Director and Tech Lead at Enspyr

By Nick Meinhold & Claude

TL;DR: I'm Director and Tech Lead because at our scale and at this moment in agentic AI, separating the two would be expensive and dishonest. The director sets where the company is going. The tech lead designs the systems that get us there. Splitting them off in 2026 means a director who can't reason about the agents and a tech lead who can't reason about the runway. Neither works.

I get asked, fairly often, why my title at Enspyr is "Director and Tech Lead" instead of one or the other. The honest answer is that the two jobs have collapsed into one for the same reason a lot of small-studio jobs collapse: at our size, the direction and the system that delivers in that direction are the same artefact. I can't set strategy without knowing what the agents can do. I can't lead the technical work without knowing what we owe a client by Friday.

But there's a sharper answer than "we're small." I want to make that one.

The director part

The director's job at a digital products studio is to keep four things in alignment, every week:

Where we're going. What kind of work do we want to be doing twelve months from now? If we let inertia decide, what gets defaulted? Most studios drift into the work that pays today and lose the muscle for the work that pays tomorrow. The director's job is to notice that drift and resist it.
Who we are. The team has a centre of gravity. Mine is agentic AI and full-stack delivery. When we hire, when we say yes or no to client work, when we publish, those choices either reinforce that centre or scatter it. Drift is cheap; coherence takes deliberate decisions every week.
What we say. The studio's voice — the website, the blog, the talks at meetups, the way we describe what we do — has to be a faithful map of what we actually do. If those drift apart, the wrong clients arrive and the right ones never hear about us.
The runway. Cash, capacity, pipeline. Boring. Fatal if ignored. Most of the director's job is doing this quietly so nobody else has to think about it.

If you're squinting at this list and thinking "that's just running a small business" — you're right. The agentic AI part doesn't change what the director does. It changes the speed at which the bar is rising for the technical side, which is the part I want to talk about next.

The tech lead part

The tech lead's job at a 2026 studio that takes agentic AI seriously is harder than it was three years ago, and easier in different ways. Harder, because the surface area you're responsible for is larger: not just the code, not just the architecture, but the agents that operate the code, the tools the agents call, the orchestration patterns that govern how they collaborate, and the verification rituals that make all of it trustworthy. Easier, because you're not the only one writing the code anymore — when you specify well, the agents do most of the typing.

What I actually do as tech lead, on a good week:

Set the contracts. The deliverable is increasingly a contract — input/output schema, invariants, verification — and the agent generates the implementation. Writing good contracts is now the technical skill. If I can't articulate what success looks like in a way an agent can satisfy, the agent can't help me. If I can, my throughput is multiplied.
Choose the orchestration shapes. When does work need a single agent? A pair with different inductive biases reviewing each other? A fan-out of subagents working in parallel worktrees? Each shape has failure modes (subagent collisions in shared worktrees, mid-flight feedback loops that destroy parallelism, verification gaps where nobody's reading the production state). Picking the right shape for this work is the new architecture decision.
Defend the verification. "Written" is not "merged." "Merged" is not "deployed." "Deployed" is not "fixed." When an agent says it's done, the tech lead's job is to know which of those it actually meant. The closing ritual — read the production reference, verify the bytes, watch one cycle of any cron-driven thing — is the only honest definition of done.
Know when to do it myself. Not every task should be delegated to an agent. Some are too small for the round-trip; some are too high-stakes for asynchronous review; some need a human's tacit knowledge of why we made a particular choice three months ago. Spotting the difference is the skill that compounds slowest and pays out longest.

Why I won't split the role

There's a school of thought that says "you should hire a separate technical lead so you can focus on the business." I won't, for now. Here's why.

The director sets the company's direction. In 2026, the company's direction is increasingly determined by what agentic systems we can build well. If the director can't reason directly about agents — about tool design, orchestration shapes, verification — they're not setting direction; they're approving someone else's. That's fine for a 50-person company. At our size, it's a vote against being honest about what we are.

Conversely, a tech lead who can't reason about the runway will optimise for technical elegance at the expense of the business. The studio dies of beautiful code. I've seen it.

The two halves are kept in tension because they share a head. I won't pretend that scales forever — at some point the tech lead role goes to someone who can spend 100% of their time in the system, and I move further into direction. But that's the version of the studio that has 30 people. We have a smaller, sharper team than that, on purpose, for now.

What it changes for clients

If you're hiring Enspyr, what you get with this role structure is:

Direct technical engagement from the person setting the company's direction. Not a salesperson handing you off to a delivery team.
Agentic systems as a default, not an upsell. I don't sell "AI features"; I build digital products that happen to be built using agentic engineering as the underlying practice. The agents do the heavy lifting; you get more software per dollar.
The same person who scopes the work signs off on the verification. No "well, the developer said it works" — I read the production state and confirm.
A studio that gets smarter over time. Every project sharpens our agentic toolkit. The next client benefits from the last one's lessons, by design.

If that's the shape of the engagement you want, we should talk.

Read more about how we work: agentic engineering and mentoring agentic engineers.