Friends,

It wouldn’t be a stretch to say that one of the favourite parts of my job is getting the chance to meet with my community. People like you reading this newsletter.

One of the things I miss the most from working in house is the chance to see what they’re working on, and what challenges they’ve overcoming and just talk it out.

One of the things they don’t put on the brochure when you become a solopreneur, is just how isolating the solo part actually is. It’s something I have to consciously solve for in how I run my business, because left un-checked, it can genuinely start to bring me down 😢

Suffice it to say, when I go to events, tee up chats and actually get to connect with those of you working in the HR space, I’m buzzing for days. (I guess that makes me an extrovert)

Sometimes, those chats are so good I leave with a whole new appreciation for where the function is going and the incredible things people are doing in it. One such meeting was one I had at an event a while back, with Nilan. So much so, I invited him to share a bit of his thinking with you all, through today’s edition.

Enjoy this week’s edition ✌️

Matt

PS: I know how many reading this are themselves standalone HR folks, which is another kind of isolating sometimes. If you ever want to chat with someone, just hit reply.

In case you’re new here (or just missed it) here’s the past three editions of the FNDN Series:

Every HR process lives in one place, from your people directory and time off to performance cycles and analytics, with no spreadsheets or scattered systems in sight.

AI agents handle the admin load end-to-end: answering employee questions instantly, running onboarding seamlessly, and flagging patterns or issues that need your attention.

Using Vibe Views, any people leader can build the dashboard or tool they've always needed, just by describing it. No engineer required.

Connect your people data wherever AI works, through MCP, enabling deep analysis, workflow triggers and custom tooling without ever exporting sensitive data.

One clear price.
Quick to implement.

Built for HR leaders ready to turn off the noise and turn up the strategy.

Interested in sponsoring the FNDN Series? Drop me a line!

Know a startup Head of People looking for answers 🙋 why not forward this to them for some instant karma? ✨

From Matt: I’m excited to welcome this weeks guest contributor, Nilan. I had the chance to meet Nilan in person a few months ago and was so impressed with what he was building with AI at Zip. So much so I asked him to share his thoughts here so more of us can do the same.

PS: If you’re interested in being a guest writer for FNDN Series, let me know. I'm a big fan of platforming people doing amazing work.

A lot of growing companies don’t really have one coherent People system.

They have:

On paper, that stack can look workable.

In practice, the People team often becomes the integration layer between it all.

A request comes in through one channel.
Context sits somewhere else.
Approvals happen in inboxes or Slack.
A document gets drafted manually.
A spreadsheet gets updated.
The HRIS gets changed at the end.

If anything changes halfway through, someone has to work out where the truth now lives.

That is the operating problem I keep seeing.

Not that companies lack systems.
That they are often designing around the final record rather than the decision path that produces it.

And that matters because a lot of People work is not hard at the point where the record gets updated.

It is hard earlier.

When context needs to be gathered.
When different people need different visibility.
When approvals depend on the type of case.
When exceptions matter.
When the final decision needs to be defensible.

When I say designing HR, I don’t just mean programmes or policies. I mean the systems, workflows, decision paths, and governance logic underneath them.

Are we designing around the final record, or around the decision path that produces it?

That distinction matters because not all People work needs the same redesign.

Some work should stay administrative.

Some work needs better visibility.

And some work needs to be treated as a governed decision workflow.

The mistake is treating all three like HRIS transactions.

I’ve started using a very simple test for this:

If the work is mainly about updating a record, keep it administrative.

If the work is mainly about getting the right decision made, by the right people, in the right order, with the right visibility and audit trail, redesign it as a governed workflow.

That’s what I’d call the orchestration test.

I’ve seen more writing lately about HR needing to operate more like a product organisation. I think that’s directionally right. It gets us away from “did we roll this out?” and closer to “does this actually work for the user?”

But for me, that still stops a layer short if we only redesign the surface.

The harder job is designing the operating logic underneath it:

Routing.
Sequencing.
Approvals.
Role-based visibility.
Exception handling.
Auditability.

In other words, infrastructure.

This is not to say workflow is new. Plenty of enterprise platforms already embed more of it natively. The gap shows up when the real process is more context-heavy, more exception-heavy, or spread across more systems than the standard workflow assumes. In fragmented People stacks, that tends to happen quickly.

In practice, I’ve found most People problems fall into three buckets:

Admin.
Visibility.
Governed workflow.

They need different kinds of redesign.

And that distinction matters.

If you treat all three like system-of-record problems, you either over-engineer simple work or under-design the places where judgment, context, and governance actually matter.

One thing that has become very obvious to me recently is that a lot of People friction is not a knowledge problem.

It’s an intake problem.
A workflow problem.
An orchestration problem.

That is why a chatbot on its own so often turns out to be the wrong answer.

A good contrast for me was our monthly headcount reporting.

This was never a governed decision workflow. It didn’t need approval chains, exception paths, or staged decision rights. But it also wasn’t good enough as it was.

The original process was manual, clunky, and hard to use well. Each month, data had to be pulled together, reshaped into something leadership could consume, and turned into a report that was more static than useful. It technically produced an output, but it did not create much real visibility. Leaders could see the headline numbers, but the moment they wanted to interrogate what was driving movement, compare views, or layer in additional context, the process became slow and manual again. 

That was the thing that forced the redesign. 

What I changed first was the reporting backbone: automated generation, a clearer executive snapshot, and a more usable view of the data.

But the more interesting shift came next.

Once that backbone existed, I could expand it beyond basic headcount reporting into broader workforce intelligence — bringing in performance data, parental leave data, and now thinking about engagement as the next layer.

That matters because not every People problem is a workflow problem.

Sometimes the issue is visibility.

The value wasn’t “a dashboard.” It was turning a brittle monthly report into a reliable decision-ready layer that leaders could actually use.

Calibration sat in a different bucket entirely.

For our upcoming performance cycle, I’ve started building an internal Calibration Engine because the hard part of calibration was never where the final rating got stored. The hard part was the decision path before that.

Our performance system could hold the final outcome, but it couldn’t support the weighted average curves, visibility of movement, or the analytics needed to run calibration properly.

So the real process spilled outside it very quickly:

Manual extracts.
Spreadsheet rework.
Divisional and regional cuts.
Separate handling of promotion cases.
Executive roll-up stitched back together later.

That’s why I started building the internal Calibration Engine.

Not to create another place to store ratings, but to bring the actual workflow into one place:

Async triage before the room.
Live calibration.
Promotion guardrails.
Role-based views.
Divisional completion.
Regional roll-up.
Executive review.
Better visibility of movement and patterns across the cycle.

The point is not better storage.

It’s designing the workflow around how those decisions actually get made.

That is a governed decision workflow.

Internal moves pushed me to the same conclusion from a different angle.

In a lot of growing companies, the People stack is not simple. It is fragmented. One tool solves one problem, another solves something else, and over time the real process runs across multiple systems with people acting as the integration layer between them.

The HRIS still holds the final employee and job data.

But the real work sits earlier:

Intake.
Context gathering.
Approvals.
Compensation checks.
Document generation.
Signatures.
Status management.
Handoffs across multiple actors.

And this is where standard workflows often break down, because context matters all the time.

One move may be structural only.
Another may have compensation implications.
Another may need Reward input.
Another may require a signed variation before anything can progress.

That is also why I’ve become much less interested in “fixing the form” and much more interested in designing the workflow properly.

In a lot of these processes, the friction is caused by approvals, handoffs, and document control — not just intake.

That is why I’ve been thinking about internal moves as a new system design problem, not just a process tidy-up.

Capture the request once.
Pull in the right context.
Separate standard paths from exception paths.
Route the right approvals.
Generate the right documents.
Manage the handoffs.
Then write the approved outcome back into the system of record.

Again, the record system matters.

But it is not where the real operating logic lives.

AI is fundamentally reshaping the HR function. But the interesting shift isn't just that AI can do more work. Routine operational work will compress. Some of it already is.

But the interesting shift isn’t just that AI can do more work.

It’s that the economics of building better workflow layers have changed.

AI hasn’t invented workflow. It has lowered the barrier to building targeted workflow layers around fragmented People stacks, faster and more in-house than before.

That matters, especially for companies that are not about to rip out their stack and start again.

A lot of AI in HR still sits at the surface layer:

Summaries.
Drafts.
Chatbots.
Dashboards.

Useful, yes. But that is not the real shift.

A chatbot won’t solve a workflow problem.

The real shift is when AI sits inside a better-designed workflow and supports human judgment without becoming the control layer.

Summarising context before a review.
Flagging outliers.
Drafting rationale.
Surfacing patterns that would otherwise be missed.

That is useful.

Letting AI become approval authority, status truth, or governance logic is where I think people start reaching for it too quickly.

AI should help the workflow think. It should not replace the workflow’s governance.

A big reason my own thinking has moved this way is the environment I work in at Zip. Enterprise access to ChatGPT and Gemini, structured learning, now Codex, and a culture where people constantly share what they’re building — that changes how you think.

It doesn’t just make you faster. It changes what feels buildable.

So the practical takeaway I’d leave People leaders with is simple:

Don’t start with the HRIS. Start with the decision path.

Ask what the process is really doing.

Is it mainly about recording an outcome?
Keep it administrative.

Is it mainly about helping the right people make, review, approve, and defend a decision before that outcome gets recorded?
Design it as a governed workflow.

That’s the shift I think matters most over the next few years.

Not just HR as a service.
Not just HR as a product.

HR as infrastructure — workflow, governance, visibility, and AI-supported judgment designed around how people's decisions actually get made. The companies that get there first won't just have better People teams. They'll have a genuine operating advantage.

If you enjoyed this post or know someone who may find it useful, please share it with them and encourage them to subscribe.

A roundup of the most interesting stuff from the week:

When you’re ready, here’s three ways I can help you:

1. Tools & resources
Resources and tools that give you what you need to build your own startup compensation practices.

2. Comp consulting
Building startup compensation practices that are clear, fair and competitive.

3. Startup People Summit
A 1-day annual event for People professionals in scaling companies. Creating the playbook for startup people practices. Grab recordings from past events, or subscribe to join the next summit.