How the tech do you build a roadmap in 2025?
Trends to watch, and a practical exercise for determining how to ship product efficiently in an AI world
“Vibe Coding” is in the spotlight right now, but it’s not the most important trend in software building in 2025.
Software development IS changing, though, especially for teams building vertical B2B software. Along with the erroneous idea that vibe coding can now replace legitimate software engineering, here are some other common trends you’ll find in tech blogs and media this year:
Death of SaaS: Many companies are cutting down on the number of software tools they use. Blame is directed to both the economic climate (reasonable) and to AI enabling companies not to need outside software anymore (silly).
AI Adoption: The adoption of generative AI happened faster than any previous human technology. Trying to keep up with all the new capabilities in AI can be dizzying.
Vibe Coding: AI will cut down on software costs by generating code in the hands of non-engineers. The truth is the cost of software development is coming down in the hands of good engineers. Vibe coding may be more valid for non-engineers sometime in the future, but not yet.
AI/Agentic Startup Wave: The AI wave has led to so many startups it’s hopeless to track them all. This was true of the SaaS era also but is compounded now. One of the dominant themes is AI agents that deliver outcomes, not just responses.
AI Attrition Woes: A huge proportion of AI startups that achieve massive adoption curves also suffer massive attrition. Easy come, easy go.
Uncertain Product Strategies: Companies who are not AI-native sense they need to do something relevant to AI, maybe pilot something, but struggle to nail their strategy due to the velocity of changes. In many cases, the classic Innovator’s Dilemma takes hold.
Whether you feel empowered or overwhelmed by all this, the reality is the software landscape is shifting beneath our feet. AI can’t solve everything, but it will leave little unchanged. The balance between configuration and code is no longer a simple either/or decision—it's become the crux of how you design, build, and scale in today’s environment.
Configure vs Code
In 2011, I was a 22-year-old fresh out of college, and I landed my first job in the big city: working at Google in Sales Operations.
In my first year, I built out a bunch of software functionality without the help of any software engineers, despite not being a software engineer myself. Was I vibe coding? Nope. I was configuring (and doing some light coding) in our Salesforce CRM instance. This was internal tech used to track customer pipelines, forecast revenue, and automate certain sales processes, not one of Google's customer-facing products. The internal "customer" was an internal business group of ~100 account executives, account managers, and various other functions.
Similar configuration work is being done by both Revenue Operations and "IT" employees at businesses of all sizes today. Starting with a business goal, these employees scope out the functionality needed by their team, and then get to work customizing third party tools, often integrating multiple external platforms to make it easy to contact customers, harness data, produce quotes... the stuff of "growth".
A few years into my career and with a strong itch to bring my own company to life, I recruited a former Google software engineer to co-found my startup in the real estate finance industry. We scoped out and implemented a ton of custom code. The centerpiece of our offering in the market was a custom designed, custom developed web application that brought the user on a digital journey that we had devised. We had our web app connected to our own internal systems, including a CRM, Google Sheets, Slack, etc. via APIs, but that complexity was hidden from outside users who received an intentionally simple interface.
This is the traditional work of Product Managers in tech companies: determining what custom technology can be built to drive value for users.
The parallel yet different approaches between Revenue Operations (find and configure) and Product Managers (scope and build) are valid ways to achieve different business goals, and rarely intersect one another besides, perhaps, collaborating on how they can share a unified view of customer data between the two camps.
I believe the era of Revenue Operations and Product Management staying in two separate camps may be nearing an end.
Configure AND Code
For years, Product Managers and Rev Ops were like parallel train tracks—heading in the same direction but rarely crossing paths. Product teams were heads-down building customer-facing features through engineering sprints, while Rev Ops was busy in the engine room, stringing together internal systems to boost GTM efficiency.
But the AI era demands something different.
With agentic AI tools, low/no-code platforms, and more composable APIs than ever, the walls are crumbling. Configuration and code are no longer two separate workflows—they’re part of the same blueprint.
Modern teams must plan their roadmap with this blended toolkit in mind. The smartest teams don’t ask, "Do we build or do we buy?" Instead, they ask, "Where can we configure to accelerate, and where do we need to engineer for strategic advantage?"
Whether optimizing internal CRM workflows or shipping AI-powered features externally, this hybrid approach is now mission-critical. The roadmap is no longer a baton-passing relay between Rev Ops and Product—it’s a co-authored playbook.
Today’s builders configure at the speed of thought using no-code platforms, but also dig deep into custom code where differentiation and defensibility live. The magic is in knowing which lever to pull, and when.
Bucketing exercise
To cut through the noise and get practical, here’s a simple exercise founders and business leaders can run to clarify their next moves. Take a business outcome that you’d like to achieve with a tech initiative, and run each component through a waterfall of implementation options:
CRM Configuration: What can you adjust or automate directly within your existing CRM? Think workflows, fields, automations, and user permissions. This requires an understanding of CRM capabilities, so an experienced IT or Rev Ops voice in the room is helpful.
CRM Integrations: What tools and systems could be plugged into your CRM to improve data flow and team efficiency? (e.g., marketing automation, billing platforms, customer support tools). The sales enablement software landscape is constantly evolving, so keeping an eye out for best practices employed by other companies helps tremendously in this step.
API Integrations: What third-party platforms need to talk to each other via APIs? Where can you stitch together best-in-class tools without reinventing the wheel? This may take some research and discovery effort with a technically savvy team member.
Custom Development: Where does your business require unique, custom-coded functionality that off-the-shelf tools can’t provide? This is where you build your moat, particularly if you can leverage deep expertise and proprietary data. It’s the highest leverage, and also the costliest investment, which is why it should be reserved as the final option.
Plotting your projects across these categories, often piece by piece as part of a comprehensive roadmap, brings clarity to optimize tech strategy and reveals gaps where your team might be over-engineering—or under-investing.
If you'd like some outside perspective as you run this exercise, I help founders and teams with this work at SaltandWisdom.com.
We’re still early in the AI era—how you structure your tech roadmap now could define your competitive edge for years to come.