Co-Written by Britt Gage, of All Trades & Hannah Griffin, Visible Hands
There’s no hiding the fact that AI is changing the way we all work. Whether it is directly changing our day to day or just beginning to make small waves in the work done around us, it’s important to consider what the future holds as it relates to team structure.
The conversation around hiring and AI has come up a lot for us lately. We often hear questions from our community like “How do I know when to hire a real person versus just using AI?” or “What if I don’t know exactly who to hire for? Should I just use AI?”.
If you’re running a startup or small business, you’re facing dozens, maybe even hundreds, of decisions daily. When you reach a point where you know you need support, but don’t know the direction to go, it can feel overwhelming. There’s no doubt that AI can support your workflow and make things more efficient for you as a founder, however, it’s also important to really consider when AI should be a tool and not a solution.
The problem for most early-stage startups is that only a small fraction of the daily work being done fits neatly into buckets. Startup workflows tend to be messier: unclear inputs, undefined outputs, problems that cross every boundary you've drawn. You're figuring out pricing, rebuilding onboarding, trying to understand why your retention dropped last month, really understanding your customers’ pain points, etc., all while maintaining day-to-day business functions. It’s exhausting.
But, oftentimes, there is beauty in the chaos. This is the work that moves companies forward; those collaborative moments when you’re trying to solve for X and actually come up with a creative solution for Y.
When companies, specifically startups, try to fit challenges or people into boxes, there can be a structural mismatch. You hire a growth marketer because you need to scale acquisition. A few weeks in, it becomes clear there’s a different problem you’re facing. You have broken onboarding, unclear positioning, no shared understanding of which customer actually converts. The growth marketer is capable, but the conditions for growth marketing don't exist yet.
As a founder of behavioral intelligence platform Mapp, Sarah Lucena, shared this experience after going through this herself.
"I was hiring for what I thought were the right skills. It looked great on paper but didn't really work in real life." (TechCrunch Build Mode, February 2026)
This is worth naming not because it's a failure of judgment, but because it's a failure of design. The org chart creates a kind of shared assumption that someone else owns the work.
There's a version of this conversation that treats AI as a solution to the problem above. And in some ways, it is. If a task is well-defined, repeatable, and has clear inputs and outputs, AI can handle it at a scale and speed that fundamentally changes the economics of building a team.
But that framing can become a trap. AI is only as useful as the clarity of the work you're pointing it at. It executes brilliantly, but it lacks human discernment. It doesn't diagnose. It doesn't know which problem to solve first, or whether the problem you've named is the real one. That part still requires a human who can hold the whole picture.
It also requires maintenance. Once built, these agents need consistent updating and support to ensure they’re working the way everyone expects and needs. It’s common but not true to assume that these can be built to “set-and-forget.” Regardless of the builder’s technical competency, AI just isn’t there yet.
The ever-so-common "just implement AI, don't hire" narrative, while directionally interesting, misses something crucial. The question isn't whether to hire. It's what you're hiring for, and whether your model for organizing work is actually suited to the stage you're in.
What we'd suggest instead is thinking about your organization less as a chart and more as two modes running simultaneously.
One mode is built for stable, repeatable work: the processes that have clear inputs and outputs, that benefit from specialization which can be systematized and eventually automated. This is where a traditional org structure makes sense, and where the right specialist hire creates real leverage.
The other mode is built for everything that isn't as clear: the ambiguous problems and the cross-functional work; the things you're figuring out as you go. This is harder to hire for, because the job description doesn't really exist yet. But it's often where the most consequential work happens.
McKinsey's recent research on what they call "the agentic organization" points in a similar direction, arguing that companies need both deep specialists for stable functions and broad generalists who can orchestrate work across domains as AI reshapes what those domains look like. (McKinsey, "The Agentic Organization," September 2025)
The shift worth considering isn't choosing one mode over the other. It's building an organization that can hold both and knowing, in any given moment, which one the work actually calls for.
The word "generalist" can mean a lot of things and the version that's useful here is specific.
While a common mental model for a generalist is “someone who does a little bit of everything without going deep on anything,” the term is also applicable to the T-shaped operator: genuine depth in one area, paired with enough range to work meaningfully across adjacent functions. Your startup can benefit having someone who can lead a customer research project, inform product decisions, pressure-test positioning, and help you think through your hiring strategy. They’re the perfect orchestrator and implementer of AI crossfunctionally in a way that truly makes an impact.
These people are genuinely rare. They're also, in our experience, exactly what early-stage companies need most right now because they can handle the work that doesn't fit the chart. They can diagnose before they prescribe. They can move between strategic and tactical without losing the thread. And critically, they can work alongside AI in a way that multiplies their output rather than outsourcing their judgment.
Reuven Gorsht, building across real estate law, technology, UX, and marketing simultaneously, described this well. With AI as a tool, a generalist can now "execute at a level that previously required entire specialized teams." This doesn’t mean they know everything, but they know enough to direct the right tools and ask the right questions. (Medium, January 2026)
At Visible Hands, when we look at founding teams, one of the things we're paying attention to is whether someone has this quality. It’s not the title, or the pedigree. The actual capacity to operate in ambiguity, across functions, without waiting for perfect conditions tends to be a strong signal both of what the founder can do and of the kind of team they'll eventually build.
None of this means specialists don't matter, or that structure is the enemy. It means the question worth asking before any hire is a more fundamental one: what kind of work is this, and what kind of person (or tool) is actually built for it?
If the work is defined and there's a repeatable process, clear outputs, a function that needs to run reliably, then a specialist is the first solution. Or maybe you’re ready to fill that need with an AI agent. These can work deeper and faster than a generalist would.
If the work is still being shaped and if the inputs are unclear, or if the problem crosses multiple functions and you're still figuring out what “good” looks like, then you probably need someone who can hold the ambiguity, diagnose the real problem, and build the conditions for more specialized work later.
Shopify made a version of this argument at the organizational level in 2025, directing managers to demonstrate that AI couldn't handle a given function before adding headcount. The instinct is worth borrowing: before you hire, be honest about what the work actually is. (Techclass.com, "Building and Leading Effective AI-First Teams," January 2026)
The org chart can still be a viable place to begin team building. Structure matters. But it works best when it's built around the work you're actually doing, not the work someone imagined you'd be doing at this stage.
The underlying question is consistent: does your structure serve the work, or constrain it? And as AI continues to reshape what work even means, that question is worth revisiting more often than most founders think.
Visible Hands is a venture capital firm that invests in exceptional, overlooked founders building high-growth startups. of All Trades helps early-stage startups find and hire T-shaped operators, building dynamic teams outside of the traditional org chart for the AI world of work.
