Go AI Mode: The Playbook for Becoming Invaluable at Work Right Now
Domain expertise + AI is the most valuable combination in any org right now.
Every company I talk to is going through the same thing right now.
Some people on the team are shipping faster than they ever have. Work that used to take 2 weeks are getting done in 2 days. Competitive analyses that required an analyst and a week of research are happening in an afternoon. These people are operating in what I’ve started calling “AI mode”, pairing their domain expertise with AI tools and producing work that would’ve required a small team 12 months ago.
And then there’s the other group.
They’re also using AI. A lot. But the output is... different.
They’re churning out more memos than anyone can keep track of. Each one reads like they’re about to give a TED Talk, ready to change the minds of all 1,000 attendees with a transformative story. Ask how we can improve email delivery rates. Get back a memo that says it’s not about improving email delivery rates — it’s about “Reimagining the fundamental human connection between brand and inbox.”
They’re producing positioning statements as if there were an army of wordsmiths perfectly crafting every word. Sure, you’re selling a backend accounting platform. But you need to position like David Ogilvy. You’re not selling accounting software, you’re “empowering finance leaders to unlock strategic clarity and redefine what’s possible in the modern back office.”
They’re answering every question with Elon Musk’s level of reasoning. Starting every conversation from first principles. Breaking every single query down to its core parts before giving a world-class and succinct answer.
Ask if the monthly numbers are correct, it looks like one of the metrics is down much further than expected, and get told: “Before we look at whether the numbers are correct, we need to ask ourselves: what does ‘correct’ actually mean in the context of a dynamic, multi-variable growth environment? If we strip this back to first principles, a metric is simply a proxy for underlying behaviour. So the real question isn’t whether the number is down, it’s whether our measurement framework is fundamentally aligned with the outcomes we’re optimising for.”
YO. The number is wrong. Just tell me why.
This is the AI slop machine. And it’s the most dangerous employee in every organisation right now.
The 2x2 That Explains Everything
I created this framework because I kept seeing the same pattern everywhere:
Top right → Pay them more. High AI output + deep domain expertise. These people know their craft, and they’re using AI to multiply it. They understand what “good” looks like, so they can direct AI toward good outputs. They use AI as a thought partner, not a copy-and-paste machine. These are the people operating in AI mode.
Top left → AI Slop. High AI output + low domain expertise. This is the danger zone. These people are producing more than ever, but they don’t have the domain knowledge to tell good from bad. So everything comes out sounding impressive but saying nothing. Volume goes up. Value goes down. When you meet these people, their expertise is incredibly far removed from the work they’re submitting (because it’s all AI-created).
Bottom right → Onboard to AI. Low AI output + deep domain expertise. These are your best people who just haven’t made the switch yet. They know the work; they just need the AI toolkit. This is the highest-ROI investment you can make right now. Get these people into AI mode, and the results compound fast.
Bottom left → ✕ Low AI output + low domain expertise. No amount of AI fixes a lack of skill and knowledge. Unfortunately, these people will find themselves in danger of being let go at most companies.
The thing most people miss: the AI slop machine isn’t a bad employee. They’re often enthusiastic, eager, and genuinely trying to be productive. The problem is that AI without domain expertise doesn’t just waste time; it creates more work for everyone else. Someone has to read those 3,000-word memos. Someone has to untangle the first-principles positioning that says nothing. Someone has to explain that the monthly numbers are just... wrong.
Having real domain expertise is a necessity. It’s how you understand where AI adds value to your workflows, how to use AI as a thought partner vs. a copy-and-paste machine, and how to use AI for iterative thinking vs. all thinking.
Most companies will go through a messy period with AI. It will cause some productivity upside for some, and a lot of productivity downside for others. The question is how long you stay in that messy middle.
What AI Mode Actually Looks Like
I’ve been watching people operate in AI mode for a while now. I’m also one of those people. Work looks different.
I started this morning by firing up my Claude Code. I wanted to do some deep analysis of ASPs across our different business segments. Claude Code has access to all of our internal systems. I ask it to go into plan mode. We align on an analysis plan, the hypothesis, the questions to ask, and the charts to build. We convert it to a .md file. I then ask Claude to create a folder structure for the analysis. We create a progress.md file to capture all the analysis as we move along. I create my Claude.md file to ensure the analysis is tailored to how I want.
And then, we work.
Claude starts acting like a team of analysts, pulling back data, creating charts. We go back and forth, iterate on queries, change questions, and note interesting trends. After two hours of working in a deep flow state, I’m amazed at what I’ve done. Working in AI mode feels completely different. It’s hard to explain. It’s incredibly addictive. I don’t recognize my working day anymore. I also work all the time. I’ve never worked so much. It feels exhilarating. I can’t work out if that’s a good or bad thing.
After lunch, I catch up with one of our top sales managers. I’ve been on a tour digging into how our sales reps are using AI. They’re doing a lot of incredible things with HubSpot + AI to supercharge their results. He gives me a tour of his AI workflows. He’s building agentic workflows with Agent.ai and HubSpot to automate more of his work. He’s using Codex to build a team of agents, yes, an entire team, to help with campaign marketing to his book of business. He’s using Claude Code to do analysis in HubSpot and build playbooks for his reps.
He’s a system builder.
Work is different. Sure, there are still fundamental skillsets that are timeless for sales, but work is just different. When you’re in AI mode, how you work starts to feel like a total paradigm shift from what came before.
It’s late evening, and I need to work on a book (I can’t say much more right now). Working on a book is hard; you have to ensure there are consistent threads across many chapters. That every chapter builds upon the others, the narrative is consistently at the right altitude for your audience. Much different from writing a Substack in Grammarly.
So I go AI mode
I built an app with an agent built in that is trained on our audience, goals of the books, criteria needed to make a best seller list, and more. The app stores all chapters, it annotates weak narratives in the book, suggests holes to fill, provides feedback like a reader, and makes tracking edits easy.
It’s different, I feel like a superhero, I work into the late hours revising, rewriting, improving.
AI Mode just feels different.
The Signal From the Top
If you’re reading this and thinking “sure, but is this actually happening at scale?”, the answer is YES!
Shopify made AI usage mandatory. Not suggested. Not encouraged. Mandatory. CEO Tobi Lütke sent a company-wide memo in April 2025 making AI a “fundamental expectation” of every employee. Teams now have to demonstrate why they can’t get something done using AI before asking for more headcount. Lütke described AI as a “thought partner, deep researcher, critic, tutor, or pair programmer.”
This is a company that grew 20-40% year over year while shrinking headcount from 11,600 to 8,100. The people who stayed became dramatically more valuable.
Product designers at Shopify now use AI for all platform feature prototypes; the results are more exploratory and faster to produce. That’s AI mode in action. Making the designer 10x more capable.
Small companies are punching way above their weight. Svenfish, a seafood brand, not a tech company, attributed 82% of its e-commerce revenue in 2025 to AI-powered emails with optimised subject lines. Skincare brand Tata Harper saw a 65% increase in form submissions within 30 days using AI-tested pop-up designs. These aren’t billion-dollar companies with AI teams. They’re small brands using AI to compete like they’re 10x their size.
75% of small and medium businesses are now investing in AI. Growing businesses are nearly twice as likely to do so compared to struggling ones.
The productivity data backs it up. The Stanford AI Index 2025 examined multiple studies and found a consistent pattern: AI shortens processing times and improves quality. Studies show 10-25% performance gains in typical knowledge work like writing, researching, and programming. With the latest models, I feel this is orders of magnitude more.
How to Go AI Mode This Week
Where do you start?
1. Map your team to the quadrant.
Print the 2x2. Be honest about where everyone sits. Who’s in the top right, producing high-quality work faster, with AI? Who’s in the top left, producing more volume, but you’re not sure about the quality? Who’s in the bottom right, great at their job, but hasn’t adopted AI yet?
The bottom-right group is your highest-ROI move. Get them the tools, show them what’s possible, pair them with someone already in AI mode. The return will be much faster.
2. Pick your most painful recurring workflow and run it through AI this week.
Not your most complex workflow. Your most annoying one. Campaign briefs. Competitive analysis. Weekly reporting summaries. Monthly board deck prep. Whatever eats your team’s time and makes everyone groan.
Carve out an afternoon. Ask Claude to plan the workflow with you first, then move to Claude Code and have it automate for you. You’ll be shocked at how far you get.
3. Judge the thinking, not the tool.
The wrong question: “Did they use AI for this?”
The right question: “Did they know what good looks like before they hit enter?”
A 3,000-word positioning doc that sounds beautiful but says nothing is worse than a rough brief with one genuinely sharp insight. AI is the multiplier. Domain expertise is the thing being multiplied. Zero times anything is still zero.
4. Reset your assumptions every 4 weeks.
Here’s the thing about AI right now: the capabilities are moving faster than your mental model of them. Something that didn’t work in Q4 might work today. Something that works today will look primitive by summer.
Build a habit of retesting. Every 2 months, revisit the workflows you ruled out. The teams that treat AI capabilities as a fixed snapshot are already working with yesterday’s tools.
Timeline for AI Mode
Here’s the maths …
Someone in AI mode today is getting better at it every single week. Not because the tools are getting better (though they are), but because they’re building muscle memory. They know which prompts work. They know where AI adds value vs. where it’s a waste of time. They know how to direct it with domain expertise.
That knowledge compounds. And the longer your team sits on the sideline, the wider the gap gets.
6 months from now, AI-native workflows won’t be a differentiator. They’ll be table stakes. The question is whether your team builds that muscle now, while there’s still an advantage to being early , or scrambles to catch up when it’s already the baseline.
The companies in the top-right quadrant aren’t waiting. Neither should you.
Pick a workflow. Open a tool. Go AI mode.
This week.
Until Next Time,
Happy AI’fying
Kieran
Preach
Domain expertise is the thing being multiplied, that line says it all.