With coding ‘solved’ by AI – the role of a good CTO is about to change dramatically
This week I hosted another CTO roundtable with Valor portfolio CTO’s from across the globe (US, Colombia, and Italy!). Our previous roundtable was just four months ago – when we were still marveling at the first AI reasoning models, the meme ‘Vibe Coding’ was only weeks old and MCP was not a thing yet! The progress since then … I think coding is more or less ‘solved’ today as far as AI is concerned and soon (very soon!) most coding will be done by AI.
This means the role of the CTO will evolve from managing code production to orchestrating AI-driven systems across the entire business. The CTO will be the new Chief Automation Manager
Here are the key takeaways from our discussion on the tools and talent
The Modern AI Toolbox: From Agents to Context
Everyone at the table was using a suite of AI coding tools, from Devin and Cursor to GitHub Copilot. But the conversation quickly moved past brand names to a more crucial point: success isn’t about the tool, but the setup. Thanks largely to MPC (and increasing context windows of the base models) lack of context is no longer holding back development. And tools like Devin are able to not just code, but also run and test the code they build.
Erik shared a powerful use case. His team uses an MCP to give their AI agent direct, read-only access to their PlanetScale database. Instead of manually pulling IDs or running queries, his developers simply ask the agent.
Julian, uses MCPs to solve the “outdated knowledge” problem. By connecting an agent to the latest API documentation, it stops generating code for old, deprecated versions—a common frustration that can halt development in its tracks.
The Human in the Loop: Redefining the Developer’s Role
Integrating these powerful tools isn’t always a smooth process. Paolo perfectly described the adoption journey his team experienced as a “multimodal curve”—starting with resistance, moving to over-enthusiasm that created over-complex code, and finally settling into a productive, balanced workflow.
This journey changes what we need from our developers. The most valuable skill is no longer just the ability to write flawless code, but the ability to guide, validate, and prompt an AI to do it for them. As Paolo noted:
“What we need to do now is, I think, no more looking for a very good coder, but for a very good prompter… people that are able to put the thing together more than building the little piece.”
This shift completely changes how to hire. The traditional coding challenge is becoming obsolete. Why test for a skill that can be automated? Instead, we need to test for the ability to leverage AI effectively. Erik shared his innovative hiring process:
- Initial Screen: Filter hundreds of applicants down to the most qualified.
- Paid Hackathon: Invite a small group to a six-hour, paid hackathon over a weekend where they are encouraged to use AI tools to build a project.
- Paid Internship: The top performers from the hackathon are offered a one-month, paid part-time internship to see how they perform in a real-world environment.
- Full-Time Offer: Finally, one or two are brought on full-time.
This process finds not just coders, but true AI-native problem solvers who can deliver results.
Is Coding a “Solved Problem”? The Rise of the AI Architect
This led to the central debate of our discussion: if an AI can write the code, is coding a “solved problem”? The group’s consensus was nuanced. The physical act of writing code is rapidly becoming commoditized. However, this only elevates the importance of higher-level skills: software architecture, system design, and rigorous validation.
Zach, compared it to the difference between an architect and a senior developer. The AI can build the components, but it often lacks the “bigger picture” and can introduce subtle biases based on its training data.
This is where the human-in-the-loop becomes a crucial guardrail. Julian issued a critical warning about the risk of AI massively accelerating the accumulation of technical debt. Without a skilled human validating the output, an AI can quickly produce a “very patchy, very complex to maintain” codebase. The speed is seductive, but the long-term cost can be enormous.
The New Frontier: Orchestrating AI Across the Business – the new role of Chief Automation Officer
Perhaps the most significant evolution of the CTO role is its expansion beyond the engineering department. Tech leaders are now in the best position to deploy AI to create value across sales, marketing, and customer success.
Zach shared a perfect example. He repurposed an AI chat extension originally built for customers into an internal tool for his sales team. By feeding it a prospect’s URL, the tool instantly generates tailored value propositions and a sales script, highlighting exactly how their services can help.
At Valor, we’ve implemented a similar “funnel” methodology to manage our deal flow. We treat every interaction as a data source. Every call with a startup is recorded and automatically summarized into key “building blocks” (Product Story, Team Story, Revenue Story, etc.). This structured data feeds a living “one-pager” for each company, allowing our team to make faster, more informed decisions without getting lost in unstructured notes and documents.
This is the new strategic frontier. The goal is to capture every interaction—every sales call, support ticket, and customer email—and turn it into a structured asset that can be used to automate other parts of the business, from driving your product roadmap, to better customer support. If every sales call, every support call, every support ticket, every customer email becomes part of your ‘data’, AI will be able to help along the way with every step. Customer onboarding that is ‘smart’ because it knows so much about the customer already from all those (pre) sales calls. Roadmaps that automatically prioritize items based on support interactions. If coding is ‘solved’ it’s easy to see how the processes before and after coding are just as much up for AI automation.
From CTO to Chief Automation Officer
AI is already transforming how software is built. But it will transform every aspect of a software company over the next few years. A good CTO will lead that transition is the Chief Automation Officer. You will help the company build the internal data collection processes needed, train the talent, and champion an AI-first mindset.
A few examples of things on the internal roadmap for our roundtable participants: automating quarterly business reviews, improving AI-driven lead generation, and building early-warning systems for customer success. It’s the CTO moving up the value chain from overseeing building software to overseeing building the businesses of tomorrow.