How Crewmojo Scaled a Lean Dev Team with Tekkon Engineers and AI-Assisted Coding
A conversation with Joel Duckworth, CTO at Crewmojo
The Setup
Crewmojo (based in Sydney) is a performance management platform built by a small, tight engineering team. Like most startup CTOs, Joel was being pulled in two directions at once: writing code to ship features, and doing the architectural thinking the business actually needed him for.
His solution?
Bring in a few awesome Tekkies (Tekkon Engineers) and pair them with AI-assisted development from day one.
Here’s how it’s played out.
Cursor From Day One
When their first Tekkie, a frontend developer, joined the Crewmojo team through Tekkon, Joel handed him one instruction: use Cursor for everything.
“There’s no point trying to code it yourself. It’s just way too slow now.”
There was no formal onboarding. No hand-holding through the codebase. Joel pointed him to YouTube, told him to watch a few videos, and said if he wanted to understand how anything worked, ask Cursor, not me.
“I just filled in the blanks where I needed to, which was pretty minimal.”
That set the tone for the whole engagement.
Frontend Devs Delivering Full Stack
At first they were cautious and tested our model with one Tekkie (or Tekkon frontend engineer), and once they were happy with the way that developer had worked out – they scaled to 2 more frontend devs. Bringing them to 3 in total.
From the outset, all 3 Tekkies came on as frontend specialists. But that didn’t last long.
Using Cursor to explore the codebase, the first Tekkie they hired started working across Go APIs and Python-based AI services, languages he had limited prior exposure to. Out of their other 2 developers, one is now consistently delivering full stack feature enhancements. the other, focused on frontend, has expressed interest in expanding further.
Joel’s take: “I’ve just been chucking them in the deep end as far as I could and seeing where their limits are.”
The AI tools didn’t just speed up delivery. They compressed the learning curve. Developers no longer need to memorise every syntax detail of a new language. They need to understand how the pieces fit together, and then direct the AI to execute.
“You don’t need to understand every line of code. You just need to understand how it fits together.”
What the New Workflow Actually Looks Like
Joel’s process has evolved into something that looks less like traditional development and more like architectural direction:
- High-level guidance goes to the team, sometimes a Loom video, sometimes a spec doc generated with Cursor, sometimes a simple checklist
- Engineers use Cursor to explore the codebase, frame the task, and generate implementations
- Joel focuses on architecture, integration, AWS cost implications, and whether components will actually play nicely together
- A Cursor bot runs automated security reviews on every release PR
The daily standup holds it all together. Usually 15 to 20 minutes. Sometimes five. Sometimes an hour. Joel doesn’t skip it.
“Even a quick five-minute catch up is better than nothing.”
The Role of the Developer Is Changing
One of the more pointed observations from the conversation was around what kind of developer actually thrives in this setup.
It’s not the person who just wants to put AirPods in and write code all day. And it’s not just someone who can talk to a client and ask an AI to do the rest either.
The sweet spot, as Joel describes it, is someone who understands architecture well enough to know when the AI is going down a bad path, and who can communicate clearly enough to guide it back.
“Low-level coding is becoming commoditised. The value now is knowing how to structure the solution.”
That also has implications for hiring. Years of hands-on experience in a specific stack matters less than it used to. Quick learners with strong problem-solving instincts, who are genuinely curious about new tools, are the ones getting the most out of AI-assisted development.
What’s Coming Next
Joel flagged two things on the horizon worth paying attention to.
First, the mono repo. Crewmojo currently has 10 to 15 separate code repos. AI tools can only see what’s in front of them, so reviews and feature builds are fragmented across multiple PRs. Consolidating into a single repo means the AI can operate across the full stack in one pass, with full context. Game changer for both development speed and review quality.
Second, testing. As the build side accelerates, the verification side becomes the bottleneck. Joel sees QA and testing experience becoming increasingly important for developers, not as a separate function, but as part of the profile.
“The actual building is going to be massively unleashed. You need to make sure what you’re unleashing is good.”
The Bottom Line
Joel put it plainly when asked whether he could quantify the productivity impact of AI tooling:
“I can’t imagine not having it now. It would be painfully slow. It’s like going back to typing on an analog typewriter.”
Three Tekkies. AI-assisted workflows. A CTO finally able to focus on the architecture work that actually moves the company forward.
That’s the setup. And they’re not going back.
Tekkon supplies seriously good developers from Nepal who embed directly into your team, and as Crewmojo shows, pair brilliantly with AI-assisted workflows.
Want to explore how a Tekkie or two could slot into your team and help you scale? Get in touch – we’re always up for a coffee or chat!