FORWARDED THIS? Subscribe here to get the next one.
TL;DR
The map of how comp teams actually use Claude, the top tier in action with Ivan Nosov on pay equity, and the fastest AI moves worth making this week. Also, I moved to Madrid.
A NOTE FROM GIAC
 |  |  |
Wheels up over England, down over Madrid, and the two of us with far too many bags.
Amigos,
I'm writing this from Madrid. This weekend my wife and I packed up and made the move. Somewhere over the Channel the grey English drizzle turned into hard Spanish sun, which felt about right.
It's the start of a new chapter, and I'll keep building Range in the open from here, with all of you.
It's been a week on the work front too. The map I shared on how comp teams are actually using Claude took off. Over 35,000 impressions in a few days, my most read post in months. So this issue pulls that thread properly, and Ivan Nosov shows what the top of the map looks like in real life.
☕️ Think of this as the quick catch up I'd give you over a coffee.
One thing before you scroll. If you’re also going through a period of change this Summer, whether you’re moving house, relocating, changing jobs or changing how you work with AI, hit reply and tell me what’s on your mind. I read every reply.
Giac
RECENT EDITIONS
New here, or catching up? Two good places to start.
THE BREAKDOWN
How comp teams actually use Claude
I spoke to over 100 reward practitioners and mapped what they genuinely do with Claude. Three tiers came out of it.
Chat still does the heavy lifting. Most people use it to draft comms, answer the same salary questions, and put equity into plain words. It's boring, high volume work, and it's the safest place to start.
The real shift is Claude Code and Cowork. This is where people stop chatting and start building. They parse job descriptions against survey data, keep their bands current, and run their own pay equity checks. Fewer people work this way, and the ones who do rarely have a peer inside their company.
The edge is agents and delegated work. It's rare, powerful, and where most of the fear still sits. The practitioners seeing real gains all have one thing in common. They started with the boring, high volume work and only moved up once that was solid.
Quality comes first. The time savings follow. You still can't fire and forget a pay decision.
Know a comp person stuck on this? Forward it over, that's how the room grows.
RANGE PODCAST
 | Ivan Nosov Head of People AI & HR Technology, Campari |
Pay equity you can defend to an auditor
The map shows the top tier. Ivan lives on it.
He runs pay equity analysis without being a data scientist. His method is the one every reward team should steal. He uses the model to write the code that does the maths, rather than asking it to do the maths itself. The output is a Python script he can read, run on his own laptop, and hand to an auditor.
And that's where most of us get stuck. Could you defend your use of Claude if someone audited you tomorrow? Ivan's answer is the calmest five minutes you'll spend on AI governance this week.
Let the model write the code. Let the code do the maths.
Where to find Ivan: LinkedIn →
FROM THE COMMUNITY
The question that keeps circling the room
Is it actually safe to put employee data into Claude? It comes up in almost every conversation.
Ivan drew a line the room liked. If the information is generic enough that you'd happily post it online tomorrow, you have plenty of room to experiment. The moment it's identifiable pay data with names attached, you stop. Most of our work, the salary ranges, the methodology, the comms, sits well on the safe side of that line. Far more than people assume.
Where do you draw yours? Hit reply. I'm collecting the best answers for a future issue.
STEAL THIS
A good comp prompt does three jobs. It sets the role and context, points to the data or benchmark, and names the exact decision you need out. Get those right and the quality jumps. The full plug and play prompts live in the Starter Kit and inside the room.
ON THE RADAR
[Anthropic] The top model for heavy analysis leaves subscriptions on Monday. Fable 5, the most capable model for heavy analysis and long documents, stays included in Claude Pro and Max only through 7 July. Front load your biggest comp and survey work this weekend, before it moves to paid credits. Read →
[EU AI Act] AI pay tools are now officially high risk, and you just got 16 more months. Brussels confirmed that AI used to set pay, score performance or support promotions counts as high risk under the AI Act, then pushed the compliance deadline from August 2026 to December 2027. Use the runway to map where AI already touches your pay decisions, rather than scramble later. Read →
[New research] Old pay data just scales your AI mistakes. Feeding backward survey data into a comp model doesn't fix your decisions, it produces flawed pay assumptions faster. A 2026 survey of comp professionals puts the top AI risks at too much trust in the tool (53%) and unaudited bias (44%). Fix the data foundation before you automate. Read →
Reading this because a friend forwarded it? Get Range Weekly in your own inbox every Sunday. Subscribe free |
 | That's all from me this week, my first one written from Madrid. This is technically where the newsletter ends, but we don't have to end here. I'd love this to be a two way thing, so hit reply and tell me what landed, what you're building, or where you're stuck with AI. I read every one. Until next week, Giac |
 |
When you're ready, here are three ways Range can help you.
1. Join the community
Find your people. Inside Range, reward pros learn to build with AI together, the prompts, the setups, the band workflows that actually hold up. Free is the signal, the room is the substance. Apply to join →
2. Grab the free resources
The Comp AI Starter Kit gets you running with Claude for compensation in five steps, no code needed. Get the kit →
3. Reward Rewired 2026
Our in person gathering for the people actually building this in comp. Register your interest →