Fraction ads

Startup Story 35: Engineers are spoiled, and have been for decades. I should know - that was me! It's only just starting to change... I started my career during the dot com bubble when demand for engineers was insane. Salaries inflated rapidly. Companies treated developers with kid gloves. Everyone was desperate to retain and recruit talent. And engineers behaved accordingly. At Intralinks, we had a ping pong table behind a work area with about a dozen people. My friend Dave would serve easy and I'd try to spike the ball to hit employees at their desks. Our CTO constantly talked about creative disruption. So one day my friend Raj shouted "you want to see some creative destruction?" and threw a chair across the room. We had an employee led showing of Office Space just to antagonize management. My cousin at Cisco told me employees brought pillows to sleep under their desks. Not to work 24/7. Just to nap during the day and go home having done nothing. The dot com bust slowed this down temporarily. But from 2003 to 2022 we had an epic bull market in software engineering hiring. An entire generation of engineering managers, PMs and leaders learned to handle developers with kid gloves, and the engineers themselves gained a certain hubris. Then 2022 hit and everything changed. Job postings for engineers fell almost 70% from peak to present. Now we're in the era of 996 culture and people bragging about how hard they work. The engineering culture at large is shifting. I empathize with anyone who lost their role these past few years. But I think this transition is overall healthy for the industry. If you look closely at the latest numbers, software engineering hiring has ticked up slightly. Here's hoping we get a bounce back in hiring. But this time with more realistic expectations from both sides!

Profits up, employment down. Is it AI? Is it the Fed? Something else? The answer is pretty simple here. The share of revenue flowing to corporate profits is at all time highs, while the share going to labor is flat. Basically, in the battle between labor and capital for their share - capital is issuing a Titanic beatdown. This started long before AI. Computers + the internet + smartphones and now AI, every new generation of tech is increasing the ability to do more with fewer employees. This is the essence of technology - saving humans time. But what if it saves us so much time there's nothing left to do?? That's the question for the AI era. My belief: new occupations will come along, but income inequality could accelerate, and the road to the other side will be bumpy...

Hallucinations? Or our next features? Why shouldn't DevHawk be able to pair and walk the Postman flow with a developer? I love that the agent is giving us new product ideas! Of course we have to add guardrails for now, the last thing we want is a Claudius moment where DevHawk tries to go to someone's office in person... But we're actively at work giving DevHawk more tools/MCPs to use in order to be as helpful as possible. Scheduling/calendar work is an obvious part of the flow. What else should DevHawk be able to do?

The difference between an agent that sucks and one that nails it? Context, context, context... Sharing some early versions of our experiments building DevHawk here: In the first example, we had a very detailed prompt telling the agent how to follow up with a developer that was underperforming on certain metrics. Result = a bland/typical nudge message Then we injected tons of context - everything an engineering manager would know about the developer's work. Result = A detailed rundown of technical steps to take, and the ability to go further into that conversation with the developer! With a prompt less than half as long. Bottom line? AI agents need context to succeed!

"He signed up for soccer classes, after having been beat by some eighth grade kids" - Noah Labhart and I getting into the weeds on CodeStory! It was great to sit down, laugh about that, and walk through the origin story of Fraction and now DevHawk. HiddenLevers (used fractional talent from 0 to 9-figure exit) -> Fraction (bringing fractional talent to other dev teams) -> DevHawk (using AI to manage dev teams) So my last three startups form a chain that we talked through - thanks for the sitdown Noah! Link to the podcast in the comments

Fraction is now DevHawk: a wild ride from our start as a fractional talent marketplace to our AI Software Factory today. We started in a different environment - in July 2022 engineering shortages were everywhere! Little did we know that engineering hiring would fall 70% by last July. We grew against the trend, hitting a high with almost 120 folks (fractional and FT) working by Feb 2025. Markets change - so we gotta change with it. First we tried building an Agentic Project Manager, and DevHawk V1 was just that. That didn't take off - nobody wants an AI boss? Who'da thunk it lol! But we started to see success with our AI-driven engineering approach this January, and so the pivot was on. Now there are tons of "AI software factory" products out there, how are we different? We customize the agents and setup for every client - and then we keep running the factory or hand over the keys if they prefer. We'll win if we serve clients well - and the proof is that client satisfaction has been much higher than with the fractional approach. One more bit of big news - I'll take it as a good luck sign - we closed a deal today that takes us past our previous revenue high water mark! Only this time we're a lean team (not 120?!) based mainly in ATL, with an army of agents to boost us - and that's exactly what we're bringing to clients.