People | Tech | Community

FNDN
• Building startup compensation practices that clear, fair and competitive.

FNDN Series
• Insights and interviews showcasing startup people and compensation excellence.

Startup People Summit
• The 1-day virtual summit delivering the playbook on modern practices for building and scaling your startup people practice.

How to Become a HR Consultant
• Sharing every detail of taking a business from zero to $500k in revenue per year.

Matt McFarlane

Thought I'd test out the new Claude HR skill... 

Guess our jobs are safe?

Building startup compensation practices 👉 Compensation Philosophy + Job levels + Salary bands.

After doing comp for a while, here's the 6 things I'd do immediately to get comp under control if I was a new Head of People.
 
I'm assuming you've come into a company where:
- Nothing exists
- You're the first people hire.
- There's <50 employees.
 
Yee-ha.﻿ Let's go.
 
1. Define a basic job level framework.
 
I'm talking basic basic.
 
Even something like: Jnr > Mid > Snr (+Founder = 4 levels).
 
That's all.
 
Just get a basic framework for distinguishing different role sizes. You can expand it later.
 
2. Level your people against the framework
 
- Jim = Junior
- Mary = Mid-level
- Tammy = Senior
 
Bonus points if you want to define the basis for putting them in those buckets (always good to have a rationale).
 
3. Build salary bands based on who you already employ (and pay)
 
Market data is nice but maybe not critical at this stage.
 
Instead: find the median salary at each job + level for the people you already pay
 
For example: junior software engineer.
 
See what the range is between lowest and highest paid.
Decide if it's too wide (20%+) or too narrow (<5%).
Congrats, you've got salary bands.
 
4. Define why someone is at the bottom vs the top
 
Important if you're going to mature these things. 

What dictates that someone is paid higher vs lower in your new band?
 
Don't have performance ratings or similar in place? Use something objective like tenure.
 
Not perfect, but a useful proxy in the short term (especially if you genuinely rate the people you've hired and kept).
 
5. Adjust your bands for hiring
 
Find your bands are holding you back in the market? Lift them.
 
Just remember to bring the current employees up with them.
 
Also make sure you're not accidentally compressing bands too close to one another, so lift the ones above and below, too.
 
6. Write all this stuff down (i.e. comp philosophy)
 
Help your people to understand how this stuff works.
- They'll appreciate it,
- You'll have more trust,
- And now questions can be answered a little more quickly.
 
Congrats, you have a halfway decent comp practice.
 
PS. let me know what is one thing you would add?

The best leader I ever worked with introduced themself with a ‘How to Work With Me’ guide. Here’s what it included — and why it worked so well.

I was new to the team, and so were they — I didn't know what to expect.
 
They didn't hire me — which led to feelings of:
- Would we get on together?
- Would they like me?
- What if they're a terrible boss?

But before we had a chance to connect over a call, something landed in my inbox and the inbox of each of their new direct reports.

It was a 'how to work with me' guide.
 
It introduced themselves and how they liked to work.

This was the first thing that set my impression.

- Not a flashy PowerPoint.
- Not a long-winded speech.
- Just a raw, honest doc.

The doc wasn't lengthy, but it was comprehensive — it covered things like:

• What they look for in their team
• How they like to communicate
• What great work looks like to them
• Their strengths and where they struggle
• How we could help them be better

They didn’t know us yet.

But this gave us immediate insight into who they were — and how to build trust and momentum quickly.

- There was no pressure to respond immediately.
- We read and digested it in our own time.
- And then we responded in kind.

That moment set the tone for our entire team culture.

We started sharing our own guides.
It encouraged us to do the same with our teams.
It scaled trust without needing synchronous time.

We talk a lot about onboarding processes and leadership styles.

Start here.

It’s simple, human, and unforgettable.

Have you ever received or written a ‘How to Work With Me’ guide?

I’d love to see your take 👇

If you're a People professional, your goal isn't to be liked, it's to be respected. Here's ten things I've seen teams do wrong, and the ten who get it right.
 
The true indicator of success in a People function isn't personal popularity, but creating a work environment that empowers its people.
 
Those that work work in People teams because they "like people":
 
1. Overpromise but under-delivers.
2. Give in to every demand.
3. Shy away from having difficult conversations.
4. Play favourites among people.
5. Avoids making tough decisions.
6. Avoids conflict at all costs.
7. Apply policies inconsistently, fearing backlash.
8. Side with popular opinion, not what's right.
9. Overlook performance issues.
10. Focused excessively on office parties and perks.
 
People professionals that earn respect:
 
1. Set clear expectations.
2. Take responsibility for mistakes.
3. Communicate effectively.
4. Balance employee needs with business goals.
5. Enforce policies fairly and consistently.
6. Handle conflicts professionally.
7. Supports addressing performance issues promptly.
8. Provides valuable, actionable feedback.
9. Develop effective training and development programs.
10. Uphold ethical standards at all costs.
 
The real currency in this profession isn't being liked, it's fostering an environment that enables employees to thrive.

What is one you would add to either of these?

Yesterday, ~900 people learned what to charge their clients as a HR consultant.

It's been the most uncomfortable part of my journey from in-house to a career that now earns me twice what my highest paid role ever did.

In this edition, I break down not only how to land on your rate, but the different pricing models I've used it for.

All in painstaking detail so you don't have to guess through it like I did.

And you can read it for free here: https://lnkd.in/gCrNXM-n

The god father of AI built a site that scores AI impact on jobs and, brb I'm going to re-skill as a Gardner.

Andrej Karpathy is one of the most influential AI engineers of the modern era — you can read more about him here: 

https://karpathy.ai/

He vibe coded a site that pulls US Bureau of Labor Statistics info and ranks their exposure to AI capability.

Check it out here:

https://lnkd.in/gYP-CXXy

HR Managers have an AI exposure of 7/10.

What's cool though is the brief description it gives on *what* will be disrupted.

And honestly, it's the parts I'm ok with being disrupted.

Check it out and let me know what you think and if you agree with its scoring.

I've rarely seen a careers page that actually made me want to apply.
 
- Stock photos
- Five values that could belong to any SaaS
- A culture paragraph that says everything and nothing
 
Meanwhile, the companies with real pull?

You hear about them because an employee jumps on LinkedIn and writes, "we just shipped this and I'm pumped"
 
If you're running the People function at a fast-growing SaaS and can't match Big Tech on salary, you know exactly why that matters.
 
The companies with pull have built something candidates already believe in before they've even applied.
 
In the coming edition of FNDN Series, I sat down with Lachlan Bradford, a brand and community builder in SaaS, and it reframed how I think about this for employees.
 
I'm breaking down the playbook in this week's newsletter.

Subscribe here (it's free): https://lnkd.in/gUTnpE_s

A lot of meetings could have been an email. Here's one of the best ways to filter the fluff.

One of the most powerful things I learned (way too late in my career) was how to become hyper efficient through meeting diligence. And there's one framework at the cornerstone of this.
 
It wasn't until I began working for truly remote companies that it dawned on me how much of an unlock it is to cull them wherever possible.
 
Historically, meetings were booked for anything and everything.
- Progress updates on projects/work.
- Regularly scheduled because someone decided it would be good to check-in.
- The ad hoc 'can I ask you something' that turns into a full blown half hour to an hour of your day - gone.
 
The fact is that these kinds of meetings kill the deep work necessary to deliver critical outcomes for you and your business. They distract you from true productivity, and harm your career in the process.
 
But then I was introduced to the POST framework — and we made it a pre-condition necessary for any meeting to go ahead had. 

The person who wanted the meeting had to include this acronym in the meeting request.
 
Here's what it stands for:

- Purpose: why are we having the meeting?

- Outcome: what goal should we accomplish by having this meeting?

- Structure: what are we doing in the meeting? Are we making decisions, or are we brainstorming something? what is everyone's role?

- Timing: how long is it going for and how does the agenda fit within this duration?
 
When you have to do even the most basic amount of work to even have a meeting you'd be surprised how many of the ad-hoc meetings fall away, and even the cyclical one's (each meeting needs to be justified).
 
Take this a step further.
 
If the meeting is a progress update, or is otherwise sharing information one-directionally, make it asynchronous and do it via written or video recorded update.
 
You'd be surprised how much faster things get done when you can read it and respond in your own time (with deeper thought or while listening at 2x speed).
 
It goes without saying that something like this needs to be adopted from the top down, but by piloting an approach and measuring the outcomes (reduced meeting time, improved output) you'd be surprised how easily the case stacks up to adopt this approach across a company.
 
Have you ever tried an approach like this at your company?

If Jeff Bezos says no team should need more than two pizzas, and AI means I’m now doing the work of a whole team — do I get both pizzas?

I speak to 100's of HR team's a year. Here's 5 steps the best ones are doing to embed AI.
 
Too many HR teams are either intimidated or overwhelmed when it comes to AI, and they're falling behind.
 
Instead, embrace the below steps and you'll be in the top 10% of People teams using AI.
 
1. Create dedicated, protected time
 
You can't just slap AI on top of your day job, or squeeze it in between meetings (duh).
 
The work you're doing today will always trump the deep focus time you need to be successful.
 
So carve it out, choose 1 of these:
- Commit 1-2 hours a day to AI, or
- Give 1 day a week to AI.
 
If you need to drop something to make this happen, do it. This must be higher priority than at least one thing you're team does or it won't get the attention it needs.
 
Protect the time, remove distractions — go into deep focus and spend it doing one of the things in the next step.
 
2. Time box it
 
If you're just starting out, commit to doing a short term blitz and then work out how to embed it longer term.
 
Here's a format you can steal:
 
6x week sprint:
 
Weeks 1-2: Focus only on learning/immersing yourself in what's possible

Weeks 3-4: Focus on testing the tools you saw in weeks 1-2

Weeks 5-6: Move from testing to actually building (what to build is covered in step 5)
 
3. Curate learning, don't create
 
There's so much free content available on using AI that you'd be crazy to spend money on it.
 
- Greg Isenberg
- Ruben Hassid
- Jeff Su 

These are just three (from dozens of) creators I love, who share genuinely actionable insights on embedding AI.
 
Spend time watching and seeing what can be done - this will get your brain subconsciously thinking about AI solutions for your team.
 
4. Empower your team
 
The key message here is: 'remove friction'
 
Time is a friction (tackled in step 1), but the second biggest one I see, is money.
 
Don't let budget be a blocker here.
 
If you're CEO wants you to embed AI practices (and you should to quit if they don't) — then securing a small budget is a no brainer.
 
There's so many tools - give your people a chance to trial a bunch of them and see what works.
 
Then:
- Keep the ones that work
- Remove the ones that don't
 
Don't make a big procurement process block game changing improvements.
 
5. Start small when building
 
Don't try to boil the ocean.
 
You won't be able to replace a HRBP in a week.
 
Instead, start small.
 
Pick the smallest workflow/process/interface under your remit, and start there.
 
Use your success here to build momentum and move up the value chain.
 
Don't let overwhelm and intimidation stop you from becoming a more effective People team.
 
Tell me, which step do *you* think is most important to embedding AI in People teams?

Even though my team smashed his team in the Great Debate on Pay for Performance, I actually agree with a lot of what James A Seechurn has to say.
 
So much so that after reading his book, I reached out to have him on the FNDN Series.
 
The book is 'What Pay Costs'.
 
It's incredibly good. The kind of book that makes you annoyed you didn't write it first.
 
I thought I grilled him for over an hour, poking holes in his idealistic views on how pay can work.
 
But actually, I just nodded along.
 
If you think most of the pay practices you see at work feel a bit broken — whether as a people professional, as an employee, or as a manager — this one's worth your time.
 
Subscribe to the FNDN Series to get a digest of the conversation, and a link to the full episode where we break it all down for you.
 
Join 5,500+ startup Heads of People here (it's free!):
https://lnkd.in/gUTnpE_s

Over the past 2 years I've been gradually refining my newsletter, written specifically for startup Heads of People. Lately, the investment has been paying off.

If you're the Head of People / CPO in a scaling company (50 to 500 people) 
👉 This is for you (it's free): https://lnkd.in/gUTnpE_s

Paying employees more money won’t motivate them.
 
But HOW you pay them will.
 
Here’s why
+ 4 steps to motivate your employees (without paying them more):
 
Employees don’t disengage because they want higher salaries.
 
They disengage because they see unfair pay practices.
 
When two people do the same job but one earns significantly more:
 
- Creates a lack of clarity on how pay decisions are made
- Makes pay feel random, not earned
 
But here’s where something interesting happens:
 
Double Demotivation Theory.
 
Double Demotivation Theory says there are two ways pay can make employees disengage:
 
1. People who believe they’re earning less than they should lose motivation.
 
That's the obvious one, but here's where it gets weird...
 
2. People who are being paid MORE compared to their peers ALSO lose motivation.
 
Now, this proves ONE thing:
 
It’s not just about more money. It’s about TRUST.
 
If people don’t trust the system, no amount of raises or perks will keep them engaged.
 
Fixing motivation doesn’t start with throwing more money at people.
 
It starts with paying fairly.
 
Here’s 4 steps to start paying fairly + keep your people motivated:
 
1. Have a clear, structured approach to pay
2. Make sure employees understand how salaries are set and why
3. Be transparent. Even if you don’t share every number, explain how decisions are made
4. Benchmark pay regularly. Ensure internal pay aligns with market rates — and that pay gaps aren’t eroding trust
 
TL;DR
 
Motivation doesn’t just come from bigger pay checks.
 
It comes from people knowing they’re being treated fairly.
 
People aren’t just demotivated by being underpaid.
 
They’re demotivated when they see OTHERS getting UNDERPAID. 
 
Pay fairly -> build trust -> drive motivation.
 
What’s one thing companies you'd like to see companies do today to make their pay practices feel more fair?

This is the most powerful tip I learned for growing a personal brand on linkedin:

You can just say stuff.

Block cuts 40% of workforce because AI - Stock pops 20%. Atlassian cuts 10% of workforce because AI - Stock drops 7%. 

Does the market think they should have gone deeper?

Or something else?

I help scaling companies actually use AI to get work done, not just talk about it.

I’m Matt, Founder of PxAI, a modern People Ops and AI enablement consultancy working with startups, scaleups, and PE-backed teams (UK, Europe, US).

Most companies aren’t blocked by tools. They’re blocked by:
🤯 messy workflows
😬 inconsistent manager habits
🧩 unclear ownership
🗂️ scattered policies and knowledge
📉 “we tried AI once, it didn’t stick”

That’s where I come in. I help CPOs, CHROs, COOs, and VC/PE platform teams build simple, stage-appropriate systems that scale.

What I work on:
⚙️ AI maturity + workflow audits (heatmaps, bottlenecks, ROI, 30/60/90 plans)
🧠 AI enablement programmes (behaviour change, training, champions, comms, governance)
🤖 Automation and agentic workflows (n8n, Make, Zapier, OpenAI), designed to run in production
📌 People foundations that stop “People Debt” building up (levelling, performance, onboarding, hiring ops, comp, policies)

My background is 16+ years across recruitment and tech, building and implementing Talent and People Ops strategies inside fast-moving companies.

If you’re trying to scale without adding unnecessary headcount, and you want AI to improve throughput, quality, and consistency, message me.

Matt Bradburn

Anthropic published the most important paper for people and org leaders this week, on AI's impact on the labour market 📊

Actual usage data from their platform, matched against employment outcomes from the US Bureau of Labor Statistics.

6 things that stood out to me 👇

🔍 The gap between what AI can do and what it's actually doing is enormous

In Computer & Maths roles, AI could theoretically handle 94% of tasks. It currently covers 33%.

That gap is the window organisations have to get ahead of this. Redesign roles. Build workflows. Upskill teams. Because that coverage is growing, and the organisations doing the work now won't be the ones scrambling later.

📉 AI isn't causing redundancies yet. It's quietly slowing junior hiring.

No measurable increase in unemployment for AI-exposed workers. That's the headline most people will take away.

But further into the paper, something more interesting: hiring of 22-25 year olds into exposed occupations fell 14% compared to 2022.

Companies aren't making people redundant. They're just not backfilling. Not posting the same junior roles.

💰 The most exposed workers aren't your lowest paid. They're your highest paid.

👉 Most AI-exposed workers earn 47% more than the least exposed. 
👉 They're almost 4x more likely to hold a graduate degree.
👉 They're more likely to be older and female. URGH.

Your biggest payroll lines are in the firing line. That's a board-level workforce risk conversation.

💻 The three most exposed roles right now:
👉 Computer programmers: 75% task coverage 
👉 Customer service representatives: significant and rising 
👉 Data entry: 67% task coverage

Most scaling organisations have meaningful headcount in at least one of these. If your 2026 workforce plan is built on 2023 assumptions, it needs revisiting.

🧩 The research validates the task-level approach: 

This is what I work on with every client, nice to feel validated! Anthropic's researchers built their entire framework by breaking occupations into individual tasks and measuring which AI is actually performing in professional settings.

Not "will AI replace this job?"
But "which specific tasks in this role are already being done by AI, and what does that mean for how we design the work?"

✋ 30% of workers have zero AI exposure

Cooks. Mechanics. Bartenders. Lifeguards. Their tasks don't appear in the usage data at all.

If you've got operational or frontline teams alongside knowledge workers, your AI strategy needs to reflect that difference. 

Anyone selling a blanket "AI transformation" without that nuance isn't being serious.

Fundamentally?

There's no mass displacement yet. But the early signals are there for anyone willing to look at them.

The window to redesign roles, rethink hiring assumptions, and build genuine AI capability into your teams is still open. I wouldn't assume it stays that way.

What are you seeing in your own organisations? 👇

#peopleops #ai #futureofwork #peopleleaders

I can’t remember the last time I reposted someone else’s content on here, but this was too accurate to not…

The atomic layer of work is your clicks and your processes.

Start with those before you move molecular and beyond…

Want to know why some companies crush it with AI while others waste thousands?


It comes down to where they start.

Every company I've worked with that got real ROI from AI had one thing in common.

They started with their processes.

After working with dozens of businesses on AI implementations, I've found there's a clear stack that separates the ones who get results from the ones who waste budget.

We call it the AI Implementation Stack.

From bottom to top:

1⃣ Process Foundation 
→ Documented workflows 
→ Clear ownership at every step 
→ Optimized before any tech touches it

2⃣ Data Layer 
→ Single source of truth 
→ Clean, connected, accessible 
→ No more copy-pasting between 6 different tools

3⃣ Integration Layer 
→ Systems that actually talk to each other 
→ Automated data flow between platforms 
→ Zero manual transfers

4⃣ Automation Layer 
→ Repetitive tasks handled by workflows 
→ Triggers, rules, and logic doing the heavy lifting 
→ Your team focuses on decisions, not data entry

5⃣ AI Layer 
→ Intelligent agents solving complex problems 
→ AI that actually has context because the layers below feed it clean data 
→ This is where the magic happens

The companies trying to start at Layer 5 without the foundation? That's where budgets get burned.

The ones building from the bottom up? They're the ones turning AI into a real competitive advantage.

Every layer compounds on the one below it. Get the foundation right and the AI practically builds itself.

Make sure to save this post for your next Ai project.

Follow me Luke Pierce for more content like this.

😘 AI doesnt know anything about you. It knows a lot about the world, BUT: 

👉It doesn't understand your role. 
👉Your business. 
👉Your communication style. 
👉How your manager likes to receive information. 👉The relationship dynamics between you and your stakeholders. 
👉The way decisions actually get made in your company. 
👉How comms flow. 
👉What "good" looks like in your specific context.

All it has is aggregated, externalized knowledge. It generates text based on patterns. That's it. Sorry to break that to you… 

🔥 So when you type "write me an onboarding email" and get something generic back - that's not a failure of AI. That's exactly what it's designed to do without context. You gave it nothing to work with.

🙃 Also, to all those morons giving you ‘300 best hr prompts’ well, you can ignore them. Completely. 

Here's what to do instead. Build a skill file. Then build more. But build them for you. 

A skill file is a plain text or .md document that gives the AI YOUR context - your tone, your process, your standards, your company's way of doing things. What good looks like. What bad looks like. You write it once. You load it every time. Output comes out consistent, on-brand, and actually usable.

📄 A prompt is something you throw away. A skill file is something you keep, refine, and give to your whole team.

Here's why this matters if you run People Ops at a scaling company:

🔥 Your context carries forward. No more re-explaining your company every conversation.

🔥 Five people use the same skill file, they get consistent output. Not identical - consistent.

🔥 You improve it once, every future use gets better. Prompts don't compound. Skills do.

"How do I build one?"

Don't stare at a blank page. Get the AI to interview you. Tell it the task, ask for 8-10 questions about how you actually do the work. Answer honestly. It structures a skill file from your answers.

Pick one task your team does every week. Build a skill file for it. Test it. Fix what's off. Share it with your team. Build the next one.

One skill file. 20 minutes. Saves hours every week. Gets better every time.

And please - ignore anyone who says "download my skill files." Their context is not your context. That's literally the entire point. Build your own. 

Want to know how? Well just hit the link in comments because I’m done with the bs… 😘 

I’ve laid it allllll out for you.

No alternative text description for this image

👉 You open ChatGPT. Or Claude. Or Copilot. You type something like:


"Write me an onboarding email for a new starter."

You get something back. It's... fine. Generic, but fine.

Tomorrow you need another one. You type a slightly different version of the same request. You get a slightly different output. Your colleague does the same thing and gets something completely different again.

🥲

This is how most teams use AI. One-off prompts, typed fresh every time, producing inconsistent results that nobody can build on.

😉 There is a better way. Skills files. 

A simple markdown doc, which when combined with many other markdown docs becomes incredibly powerful... 

This is what I showed folks on my course today... They loved it - next cohort in March btw - (https://lnkd.in/efifaQQ7)


- - - - - - - -

A skill is just a document. A text file (usually Markdown, ending in .md) that tells the AI exactly how to do a specific task.

Not a vague prompt. A proper set of instructions. Context about your company. Steps to follow. Constraints. Examples of what good looks like.

Here is the difference:

👉 A prompt
"Write me a job description for a Senior Product Manager."

You type it. You get something. You move on. Tomorrow, you start from zero.

❤️ A skill

A document that contains:
👉 Your company context and tone
👉 Your job architecture and levels
👉 The structure every JD should follow
👉 What to include and what to avoid
👉 Examples of JDs that worked well
👉 You load it once. Every JD comes out consistent, on-brand, and usable.

That is the big shift. 

From typing instructions to loading them. 

From one-off requests to reusable assets for your team.

❤️ A prompt is something you throw away. A skill is something you keep, improve, and share with your whole team - glorious! 

- - - - - - - - 

❤️ Why This Actually Matters ❤️

Four reasons.

1. Memory. A prompt starts from zero every time. A skill carries your context forward. Your company values, your tone, your processes, your standards. All of it. Every time.

2. Consistency. When five people on your team use the same skill, they get consistent output. Not identical, but consistent. That matters when you're scaling.

3. Compounding. You can improve a skill. You notice the output misses something, so you update the skill document. Now every future use benefits from that fix. Prompts don't compound. Skills do. 

4. Scaling. You start by taking a workflow, then you break down the tasks, then you build the skills for those tasks, then an orchestrator skill, then you get Claude cowork or claude code to run them... 

It's a beautiful thing to watch! 

Want to learn more? Come sign up for my course here - https://lnkd.in/efifaQQ7 

🔗 Or Drop the word "skills" for the full guide

😰 Scared of Claude code? Scared of the terminal? I was too… until I got into it and realised what you can create.

❤️ This post isn’t for everyone, but if you want a powerful stack, I have set out the what, the why and the how in the article attached.

Background: 

I've spent the last few months building People Ops automations with a stack that I think is worth understanding.

💪 The setup

Three things working together:

1️⃣ Claude Code. Runs in your terminal. You tell it what to build in plain English, it writes the code and deploys the workflow.

2️⃣ MCP connections. These let Claude Code talk directly to your systems. Notion, Slack, Google Workspace, and if you’re lucky, your HRIS. Read and write access, not just chat.

3️⃣ n8n. Open source workflow engine. Self-hosted, so your employee data stays on your infrastructure. Not routed through a third party.

🔥 How it works day to day

Say you need part of an onboarding flow, the comms and first week parts. You tell Claude Code: when a new hire lands in hibob, create their checklist in Notion, send a welcome in Slack, DM the manager, book the Day 1 calendar events, and alert me if anything fails.

It checks your existing Notion schemas, builds the workflow, deploys it to n8n, tests the endpoints. You open n8n, walk through what it built, test with dummy data, tweak as needed.

🔐 The governance bit (don't skip this)

You're giving an AI access to build workflows that touch employee data, send messages on behalf of your company, and modify production systems. If that doesn't make you pause, it should.

💪 The approach that works:

→ Nothing goes live without a human reviewing every node in the workflow

→ Data sensitivity rules get encoded into the build instructions. Comp data never appears in plain text in Slack channels. Manager briefs go as DMs only. The AI follows what you tell it, so you need to tell it clearly.

→ Policy bots must know what they don't know. If the answer isn't in the handbook, the workflow escalates to the People team. No fabrication. No guessing.

→ The attrition early warning system is framed as proactive support, not surveillance. Language matters. Your leadership team needs to understand the intent, and you should consider whether employees know it exists.

→ Access is scoped. Claude Code gets read/write to specific databases, not your entire workspace. HRIS data comes via webhook triggers, not direct access.

→ Everything gets tested with dummy data before connecting to real triggers. Activate in stages, not all at once.

This is powerful. Which is exactly why the governance has to be as solid as the build.

👏 What you actually need to get started

I've written the full walkthrough with setup instructions, all five playbooks, the access control matrix, the review protocol, and the advanced patterns for teams who want to go further.

🔗 Link below in comments.

#peopleops #claudecode #aiforhr

taplio

The Battle

Matt McFarlane

Matt Bradburn

Followers

Posts

Engagements

Likes

Performance ScoreMatt McFarlane's LinkedIn Performance

Performance ScoreMatt Bradburn's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Matt McFarlane

Matt Bradburn

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Matt McFarlane's LinkedIn Performance

Performance ScoreMatt Bradburn's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreMatt McFarlane's LinkedIn Performance