They’d given teams licenses and access, and were even building an early AI coding harness of their own. Even in 2025, they were further ahead on AI than most of Wizeline’s clients, and they pushed their partners to adopt alongside them. Broadly speaking, though, teams were using AI yet the numbers weren’t moving.
Recognizing the opportunity to accelerate with better, more mature adoption, the CTO asked every team, Wizeline’s included, to close the gap.
We leaned into it and our view formed fast: a metric and a target don’t alone make a plan. If we really wanted to help drive company-wide, mature AI utilization in coding processes, teams would need to know why the numbers matter, and they’d need education and partners to help them get there.
We brought what we’d learned from continuously evolving Wizeline’s own skill communities to create more impact here.
COACHING NEW PEOPLE-PROCESSES, TO BETTER ADOPT AI
As they say, what gets measured gets managed. Our thesis was that if every engineer could see and understand their own numbers (and had a peer to talk them through the numbers and how to affect them if questions arose), use of the AI tools would likely rise.
So Wizeline went to work, on top of the organization’s new systems, to ensure that each developer got a personal dashboard. That dashboard showed their metrics, what each one meant, the target, and how we reported it to the client. “If you don’t know how you’re doing, you don’t know how to improve,” said Claudia Weason, Wizeline’s Head of Forward Deployed Engineering, who led our team on the account at the time.
We knew the importance of extreme transparency during this time of change and were open with the client about the numbers, good and bad.
Dashboards showed statuses of who was behind and who was ahead in adoption, but of course they couldn’t show why. For that, we set up a model that the team called “performance engineering partners.” Here, senior engineers paired with whoever needed improvement, the same peer model as our skill communities, and talked through what might be getting in the way of AI adoption.
When a performance number dipped or adoption lag was spotted, we talked to the engineer and their engineering manager to work out a fix.
“The metrics are great, but as a starting point for a deeper conversation.”
More than only adopting the new program, we helped shape the measurement program itself too. When the tool had issues or data was incomplete, we worked through it with the client and ran our own statistics, often proposing new ways to measure, too. Wizeline had a key role to play in designing the measurement framework and setting the bar for how findings were worked with to drive actual change.
Lining up with a client’s goals like this is cultural here. We have operations aimed at finding these opportunities and acting on them.
OUR NUMBERS DOUBLED TWICE, AND QUALITY WENT UP
In the first six months, our team’s performance metrics doubled, then doubled again, with quality holding the whole way. It didn’t happen overnight, but came from our structured and highly intentional approach.
98% AI ADOPTION
Across our team, up from everyone at low or medium use during the period when adoption was under-emphasized at the client, with our engineers now reaching high use.
Because more speed can put quality at risk, and this was certainly an early concern with AI adoption, the framework was built to guard it. “We never dropped quality across the whole time,” said Weason. “We even increased it.”
After the client reorganized how its teams worked and realized the benefits of AI-driven engineering here, its product backlog fell to one or two weeks of work. That gave engineers time for technical debt and security work they hadn’t had room for before.
With adoption high, our engineers now work inside the client’s own AI coding harness as well. As Hugo Morales, Wizeline’s Delivery Director on the account, puts it: “[Having] Wizeliners actively building skills, tailoring [their] token consumption, optimizing for efficiency, building frameworks and adapting the harnesses [across] all of the repos is really, really good for us.”
LEANING IN EARNED DEEPER TRUST, AND THE WORK IS TURNING TO OUTCOMES
Our industry and the way that we engage with clients to drive their core engineering programs are rapidly changing in the age of AI. We’re leaning into the change.
On the back of this work and the trust earned through our help in setting the bar for change, our client and team are now focusing on more outcome-based programs, with our team owning projects end to end for maximum velocity. We’re excited to grow this kind of work with them. “If they are proven successful,” said Morales, “we’re going to be successful alongside them.”
Rolling out AI for software engineering and waiting for the numbers to move? Let’s talk.
Wizeline is a nearshore transformation engineering partner with 1,400+ team members across Mexico, Colombia, Argentina, Spain, Romania and Vietnam. Our consultative doers embed within client teams to move AI possibilities into modern products and workflow innovation. See our engineering and AI capabilities.
