From the Trenches: The Misconception About AI and Software Development

AI can generate code faster than ever. But building products that scale, improve, and help organizations make better decisions still requires product judgment, domain expertise, and teams who know how to turn messy feedback into systems.

As a Head of Product working in sports technology, I have started to notice a misconception growing across the industry.

It usually sounds something like this:

"Why should we buy software if AI can build it for us?"

I understand why that thought exists.

The tools available today are remarkable. Claude Code, ChatGPT, Cursor, v0, and other AI development tools have dramatically changed how fast software can be created. A feature that once took weeks can now be prototyped in hours. A dashboard can be generated from a prompt. An internal workflow can be mocked up quickly. A non-engineer can now participate in the creation process in ways that were not possible a few years ago.

That is a massive shift.

But from the trenches of actually building products, talking to customers, organizing feedback, prioritizing work, shipping features, fixing edge cases, and supporting real users, I think many organizations are drawing the wrong conclusion.

They are confusing the ability to generate software with the ability to build, maintain, and continuously improve a real product.

Those are not the same thing.

The Real Misconception

The misconception is not that AI is useful.

AI is incredibly useful.

The misconception is believing that because AI can help generate code, software development is now easy.

Software development has never just been about writing code.

It is about understanding problems.

It is about organizing ambiguity.

It is about taking loose ideas from multiple stakeholders and turning them into something real, usable, scalable, and valuable.

Customers rarely show up with perfect requirements.

They show up with frustrations.

• "This takes too long."
• "I wish this was easier."
• "We have all this data, but we don't know what to do with it."
• "Could we compare players this way?"
• "What if we could automate this?"

Sometimes they have a strong idea. Sometimes they have half of an idea. Sometimes they know the pain point but have no idea how to solve it.

In fact, some of the most honest and productive customer conversations eventually reach the same moment:

"I don't know how we would actually do that."

And that is completely fair.

Their expertise is coaching, recruiting, operations, player development, roster management, or team building.

Their job is not to design scalable software systems.

That is where product teams create value.

The Bottleneck Was Never Just Coding

AI is making coding faster.

But the bottleneck in product development was never just coding.

The bottleneck was translating real-world problems into systems.

Someone still has to determine what the customer is actually asking for.

Someone still has to separate the symptom from the root problem.

Someone still has to identify patterns across dozens or hundreds of conversations.

Someone still has to decide what should be built now, what should be delayed, and what should not be built at all.

Someone still has to think through architecture, data models, permissions, workflows, edge cases, integrations, testing, support, onboarding, and long-term maintainability.

AI can accelerate that process.

It does not eliminate it.

If anything, AI makes strong product thinking more valuable because the speed of execution increases. When teams can build faster, bad decisions can also ship faster.

The quality of the thinking matters more, not less.

Software Engineering Is Not Going Away

I do not believe software engineering jobs are simply going away.

I believe the role is evolving.

The best engineers will not just be people who write code manually line by line. They will be people who know how to use AI as leverage.

They will understand architecture. They will understand systems. They will understand how to review AI-generated code. They will understand security, infrastructure, testing, deployment, and performance. They will know how to prompt, evaluate, debug, and improve.

Prompting will become another engineering skill, similar to version control, APIs, cloud infrastructure, or automated testing.

The same is true for product leaders.

The best product people will be the ones who can combine customer discovery, domain expertise, AI fluency, technical understanding, and execution discipline.

In my view, AI does not reduce the need for talented people.

It increases the demand for people who can operate at the intersection of domain knowledge, product judgment, software development, and AI.

That combination is still rare.

The Scarce Talent Is Systems Thinking

This is one of the biggest things I think people underestimate.

There is a real lack of talent in the market for people who can think in systems.

People who can sit in a messy conversation with multiple stakeholders, hear ten different ideas, separate what matters from what does not, and turn it into an actionable plan.

• People who can take a loose thought and turn it into a workflow.
• People who can take customer feedback and turn it into a roadmap.
• People who can take product strategy and translate it into engineering tickets.
• People who can understand the sport, the business model, the data, the user experience, and the technical constraints at the same time.

That is hard.

It is also exactly the type of work that becomes more important in an AI-first world.

AI can help write the ticket. AI can help generate the component. AI can help summarize the meeting. AI can help produce the first version.

But someone still has to know whether the idea is worth building. Someone still has to know how it fits into the system. Someone still has to know whether it solves the real problem. Someone still has to turn scattered feedback into execution.

That is not just prompting.

That is product judgment.

"We'll Just Build It Ourselves"

Will some organizations successfully build internal AI tools?

Absolutely.

In fact, they should.

AI has dramatically lowered the cost of solving specific internal problems. If a staff needs a small recruiting dashboard, a custom report, an internal workflow, or a one-off automation, AI can make that more achievable than ever before.

That is a good thing.

But there is a fundamental difference between building an internal tool and building a product.

• An internal tool only has to work for one organization. A product has to work across many organizations, many workflows, many user types, and many edge cases.
• An internal tool can be rough around the edges. A product needs to be reliable.
• An internal tool can be maintained by the person who built it until they leave. A product needs documentation, support, onboarding, infrastructure, testing, monitoring, security, and a roadmap.
• An internal dashboard can answer one question for one staff. A product has to continuously evolve as the market changes.

That is the part many buyers underestimate.

AI lowers the cost of the first version.

It does not eliminate the cost of ownership.

What Sports Organizations Are Actually Buying

Sports organizations often think they are buying software.

In reality, they are buying a team's accumulated ability to turn hundreds of conversations with coaches into systems that actually improve decisions.

• They are buying product judgment.
• They are buying lessons learned from failed ideas.
• They are buying workflows that have already been tested in the real world.
• They are buying edge cases another organization already ran into first.
• They are buying data models that have been refined over time.
• They are buying the ability to take feedback and turn it into production-ready improvements in weeks, not months.

That does not happen just because AI writes code faster.

It happens because a thoughtful and dedicated team exists at every level: engineers, product leaders, designers, data scientists, customer success, domain experts, researchers, and analysts. People who understand the sport. People who understand the data. People who understand the customer. People who understand what can actually be built, supported, and scaled.

AI makes those people more effective.

It does not replace the need for them.

The Differentiator Is Easy to Miss

This is the part I think the market may miss.

The differentiator in great software is rarely visible on the surface.

A buyer sees the interface. They see the dashboard. They see the report. They see the feature.

But behind that feature may be years of work. Someone modeling the data correctly. Someone with advanced statistical knowledge shaping the methodology. Someone with unique domain expertise identifying what actually matters. Someone maintaining the infrastructure. Someone managing operating costs. Someone thinking about scalability. Someone handling customer support. Someone making sure the product does not break when more teams, leagues, sports, users, or data sources are added.

The value is not just the screen the customer sees.

The value is the system underneath it.

That is what AI does not magically create out of thin air.

A prompt may generate a dashboard. It will not automatically generate the right model, the right workflow, the right support system, the right architecture, the right customer feedback loop, or the right product judgment.

Those are the differentiators. And they are easy to underestimate until you have to build and support the product yourself.

The Hardware Contradiction

There is another contradiction I see in the sports industry.

Some organizations are becoming more hesitant to invest in software, yet they continue to invest heavily in hardware: GPS systems, wearables, tracking cameras, sensors, video systems, and tagging tools.

Those technologies can be valuable. They collect important data.

But hardware is becoming increasingly commoditized.

The harder problem is no longer collecting data. The harder problem is interpreting it.

Many coaches and organizations are not suffering from a lack of information. They are overwhelmed by it. They have more film, more tracking, more tags, more reports, more metrics, and more dashboards than ever before.

But the core questions remain:

• What does this mean?
• What should we change?
• Who should play?
• Who is improving?
• Where are we exposed?
• Which recruit fits our system?
• What decision should we make next?

Hardware can help tell you what happened. Great software helps you decide what to do about it.

That is where the next wave of value will be created. Not in collecting more data for the sake of collecting data, but in turning data into decisions.

AI Raises the Standard

AI will not make software less valuable.

It will raise the standard for what software should be.

Static dashboards will not be enough. Basic reports will not be enough.

Organizations will expect software to explain, recommend, predict, automate, and adapt. They will expect products to improve faster. They will expect their feedback to be heard and acted on faster. They will expect software companies to understand their workflows, not just sell them tools.

This is where strong product teams will separate themselves.

The companies that win will not simply be the companies using AI. Everyone will use AI.

The winners will be the companies that combine AI with domain expertise, customer feedback, strong engineering, data science, product discipline, and operational excellence. They will be the companies that learn faster than everyone else.

The Reality Check for Buyers

AI should absolutely change how organizations think about software. It should make buyers ask better questions.

But the question should not be:

"Can AI build this?"

The better questions are:

• Who will maintain it?
• Who will improve it?
• Who will secure it?
• Who will support it?
• Who will interpret the data?
• Who will understand the edge cases?
• Who will turn user feedback into the next version?
• Who will still be accountable when the prototype becomes business-critical?

That is the difference between generating software and owning software.

It is also the difference between a tool and a product.

My Prediction

I believe we are entering a period where more software will be created than ever before. More internal tools. More prototypes. More niche applications. More automation. More experimentation.

That is exciting.

But I also believe the gap between average software and exceptional software is going to grow.

AI will make it easier to build something. It will not automatically make it easier to build the right thing.

The organizations that succeed will not be the ones that assume AI replaces product teams, engineers, data scientists, or domain experts. They will be the ones that use AI to amplify those people.

They will understand that code is becoming easier to generate, but product judgment is not.

And in sports technology, that distinction matters.

Because the future will not belong to the organizations with the most data. It will not belong to the organizations with the most hardware. It will not even belong to the organizations with the best prompts.

It will belong to the organizations that can turn expertise, feedback, data, and AI into systems that help people make better decisions.

That is the real future of software development. And it is a future that still requires great people.