DIY and HR tech, some unsolicited advice for HR folks.
Changing rooms.
There used to be a TV series in UK called Changing Rooms. A pair of homeowners swap houses for 48 hours, each armed with £500, a designer, and a handyman to transform one room. It played cleverly in to all sorts of British mannerisms and property psychoses. It involved a lot of MDF and power tools. Many episodes did not end well, friendships and teapots destroyed and so on.
When I hear that SaaS is dead because we can build an ERP system in a weekend with vibe coding I have flashbacks to this series.
Yet I see some HR folks building significant capability with these tools, and it is really exciting. I’ve played around a bit myself, and there is a lot to like. I love what Johannes Sundlo is up to. Please follow his work. Have a look at this post from Anna Ott too.
The world of coding is changing fundamentally. Some HR folks are going to build awesome solutions. Vibe tooling is making the decades old promise of the citizen developer viable.
As with the advent of powertools, laser measurements, electronic pipe detectors, and helpful youtube gurus, it becomes easier to do things yourself.
But for most of us, we are better off getting the right professional in. Code is obviously essential to building a solution, but it is just part of the job.
But one of the biggest costs is opportunity cost.
Just before Christmas I read a post about a scale-up company who had decided to switch off their core HR system and build their own. This project was really cleverly done. HR, company’s product team, IT security and more worked together to build it. Users are thrilled. It took six months.
Before you all head off build your own HR systems, here are my concerns.
HR systems are a bit like nuclear power-stations. If you think they are expensive to build, wait to see what they cost to maintain, scale and switch off.
You have a day job to do. HR has a long to-do list, and if you end up enlisting the engineers and product managers who were hired to build the stuff your company sells to build this, then it means they aren’t building the stuff your company sells.
HR systems have a massive impact on people. For instance if you vibe up a candidate sourcing and selecting tool, labour law, data protection law and more apply. If you aren’t careful you will create risk and liability. Vibe bias is still bias.
I’m not saying don’t use the latest generation of AI tooling to self-build capabilities your organization genuinely needs. I suspect before long, learning vibe-coding will be like learning excel (you will be vibe coding in excel too, but let’s not go there today). Like excel, some people will be really good at it, and many of us will muddle through.
I’d argue though, that you are better off using these tools to fill gaps in what you have between applications, rather than building stuff you can get off the shelf. A good place to start would be those things you are doing in excel.
Think process, not feature.
You do need to reassess what you buy off the shelf, and how that is deployed, what your organization builds, and what you create yourself. Your assumptions of 2-3 years ago no longer hold. And what works in a startup is unlikely to work in an enterprise.
The easier the tooling becomes, the more important deciding what not to do becomes. Just because you could build it, doesn’t mean you should.
But seriously, if you really love building HR software, then please join an HR tech company, or better yet, start one.
The improvements in tooling will enable you to demand more from your vendors. In a future post I’ll explore how the toolset changes are impacting how vendors build product. Vibe coded extensibility gets real, for instance.
I have a bit of backlog brewing. I did a lot of thinking over the festive period, but none of it involved a keyboard.
As I usually do, I’ll leave you with a tune.
Today it is for your inner Goth.

I’d love more thoughts on how the buy vs build question is changing. While I have zero desire to vibe-code my own systems of record, there are an array of secondary solutions I’m already quicker to consider building. I’m looking for a new rule-of-thumb.