AI is changing how we (and for this article I mean we to specifically be AgTech startups) should think about engineering talent. I have been listening to and reading a lot of content about AI and its impact on businesses. One of the under-recognized aspects is the multiple impacts it is likely going to have on organization-wide engineering for startups. Based on what I’m hearing and reading, I believe the following are some of the key areas for change:
- Improved coding efficiencies are available – you have to figure out what that means for you development team. Yes, the key first point (as it has been for months to now almost years for many organizations) is the obvious truth—the AI coding tools can do a lot more coding a lot more efficiently. So, the success metrics around output for engineers needs to change. What one engineer could do from a product roadmap perspective for software needs to be looked at. Can you reduce your product roadmap in terms of cost or time and by how much? In some cases, you’re not going to know which variables are impacted and you’re just going to have to work on the product roadmap revisions in real time.
But there’s no doubt based on the successful revenue growth in the coding segment by the key players that software efficiency gains are available, and the key for most startups is to figure out what your gains are likely to look like. There are multiple podcast episodes across the tech segment talking about 10-20% efficiency gains in weeks for coding, quality control (QC)/quality assurance/testing functions, and others and in some cases continuing to lower by another 10-20%. You start to scale that across even a small engineering team and you can really have an impact on the product roadmap.
2. The cost of engineering talent is going up. We’ve all seen the Steph Curry-like contract figures for some of the true AI unicorns. Those aren’t the people I’m talking about. They are the ultimate outliers and they are hunted like NBA free agents that can shoot the 3 and have enough defensive skills to guard the 3 point line as well (the “3 and D” crowd, if you will) and they are surprisingly rare and therefore increasing in value (come on, Trae Young just signed for $212M in DC and I’m not even sure he plays defense, so you’re not getting the same value for $212M you got just a few years ago). No, I’m talking about the journeyman players out there that have nice leisurely all-star weekends but can find their way onto NBA rosters for 8-10 years or more by maxing out their skillset and hoops IQ. In this case, think the 5-10 year coding veteran that has a year or two of AI coding tools and knows his or her way around all the cool kid toys, knows what each model does, and can implement them at scale with very limited managerial support. A coding veteran with some skills in that arena that can leverage AI coding tools effectively is commanding premium value in the marketplace today. So what used to be a $200-300k developer is now in many cases a $350-$450k developer or more. So figure out where market is before getting that job description together and be brutally honest about what you can get for your money out there today. Recruiting AI talent is only getting harder and more costly.
3. And that engineering talent is going to need some care and feeding – and by care and feeding I mean AI tokens, lots and lots of AI tokens. So now you’ve hired that rock star $500,000 AI developer that knows all the latest tools and you’re just hoping they’re not a high maintenance prima donna (but you’re almost ok with that – they’re that good – or might be). But that’s not the end of the story. To quote the late great John Pinette, “I say nay nay!”
Jensen Huang (yes, that Jensen, Nvidia CEO) was on one of the tech podcasts recently and he said (paraphrasing slightly) “if I’ve got a $500,000 engineer and he’s not using at least $500,000 of AI tokens a year, we’re going to need to have a chat.” His point is why would you bring on A-grade talent and pay them A-grade talent money and then give them the coding equivalent of a Chevy Malibu. The key insight behind that point is you want to give that $500k developer another $500k+ in tokens because if they’re that good they should turn that into 3-5x performance gains and get the product available to sell faster to create incremental revenue opportunities all over the place.
From most of the CEOs, founder, and funders I’ve been listening to, it seems like most of the early action around AI coding was working hard to reduce engineering expenses through the use of coding tools. Once that was achieved, the focus for some of the more aggressive CEOs started to focus on leveraging AI further to create even more of a competitive advantage. Code that wasn’t even on the roadmap or was low priority can now get developed and released or re-prioritized.
It’s a little more nuanced for AgTech startups, but the principle remains the same. You need to balance out the level of talent you’re recruiting with the level of support (tokens) you want to give them to create the multiplier or flywheel effect for the engineering team. Maybe $500,000 each isn’t the right capital model for your engineering team, but figure out what it is and then hire and support to that level. And as with general tech, AgTech startups will have the choice to go for coding cost savings or be more aggressive with roadmap changes and quicker release time frames.
4. Retention is getting tougher as well. I already know of multiple startups that have lost candidates in recruiting mode to AI startups because of a much higher cash compensation and some AI equity that, if we’re being honest, is probably a bit to a lot more likely to turn into something meaningful on the AI side of the world then many AgTech startup equity grants. So you have two choices during the recruitment phase: (1) take the A player that is a 5-tool athlete but a known flight risk if they look or sometimes get an interesting in-bound from a recruiter after a rough day or week; or (2) take the A-minor play that is a 4.5-tool athlete but less of a flight risk (and maybe more of a team player because he realizes that can almost nudge you up to an A in a lot of situations). I’m not suggesting either is right for every startup. Your risk tolerance and self-awareness about how hard the A-talent is to replace will have a lot to say about where you come down on this one.
So that’s a few of my thoughts on the impact AI is having on tech and AgTech. I’ll be putting some additional posts out on the impact of AI in both horizontal tech segments and AgTech as a key vertical.