When AI started gaining traction a few years ago, Neville Allen and his peers at insurance technology company Vertafore didn’t wait for permission to explore the emerging technology — they dove right in.
“We have a culture where experimentation isn’t just encouraged — it’s expected,” the director of development said.
This experimentation-centric culture has encouraged the company’s engineers to embrace a wide range of tools to enhance their day-to-day work, including GitHub Copilot and Claude Code, which have enabled them to spend more time solving complex problems.
“They’ve fundamentally changed the pace and rhythm of how we build software,” Allen said.
Sometimes, the best solutions for working smarter and faster aren’t technical at all. For instance, at Mastercard, Vice President of Product Management Services Jess Greco strives to foster a sense of psychological safety on her teams while guiding team members as they work through problems together.
“A coaching approach to leading teams is better than any tool I’ve seen thus far,” Greco said.
Backed by strong leadership and a focus on experimentation, the tech teams at Mastercard have adapted to transformations across the financial landscape.
“It’s rare to see an established company constantly challenging itself at reinvention, but it’s happening every day across the enterprise,” Greco said.
Below, Allen, Greco and leaders from 11 other companies share the tools and approaches that drive their teams’ day-to-day work, how their employers encourage experimentation and how these efforts enable their teams to adapt to change.
Mastercard is a technology company in the payments space.
What tools support your day-to-day work?
We have a robust set of team-level rhythms, supported by open and candid communication norms, to ensure we can keep things moving and stay on the same page. This includes best-in-market tools for: whiteboarding and collaboration, project tracking, communication, news and market intelligence, and AI coaching.
Ultimately, tooling is about having a means to an end. The best collaboration comes from a baseline level of psychological safety and a leadership approach that focuses on coaching people as they work through a problem as a group. I don’t want to be the person telling them how to solve the problem and just enact the steps I think need to occur. I want them to figure out how to get to the goal and be opportunistic about the best ways to get there. A coaching approach to leading teams is better than any tool I’ve seen thus far.
How does your team experiment?
For a successful experiment, teams need to start with the right mindset. My team has been hired for a professional orientation that is highly adaptable and able to navigate ambiguity. We’re always looking at the work in terms of “the art of the possible” — What would need to be true to achieve this vision?
Secondly, we bring a hypothesis orientation to the work. When we’re looking at a goal, we’re discussing ways we can break it down into pieces that can be validated or invalidated to learn enough to get us closer to something realistic and trustworthy.
“When we’re looking at a goal, we’re discussing ways we can break it down into pieces that can be validated or invalidated to learn enough to get us closer to something realistic and trustworthy.”
Third, we each bring a unique point of view and have open, candid collaboration. I’m trained as a designer and bring classic service design and change management techniques. Another team member is a project management expert and another is a delivery expert.
And last but not least, we ensure our objectives are focused on achieving the business goal, not specifying how we do it. Experiments are all about adjusting an approach in response to data. This is especially important with our team’s focus on agentic AI and new business opportunities across the enterprise.
How does Mastercard adapt to change?
Established financial companies seem to go one of two ways — either they become more rigid, formal and risk-averse over time, or they become more dynamic and experimentation-oriented. From my seven years at Mastercard, we have certainly been trending in the latter direction. The R&D teams are at the leading edge, our products are innovating on what’s possible in this space, and our leadership is sponsoring new big bets that challenge preconceptions of what was versus what could be.
It’s an exciting time to see our executives ask teams what evidence for their proposal exists and to see the areas they have de-risked as they work through the problem space. It’s rare to see an established company constantly challenging itself at reinvention, but it’s happening every day across the enterprise.
Perk’s platform enables companies to book, manage and report business travel.
What tools support your day-to-day work?
As an AI-native platform, we are always looking for ways to use technology to remove the “shadow work,” or hidden, manual tasks, from our days so we can focus on the real work. Our people operations team runs on a core stack built around Zendesk and HiBob. Zendesk is our employee service hub with ticketing, internal knowledge base and AI agent, which now resolves 60 to 70 percent of HR queries that come through the chatbot without any human intervention. HiBob is our HR system of record and the engine behind our automated employee journeys. Because the two are integrated, our AI doesn’t give generic HR answers; it gives personalized ones. It knows the employee’s department and their region.
“Zendesk is our employee service hub with ticketing, internal knowledge base and AI agent, which now resolves 60 to 70 percent of HR queries that come through the chatbot without any human intervention.”
How does your team experiment?
Adoption is important for success so we always test before we commit. When we rolled out our AI agent, we conducted a pilot study first, monitoring how it handled real employee queries, refining responses and building confidence before opening it up fully. Before going live with our knowledge base, the whole people team ran a content hackathon, working in pairs to write articles for repetitive questions employees were already asking us. That groundwork, together with specific use cases built in the backend of the bot, is what got us to a 60 to 70 percent AI resolution rate for HR queries coming through the chatbot.
We also experiment with how we localize. Our AI and human agents — people operations and payroll team members — handle different labor laws by pulling from a localized knowledge base. That required the team to think carefully about how we structured content, not just what we put in it. Each change was piloted, iterated on and scaled only once it worked.
How does Perk adapt to change?
Perk’s product is built around eliminating shadow work. We apply that same philosophy internally. When I joined people ops in 2020, employee support was running through a shared inbox with zero visibility, no data and fully manual triage. As the company scaled, request volume grew massively, and that model simply broke. So, we transformed the entire operation: implemented Zendesk, built a localized self-service knowledge base, integrated HiBob for automated workflows and layered in AI. The AI now handles the shadow work, repetitive payroll and policy questions, freeing the team to focus on the complex, human work that actually requires judgment and empathy. Adapting to change at Perk means asking, “How do we do this smarter?” not just “How do we do more of the same?”
.png){kind=logo}
A subsidiary of Comcast Corporation, NBCUniversal is a global media and entertainment company that oversees multiple brands, including NBC, Focus Features and Peacock.
What tools support your day-to-day work?
At the core of any productive team is a toolset that’s intuitive, reliable and secure. It should be tools that people actually want to use. For planning and visibility, we rely heavily on platforms like Jira. It serves as our central hub for organizing work, managing sprint cycles and coordinating production releases. Just as importantly, it gives stakeholders a transparent view into progress, helping bridge the gap between engineering and product teams.
Alongside traditional tools, AI has quickly become an everyday companion in the workplace. Solutions like Microsoft 365 Copilot are particularly impactful. Because they’re embedded directly into familiar applications, they can assist with drafting emails, summarizing content and even helping to structure documents or technical solutions. Beyond productivity tools, AI capabilities within cloud platforms are advancing rapidly as well. Whether it’s integrated assistants in data platforms or cloud-native AI services, these tools are becoming more refined with each release and are helping teams accelerate development, improve documentation and uncover insights faster.
And despite all of this innovation, one thing hasn’t changed; the good ole’ spreadsheet still holds its ground. When it comes to quick collaboration, data validation or aligning on assumptions, a well-structured Excel file remains one of the most effective tools available.
How does your team experiment?
Experimentation is essential, but it works best when it’s intentional. For us, experimentation starts in controlled environments like sandbox datasets or isolated cloud projects. These spaces allow teams to explore new ideas without impacting production systems or sensitive data. The goal is always the same: Learn quickly, but do it safely and transparently. AI is playing a growing role here as well. It can help teams rapidly generate potential solutions, identify edge cases and even highlight risks early in the process. In many cases, what previously took days of iteration can now be explored in hours.
That said, speed doesn’t replace discipline. Every experiment is documented, and results, both successes and failures, are shared with the broader team. This ensures that knowledge compounds over time rather than staying siloed. Choosing the right tool is also critical. The wrong tool can slow things down more than it helps. For example, imagine trying to use a general purpose business intelligence dashboard tool to build a large-scale data pipeline. While the tool might allow limited data transformations, it’s not designed to handle complex dependencies, scheduling or large data volumes. The result is often fragile workflows, performance bottlenecks and increased maintenance overhead. A dedicated data engineering framework, on the other hand, is purpose-built for that scale and complexity. Experimentation should challenge assumptions but still respect the fundamentals of good engineering design.
How does NBCUniversal adapt to change?
Adapting to change isn’t a one-time effort — it’s a continuous mindset. Within our organization, staying “in the know” is a shared responsibility. Teams regularly exchange ideas through forums, internal communities and peer-led sessions, known as engineering guilds, where engineers showcase what’s working in real-world scenarios. This kind of knowledge-sharing helps accelerate adoption and builds confidence across the organization.
Knowledge-sharing around AI has been a key driver of conversation and focus for our teams over the past two years. Engineers actively exchange ideas through live “show-and-tell” demonstrations and Confluence documentation, where practical use cases and lessons learned are shared in real time. This collaborative approach is paired with structured upskilling opportunities, including vendor-led sessions, hands-on working labs and internal AI bootcamps that allow teams to learn by doing. Engineers also stay current by attending industry conferences where emerging AI capabilities are showcased along with practical ways to integrate them into existing architectures and data pipelines based on specific business needs.
“Engineers actively exchange ideas through live ‘show-and-tell’ demonstrations and Confluence documentation, where practical use cases and lessons learned are shared in real time.”
In addition, NBCUniversal offers a broad suite of training resources designed to meet teams where they are in their AI journey, which includes free LinkedIn Learning pathways, curated office hours and town halls that highlight real-world AI applications. One standout experience that I enjoyed was the Microsoft-led Copilot office hours, where experts guided individuals through foundational and advanced use cases in an open, interactive setting. These sessions helped employees build practical skills, which enabled us to confidently incorporate AI into our day-to-day workflows.
This approach had an immediate impact. It enabled engineers to integrate AI into their workflows quickly and safely by using it to streamline development, improve documentation and reduce repetitive tasks. At the same time, it created a foundation for building internal AI solutions tailored to our specific needs.
More importantly, it positioned the organization to lead rather than follow. By combining governance with enablement, we’re not just adopting new technology; we’re helping define how it can be used effectively, responsibly and at scale within the media and entertainment industry.
Vertafore offers insurance technology solutions that are designed to connect every point of the insurance distribution channel.
What tools support your day-to-day work?
GitHub Copilot and Claude Code have become a critical part of how my team operates day to day. These AI-assisted tools support everything from brainstorming solutions and scaffolding code to reviewing logic and accelerating repetitive development tasks. Work that once required hours of research or multiple rounds of iteration can now get started in minutes, allowing engineers to spend more time solving complex problems and less time on administrative overhead. They’ve fundamentally changed the pace and rhythm of how we build software.
How does your team experiment?
We have a culture where experimentation isn’t just encouraged — it’s expected. One of the principles we operate by is leaving the code better than we found it. Whether that’s refactoring a function, adopting a new pattern or evaluating a tool that could improve efficiency, team members are empowered to test ideas and share what they learn. That mindset helps us continuously improve our systems and processes rather than simply maintain them.
“One of the principles we operate by is leaving the code better than we found it.”
How does Vertafore adapt to change?
The rise of AI is probably the best example I can point to. When these tools started gaining traction, we didn’t wait for a formal directive to explore their potential. Instead, we began experimenting and finding practical ways to apply them to real engineering challenges.
One of the most impactful examples was a major database consolidation effort. We used AI-assisted development tools to help create scripts that merged three separate databases into a single environment while preserving schemas and maintaining backward compatibility. We also used AI to help build a framework that could dynamically update SQL references to the correct schema, minimizing changes to the existing codebase.
AI didn’t make the technical decisions for us, but it accelerated research, prototyping and implementation. We’ve seen similar benefits when adding Microsoft Graph Outlook API support, resolving memory leaks and investigating performance issues. In each case, AI helped our engineers move from uncertainty to actionable solutions faster while still relying on human expertise and judgment.
Benchling’s cloud-based platform is designed to accelerate how biotech companies discover, develop, and scale life-changing products by bringing together scientific workflows, structured data and collaboration tools in one place.
What tools support your day-to-day work?
Claude is our backbone today. I run a personal project that knows my team, voices, calendar and work in flight, and everything else feeds into it. It’s always aligning with sources of truth, architecture and structures. We use Workato to wire Slack, Google Workspace, Atlassian and our back-end systems together so the rote work doesn’t need a babysitter. Confluence holds the team’s living playbooks: SOPs, prompt and skill libraries, and training docs. App Script is the unglamorous duct tape that turns a sheet into a small app or makeshift connector, when a tool would be overkill. Slack is where the team works, so we built skills that post into it: a daily brief, a channel digest, project updates, a room booker and various pipeline trackers. With all these tools together, the team moves faster than it should be able to. The stack does a lot, but combining it with consistent training and alignment across the org is what removes toil.
“We use Workato to wire Slack, Google Workspace, Atlassian and our back-end systems together so the rote work doesn’t need a babysitter.”
How does your team experiment?
We have a culture of experimentation, especially today, when tools and ways of working aren’t set in stone. It’s an exciting time to be a traditionally nontechnical team.
“AI Day” is how we run experimentation as a department. Every quarter, we block protected time to build something: an agent, an automation or a pipeline that turns a recurring slog into a button press. Nothing is a demo for its own sake. We always ask: Does the thing we ship buy back time we’d rather spend on the employee experience?
A lot of internal tooling came out of those sessions. A daily brief triages calendar, email and Slack into a 90-second read. A channel digest auto-summarizes any Slack thread. A helpdesk ticket drafter turns a Slack message into a properly formatted ticket.
Last quarter, we removed the human from the “request to first response to ticket” creation chain. This quarter, we’re on ticket to completion: An approved swag ticket auto orders via the swag API, which is the same for desk booking and shipping labels. Anywhere we can, we’re removing toil.
The posture is the same: Pick what hurts, build the smallest thing that fixes it and ship it to the team’s skill library so the next person doesn’t build it twice.
How does Benchling adapt to change?
The clearest example is how we rolled out AI fluency. When generative AI started reshaping knowledge work, we immediately started building a program around it.
We defined what fluency means. Our AI Fluency Rubric maps four levels — emerging, capable, adoptive and transformative — across five dimensions: velocity and quality, fluency, tools, goals and pace. People know where they stand and what “next” looks like.
We set a concrete target: 100 percent of the team at “capable,” 50 percent at “adoptive,” by the second half of the year. We built the infrastructure: the AI Hub holds all training docs for each transition, a prompt library, a skills library any teammate can contribute to and an AI Fluency Coach, a shared Claude project, that assesses where you are and routes your next move.
We’ve really tried to make it a habit. “Coffee with AI” runs every Friday morning: Someone shares what they built, someone brings a problem, and we run workshops and host an open Q&A. There are additional quarterly workshops tied to transitions in addition to AI Day for protected build time.
We’ve been lucky to have space to change how we think about our work — the rubric, the hub, office hours and build days. We’ve focused on adapting culturally with these rituals and alignment.
Carbon Robotics is revolutionizing agriculture with AI and robotics to reduce costs and increase yields.
What tools support your day-to-day work?
Whatever gets the job done. We care that the job gets done more than how we do it. This is the main concern, although we also need to make sure we don’t end up with too much tech debt or use tools that don’t scale well either financially or operationally. Our stack includes: Golang; C++; Python; Flutter; Dart; Typescript and React; Postgres; AWS; Github; Kubernetes and Terraform; Docker; Prometheus and Grafana; and many other tools, like Jira, Slack, CVAT, Phrase, Zendesk, Auth0 and Postico.
Yes, this pragmatism also means LLMs are a part of our flow, either via Claude, Cursor, Windsurf or others. The main thing that we believe: You should use the tool — don’t let the tool use you, and don’t let it replace your thinking. This is true with AI, as much as it always has been with your choice of programming language or even personal use of social media or other tools.
How does your team experiment?
We get our solution “into the field” as soon as possible.
A boxer once said, “Everyone has a plan until they get punched in the face.” Well, every feature or product seems like a great plan until it gets hit with reality on the farm. This is true whether that’s the weird edge-cases (each farm is unique; it’s the real-world, so there’s infinite variability), or with an unforeseen challenge (flammable fields, tight seedlines, crazy weather, etc.), or just a creative farmer using our product effectively in a new and unique way that wasn’t intended.
We are willing to accept tech debt and the quick hack to learn what the farm really needs and to support our customers. Then, once we hit the “rule of three,” we spend the time building the right abstractions and making something scalable. In any case, we leave most of this to the discretion of engineers. We hire smart people. Why would we waste their time by making them get approval to do their jobs? The agency we give engineers is one of the sources of our strength as an engineering organization.
“The agency we give engineers is one of the sources of our strength as an engineering organization.”
How does Carbon Robotics adapt to change?
We have a mission, a defined culture and smart, hardworking people. With those ingredients, we can adapt to anything. For instance, when contractors were not doing a good enough job of building our hardware, we built our own manufacturing facility to be in control of our own destiny.
As another example, as LLMs have gotten better, we keep evaluating them to see what we can use them for effectively. This year, we created a Claude-powered Slackbot that uses past on-call history to keep already-solved issues from paging our engineers. Again, this doesn’t replace our thinking and expertise. Instead, the use of AI allows us to scale that engineering expertise dramatically, providing better service for our internal team as well as freeing our engineers to do more high-value work. This was not possible even a few years ago, but the tech has changed enough for us to use LLMs in this powerful way.
Pie Insurance offers workers’ compensation and commercial automotive insurance for small businesses.
What tools support your day-to-day work?
I have many different tools that help support my day-to-day work. At Pie, we leverage the Google software suite with Zoom, which is obviously critical for day-to-day work. We also leverage Jira, Confluence and other Atlassian tools for our project management, documentation and software incident response.
At a personal level, I leverage Claude Enterprise and Claude Code heavily for my work as a leader. This lets me pull analytics from Jira and our other metrics tools, like Github, AWS and others, to understand team performance and bottlenecks. As a leader, I’m also forced to context-switch a lot between meetings and topics. To support this, I’ve built up automation through Claude, Google Calendar and my note-taking app Obsidian to manage my notes, action items and calendar in order to stay on top of all my projects.
“I’ve built up automation through Claude, Google Calendar and my note-taking app Obsidian to manage my notes, action items and calendar in order to stay on top of all my projects.”
I, as a leader, also use Claude Code to rapidly understand different parts of our product for any support and technical questions. As I no longer perform software development day to day, these types of tools can let me understand pieces of our software much more quickly.
How does your team experiment?
As an insurance tech company, we leverage the Agile methodology. This focuses on small, iterative development processes across our organization, not just in technology. This also lets us perform experiments to test and learn quickly in the business, but this happens the most in engineering and product design.
For technology, we generally let teams experiment with new technology, processes or ways of working before bringing those results back to leadership and the larger group. This can be new software paradigms, new AI products, etc. The experiments are then presented at team all-hands, technology all-hands or staff meetings for others to discuss and adopt if successful.
Another big function we have at Pie Insurance is our bi-yearly hackathon. These are two-day events where everyone in the company is encouraged to bring ideas to our technology teams for rapid prototyping over two days. These ideas and experiments are then presented to different leaders, including the CEO, which are judged and could be immediately implemented into our products.
How does Pie Insurance adapt to change?
Where we can, especially in technology, we use the Kotter Model for change management to help drive change. This consists of creating urgency, enlisting champions, removing friction and sustaining acceleration, to name a few steps. We have used this model to force change in our AI adoption methodology. Through our product and company needs, we drove urgency, reduced friction by ensuring open access to tools and tokens, and also enlisted AI champions across the organization to mentor and bring others along. We then leveraged data available in some of our analytics tools, like developer intelligence platform DX, to see improved utilization and output from individuals.
For larger corporate-level change, we generally put together a “taskforce” that meets weekly to leverage data to drive changes within an organization or process. We leverage our enterprise data warehouse and data engineering and analytics team to provide this information. We are currently running a taskforce around some operational improvements specific to premium audit.
AlertMedia’s threat intelligence, emergency communication and travel risk management solutions help companies of all sizes identify, respond to and recover from critical events faster and more confidently.
What tools support your day-to-day work?
Technology plays a critical role in enabling our team to work efficiently and effectively.Some of the tools I personally rely on every day include custom Salesforce dashboards for tracking marketing campaign performance, pipeline and revenue, as well as AI agents running in Claude and ChatGPT, which I use for research, analysis and to help manage daily tasks. Across the marketing department, we also rely heavily on our project management system to organize and communicate about active and upcoming projects.
Across the organization, technology is also key to enabling a culture of transparency. As chief marketing officer, one of my top priorities is ensuring that every person in our department has the data and tools they need to make informed decisions and maximize the impact of their work. By providing direct access to key business metrics, every team member can help identify trends, opportunities and issues, which allows us to stay nimble and responsive to evolving market dynamics.
How does your team experiment?
At AlertMedia, experimentation and testing are baked into both our strategy and culture. Our demand-gen team conducts dozens of tests and experiments each quarter to maximize return on investment and reach prospective customers in new ways. Similarly, our product marketing, content and creative teams are constantly testing new messaging, themes and tactics to help contextualize our products in terms that will resonate with buyers. The best marketing is anchored in authenticity and a deep understanding of what your current and future customers care about, which is why we also spent a lot of time testing ideas directly with customers to ensure what we’re bringing to market aligns with what they need to be successful.
“Our demand-gen team conducts dozens of tests and experiments each quarter to maximize return on investment and reach prospective customers in new ways.”
Additionally, we encourage every team member to be curious. We ask our teams to proactively seek out new perspectives, test new tools, and bring back recommendations and ideas for consideration. For example, our entire Marketing and RevOps team recently conducted an AI hackathon, which generated several promising ideas we are already implementing. By prioritizing time for experimentation and learning, we’re able to find new and improved ways of working.
How does AlertMedia adapt to change?
Change is a constant for both our team and the customers we serve. Our mission is to help organizations keep their people safe, which means that when there are life-threatening incidents that require fast action, such as a developing geopolitical conflict or a severe storm, our teams must be flexible and willing to pivot quickly.
During time-sensitive incidents, that may mean rapidly creating collateral, launching a new webinar to help connect customers to subject matter experts or enabling new platform capabilities to facilitate a faster response. What’s great about AlertMedia is that every department can navigate these changes together because we take pride in our work and know how impactful it is for the people who rely on our platform in those moments.
MarketAxess’ platform is designed to make bond trading more accessible.
What tools support your day-to-day work?
My toolkit reflects the balance of engineering management: staying close to the technical work while keeping teams unblocked. On the technical side, I rely heavily on Kubernetes, Kafka and AWS, which power our event streaming and orchestration infrastructure. FreeLens has become a daily driver for managing our Kubernetes clusters, and I spend a lot of time in Git. I’ve also been leaning more on AI-assisted development through GitHub Copilot, which is increasingly useful for code review, exploration and getting unstuck on unfamiliar parts of the codebase.
On the management side, it’s the usual suspects: Microsoft Teams for communication, Jira for tracking, and well-structured one-on-ones. I also use local LLMs to think through problems, draft communications, and explore ideas before bringing them to the team. I have found that having a tool that lets you reason out loud without interrupting anyone is highly valuable.
How does your team experiment?
We treat experimentation as a normal part of engineering, not a separate activity. The teams I lead operate distributed systems where assumptions break in surprising ways, so we build in space to test ideas before committing to them. That includes spike branches, proof-of-concept environments and a culture where it’s OK to say a path didn’t work out.
“The teams I lead operate distributed systems where assumptions break in surprising ways, so we build in space to test ideas before committing to them.”
We also encourage exploration around emerging tooling. AI tooling is the obvious example right now. Rather than mandating one approach, our engineers try different workflows and share what works. Some of the best ideas tend to come from quick experiments that surface in team syncs. I stay hands-on with these experiments myself, both because I want a real point of view on what’s working and because the best way to know whether a tool belongs in our stack is to use it on actual problems.
How does MarketAxess adapt to change?
The teams I lead are product teams, which means we feel priority shifts directly. New initiatives land, timelines compress, and we need to deliver without compromising quality or reliability.
One pattern that’s worked well is reusing proven implementations rather than rebuilding from scratch. When there is a new product request, our first question is whether we already have a pattern, service or component that solves most of the problem. More often than people expect, the answer is “yes.” This mindset has helped us shorten time to market on several initiatives and free up engineering capacity for the truly novel work that requires custom solutions.
It sounds simple, but it requires real discipline. It means investing in shared infrastructure and resisting the engineer’s natural pull toward greenfield work. When it clicks, it’s one of the highest leverage things a platform-oriented organization can do.
GameChanger’s youth sports platform makes it possible to livestream games, schedule practice, review season statistics and more.
What tools support your day-to-day work?
It might sound trivial, but since starting at GameChanger, I’ve had to figure out how to work almost exclusively out of Slack for communications. In my previous job it was 50/50 Slack to email, so I’ve invested in AI channel summaries, scheduled sends and a truly comical number of Slack canvases to stay organized. Claude is my loyal sidekick for summarizing notes, building prioritization arguments and organizing complex projects. Finally, Google Gemini has been helpful for building compelling slide decks to present to my organization.
“Claude is my loyal sidekick for summarizing notes, building prioritization arguments and organizing complex projects.”
How does your team experiment?
Many of our new features are rolled out leveraging web labs to A/B test customer receptivity and engagement metrics. We’re also tapping Claude Code and other AI resources to build quick prototypes to get quick feedback from our engaged community members before investing our full engineering resources. We’re encouraged to move quickly and iterate as we learn.
How does GameChanger adapt to change?
At GameChanger, we strive to be the home of youth sports. Youth sports is a rapidly evolving industry, which means we need to frequently evaluate and refine our product strategy. We adapt to these shifts by keeping the customer experience at the forefront and working backwards from the core GameChanger experience.
Pager Health’s platform is designed to enable healthcare organizations to offer virtual care services.
What tools support your day-to-day work?
Aha, Jira, Claude Code, Figma, Jupyter Notebooks and Python.
How does your team experiment?
Discovery runs as a staged, evidence-driven process with explicit go/no-go decisions at each gate — not a continuous flow where ideas quietly graduate into roadmaps.
It starts with a draft product opportunity assessment. Before we commit any real cycles, we pressure-test whether the opportunity itself is worth pursuing. Is the problem real and painful for a specific market segment? Is it big enough to matter — does the TAM clear our bar? Does it fit our strategy, where we have the right to win, and what are the competitive alternatives? The draft POA forces crisp answers to those questions and drives the Stage-Gate 1 decision: go or no-go. If it can’t clear that bar, it doesn’t move forward, no matter how interesting it is technically.
If we go, we move immediately into rapid prototyping, not to build production software, but to generate enough evidence to either kill the idea or sharpen it. The prototype is a tool for market conversations, not an engineering commitment. We take it directly to customers and target-market buyers: Does the problem resonate? Does the proposed shape of a solution feel compelling? What would they pay for? What would they refuse? We conduct a week of prototyping.
How does Pager Health adapt to change?
The sharpest example right now isn’t a market shift or an org change — it’s how fast AI is collapsing the traditional boundaries between product, design and engineering. The roles we played twelve months ago aren’t the roles we will play tomorrow, and pretending otherwise is the fastest way to become irrelevant.
“The roles we played twelve months ago aren’t the roles we will play tomorrow, and pretending otherwise is the fastest way to become irrelevant.”
A year ago, a PM wrote requirements, a designer mocked the flows and engineers built the thing. That assembly line is breaking down. PMs are now prototyping working software in an afternoon. Designers are shipping functional components, not just Figma files. Engineers are operating at three to five times throughput with AI pair-programming and spending more of their time on architecture and judgment than on boilerplate. The lines between who specifies, who designs and who builds are blurring fast, and the teams that win are the ones where everyone is a builder.
The honest part: None of us knows exactly what a product org will look like in 18 months. The specific skills, the ratios, the titles — all of it is in motion. The adaptation isn’t landing on a new steady state; it’s building an org that can keep re-learning faster than the technology keeps changing.
Axle Health offers scheduling and workforce management software for in-home healthcare providers.
What tools support your day-to-day work?
At our stage, tools are less about standardization and more about speed and leverage. We lean on AI like Claude to go from idea to something tangible quickly, whether that’s a rough prototype, a workflow or just a sharper way to frame a problem, shrinking the gap between spotting an opportunity and putting it in front of users.
That speed shows up most in our product lifecycle. Pylon’s support agent triages B2B customer issues and pipes the signal into Slack, where the team discusses and pressure-tests it. From there, a Linear integration turns a raw feature request into a refined, trackable item in our agile workflow. Discovery to implementation, with automation carrying the handoffs.
We’ve also started building a lightweight “Ask Axle” Claude plug-in and a growing repo of skills and actions where people share useful prompts and workflows. It’s not overly structured, but it helps good ideas spread fast across a small team.
More than anything, the tooling is about enabling people to act. When everyone can build and iterate on their own ideas, we move faster without heavy process.
How does your team experiment?
The point isn’t running “formal” experiments, it’s continuously testing assumptions as we build. A lot of that is learning quickly what won’t work, which lets us stay flexible and put our energy into what will. It keeps us honest about what’s genuinely valuable versus what just sounds good in theory.
We try to keep experimentation as close to the work as possible. Instead of writing heavy specs, we can go straight from a customer insight to a mockup or prototype, but when a customer request comes in, we resist speccing the first description. We break the “ask” down to its raw parts to find what’s actually driving the need, then test against that so we build toward the underlying value, not just the literal request.
Because the team is small, feedback loops are tight. Claude and Figma make it easy to show what an idea might look like and get reactions before investing too heavily. We can put something in front of customers, engineering or leadership the same day and refine it in real time.
“Claude and Figma make it easy to show what an idea might look like and get reactions before investing too heavily.”
How does Axle Health adapt to change?
For us, adapting to change is less about reacting to requests and more about sensing what’s shifting around each customer, such as their market, their regulations or the Electronic Medical Records they live in, and getting ahead of it.
That matters because our customers don’t schedule the same way. A home health agency and a hospice have very different priorities, so the logic our scheduling algorithm is built on has to flex to each, and we integrate with multiple EMRs that each ship changes on their own timelines.
A clear example: A recent federal rule reshaped how hospices schedule patient visits, requiring certain follow-ups to happen within a strict two-day window. We saw how hard that would be for our customers to stay on top of and built our system to automatically flag when one of those visits was due instead of leaving staff to track it by hand before clients even asked.
That’s the discernment we try to practice: understanding the context a customer operates in, not just the requirement they hand us. When an EMR or a regulation shifts, we’d rather have already adjusted than wait to be told.
Gradient AI serves clients that specialize in group health, property and casualty insurance and workers’ compensation, offering AI solutions that perform tasks like predicting underwriting and claim risks and reducing quote turnaround times and claim expenses.
What tools support your day-to-day work?
AI has become a critical part of our daily process. In addition to in-house proprietary models that underpin our products, we’re constantly testing frontier models from every provider across the board, such as ChatGPT, Copilot, Cursor and Gemini, to see how they support different job functions. Recently, Claude and Claude Code have been giving us the most consistent results. Tools like AWS’ Bedrock and Databrick’s Genie allow access to some of the same technologies but with additional guardrails and usage metrics. We’re also exploring targeted solutions for specialized tasks, such as Attention, Granola and BrightHire.
How does your team experiment?
Experimentation feels like the gateway to creativity, which is a quality you can neither mandate nor schedule. We’ve had the greatest success by enabling our natural explorers and early adopters. These are the people that want to be playing with new tools and finding more efficient solutions. Much of what we have to do is get out of their way. There’s a delicate balance between enforcing security with compliance requirements and giving people the space to play. Our approach has been to empower everyone to try new things — vendors, tools and processes — in a separate sandbox environment that isolates company data from exposure. Then, you need a new feedback loop that constantly assesses the latest findings to determine the most value and which should be incorporated into policy.
“Our approach has been to empower everyone to try new things — vendors, tools and processes — in a separate sandbox environment that isolates company data from exposure.”
How does Gradient AI adapt to change?
We strive to remain as nimble as possible to adapt to our changing landscape. GenAI tools are a great example where, regardless of their individual strengths, a month later when new models are released, the rankings get re-shuffled. For us, we’ve tried to balance the desire for cutting-edge results with our business need to minimize churn — migrations between tools — and our need to leverage economies of scale by not having each employee on a different tool’s subscription. This has driven us to write policy defining which tools are in scope for company data versus which are still being evaluated.
