Parry Malm

I had a random sales rep DM me about Megadeth and I knew instantly it was a bot. We'd never spoken. He'd never interacted with anything I'd posted. But somewhere in my profile, the words "Megadeth" and "ice hockey" were sitting there, waiting to be weaponised by a sales engagement platform. This is where we're at now. Lazy mass-market outreach, dressed up as personalisation. A thousand messages fired into the internet, each one pretending to have done its homework. Buyers can smell it a mile away. And they hate it. Salesforce reckons three quarters of sellers missed quota last year. I can't prove this is the reason, but spraying fake Megadeth chat at strangers probably isn't helping. The tools make it incredibly easy to be lazy. Find a detail. Insert a detail. Add an exclamation mark. Hit send. Digital mosquitoes, buzzing around people's ears with absolutely nothing useful to say. I saw this constantly at Phrasee, and I see it now at Drumbeat. If the only reason you know I like Megadeth is that an LLM told you so... you can put an egg in your shoe... and beat it.

You know who writes AI slop? YOU DO. And you're not even using AI! Last week LinkedIn released their AI slop button, and the internet did what the internet does. The death of AI content! The glorious return of Real Human Writing©! But... that's not quite what's happening. In fact... it's not even close (here's why... with the stats below) Many assumed LinkedIn built an AI detector. They didn't, because detectors don't work. They measure statistical patterns in text, not whether or not a post is boring. So LinkedIn skipped that entirely and have focused on labelling. Every tap is one person saying "this is slop." Collect millions and you have a training set for the only variable that ever mattered: Perceived Quality© They're not building something that detects AI, and they even say so themselves. No, they're building something that detects boring. Which is a great outcome, and should only worry the beige amongst us. But look, it's not just me saying this. The data backs it up. Statistically, most posts are average (haha I mean, that's what average means). Most of everyone's posts are mid. But every once in a while, you get an outlier. The one that goes bonkers. That's what you gotta be targeting. So the question becomes: does using AI help or harm the chances of a post going Bonkers©? Here are the stats (based on our analysis of about 20k posts): - Untouched AI drafts broke out 13.4% of the time - AI-drafted, lightly human-edited drafts (a phrase changed here and there) hit 22.2% - Human-written? 4.9%. The lesson? AI alone? bad. Human alone? Worse. Human + AI? Best. AI can generate slop at scale. But so can humans. So this new AI Slop button? It's a lot smarter than people think. And if you have some good ideas and aren't boring... then you've got nothing to worry about.

The golden age of AI-generated LinkedIn garbage is officially OVER. Kind of. Most people think AI content is bad, and that we should use an AI-detector to remove it. But that's a very silly way to look at things, not least becasue AI detectors are pretty bad. What LinkedIn is doing here is actually pretty awesome. They're creating a huge human-labelling system for "AI Slop". this is quite nuanced and smart. AI Slop is not AI-generated content; it's lazy, bad ideas, presented in a bad, sloppy way. Sometime's it's Human Slop. This will result in feeds improving for those who have 1) good ideas; and 2) aren't lazily relying upon ChattyG to spam out a post about what they've learned about B2B sales. For Drumbeat obviously this is great news. The work we've done with our Anti-Slop technology ensures that good ideas are at the forefront of the content, and the beige AI tropes are not. And for YOU, the LinkedIn user looking at your feed - this is even better. It should, over time, get rid of bad content. AI isn't the enemy, even though it's what people love to hate. Slop is the enemy, be it of the AI or human variety... and what LinkedIn are doing is a brave, bold step to combat this. Of course, maybe it won't work. But you gotta appreciate the effort...!!!

For 46 years, my eyeballs were fundamentally broken. Like, properly useless without thick glasses or contact lenses. I always assumed laser eye surgery was still stuck in the 1990s. Like if you got it done, there was a 117% chance you'd go blind and probably need a glass eye. So there I was, doomed to a lifetime of smudged lenses, vanishing contacts, and poking myself in the eye at 6 AM just to doom scroll stupid AI reels. Just figured I had to suck it up. Nothing to be done about it. Blind as a bat (I was -7 in each eye, legit blind as a bat!!) Then, just by chance, my old boss Nik and I reconnected for coffee one day. Despite proving unsuccessful at firing me back in the day, he powered through life, and had become CEO of OCL Vision. And he changed my mind. Well almost. I first built a payback model for the lifetime cost of bad eyesight. New glasses every 5 years, contact lens and accoutrement costs, plus a time-cost of spending 5 mins a day faffing with them (assessing my time to be worth a measly £10/hr) Then I compared it to the upfront cost of getting my eyes sorted by OCL Vision. The math was a complete no-brainer. The ROI was HUGE. It paid back after about 5.5 years... and then that was that. Massive thanks to Dr Ali and the OCL Vision team. No drama, no BS, and no 90s sci-fi horrors. When I wake up in the morning now, I open my eyes, and I can see the London skyline, without having to fumble around for my glasses. It truly is magical. If anyone is sick of glasses and contacts, but is nervous about laser eye stuff - take it from me. It is probably the single best investment I've made in my life. Give OCL a call!

I retired at 44... and un-retired at 46. Here's what early retirement is actually like (tldr: it's kinda awesome but also kinda sucks) After I sold Phrasee, I stuck around for a year and a half, and then packed it in. I was zonked, exhausted, burned the eff out. So off into the sunset I rode. That was the end of August, 2024. Freedom!!! All day to do what I want, when I want, with no one to answer to. This is what all the years of stress were building towards... and I had achieved everything I had set out to. A house on a hill, FU money, and the time to enjoy it. Which, at the time, I thought meant I'd just chill for 50 years. Anyways, I woke up on September 4th... and took my kids to school. Then I got home. And... Tumbleweed. My wife was around of course and we went for lunch, and walks, and we even tried a tennis lesson (I was so good that i nailed a ball not only outside of the cage, but over the next cage. Turns out a "tennis swing" is not the same as a "baseball swing", who knew?) And you know what? It was dope. Like, legit it was dope. I chilled out, spent lots of time in the forest, went on lots of walks. But after a few months... I started to get cabin fever. Want to go for lunch today? Well, we just had lunch yesterday so... And before long... I discovered something: having lots of time to do what you want is legit great. But having something to do in that time is even greater. And people to do it with... the problem is, when you're retired at 44... there's not many playmates around, so you spend a lot of time on your own. Which I guess is OK, but it's also pretty boring after a while. So after about a year and a half, I decided I wanted to get back into the game. Not because I'm deeply passionate about B2B sales (I hate to burst your bubble) but because I get stimulation from solving interesting and hard problems. And when I don't have an interesting and hard problem to focus on, I invent problems, and that's when I get myself into trouble. Fast forward to now, and I'm no longer retired (or unemployed, take your pick). And it's cool. I don't always love it - and in fact, I'm always deeply skeptical of people who say they love work - because... have you tried NOT working? It's WAY better... so long as you have interesting problems to solve. For me, I find the best problems in an entrepreneurial setting. But I'm a masochist. So for all you who want to be FIRE, or better yet FATFIRE (which is preferable IMO), do it for a while, absolutely. Take a break, touch some grass, chill the eff out. But then get back in the game. Maybe it's not work, maybe it's something else. But idle hands and all that... The REAL G's know what you gotta optimise for: autonomy, stimulation and people. Not just money so you can "relax" for the next 50 years 😴

Agentic! Agentic! Agentic! FFS maybe i'm thick, but I don't totally understand the hype. First of all - what is Agentic? Well, in simple terms - it's when the machine determines the workflow, rather than having a predefined one. Which sounds cool except... What if there's a predefined workflow that's good? Then isn't it rather wasteful to have a magical Agentically Agentic Agent recreate a workflow each time? Workflows are great. Because you know the inputs and the outputs. And they're easy to monitor and control and measure and tweak and optimise. Where workflows fall short is when one node in it borks, and then it all falls over. So THIS, to me, is where "Magical Godlike AI Agents" should be used. When there's ambiguity in a workflow ("is this person a good target account or not, and if good, how good?") use an agent (or other forms of AI/stats) to then direct to the correct workflow direction. And even have the agent help decide what steps in a workflow to take, within guardrails. But having an agent determine the workflow every time? Wasteful, inefficient, and also environmentally damaging and capital inefficient. (it's also slow...) (Yes, I know that you can have plans, templates, memory, tool constraints, policies, and whatever, thank you Agentic-Bro's for pointing that out. But I'm using a generalisation as a rhetorical tool; Claude will confirm this for you) Look, I know how people love a buzzword, and now everyone wants some crazy agentic whatever to kiss you on the forehead at bedtime and tell you everything is going to be ok. But an agent will not replace domain expertise, specific and repeatable workflows, and also that old-fashioned, boring human jugdement. Hey, though I should caveat: what do I know?! Maybe I'm too thick to understand what agentic really is, and my abject stupidity will be proven out in the coming months and years. But also: maybe not...

Once I was in a meeting with a dude bragging about increasing ARR per head like he’d just discovered nuclear fission. Sure, revenue was way down, margins were in the toilet... but... ARR per head was up! All I could think was, "Congratulations there, big dog, you’ve discovered how fractions work" 🙃 Increasing "efficiency" by making the denominator smaller isn’t operational excellence. It’s just math. If you fire a load of people, ARR per head goes up. But doesn't make the business any better, it just means there's fewer people to fix the underlying problem. If your ARR per head goes up because your revenue is skyrocketing, you’re a genius. If it goes up because you’ve handed out a dozen P45s to try to misdirect from what's actually happening... you’ve failed your company... but most importantly... you've failed your team. If your most important metrics - revenue and margins - are going down, but your ARR per head is going up (and that's what you're focused on)... then, well, good luck to your company in your next role (they'll need it). It's just a trick grifters play for investors and boards. Does anyone actually fall for it? If they do... well, your ARR per head isn't your only problem.

Subject lines in the national news? OMG DROP EVERYTHING this is absolute CATNIP for me 😻 Everyone is wondering how a huge brand could let something so grim slip through the net. People assume there’s a mahogany boardroom where a dozen executives sip espresso and scheme up evil, edgy email campaigns for the purposes of profit. TBH... and I'm saying this having probably spent more time talking about subject lines than any other human on the planet (yay me)... The truth is, marketing execution is rarely a glossy mastermind operation. It’s usually just a haggard guy called Dave staring at a blinking cursor of doom at 4:45 PM on a Friday. Dave has to send out 4 million emails a week. Dave has been up since 4am with a screaming toddler. Dave has had success in the past making jokes about things. So he jumps on a trending hook and to him it's hilarious (and 99/100 times it probably is hilarious) and then SEND BABY SEND! We've all been Dave at some point. I sure have been. I remember once I sent out a newsletter amidst the Greek financial crisis with the subject line "It's Greeky bum time!" and got like 3 really angry emails. Which is a lot. But hey. Let's give Dave some credit. He could have taken the safe route. The beige route. The road most taken. He could have gone, "40% off spa deals today" and called it a day. He could have been lazy. "Lazy old Dave," they'd say, Never takes a risk, eats a Pret ham and cheese like clockwork. Loves a cold Brewdog at 5pm. This was the 1/100 that it didn't work, and it was a pretty brutal mistake. But, don't forget: sometimes you got to judge the effort, and accept that the outcome isn't ideal. You may be wondering, "how could they let that happen?" like "they" are a group of arbiters of good taste. But, consider this: it's a subject line. If a CMO needs to sign off a subject line, then the CMO is probably not a very good CMO. You got to let your players play. Dave took a bad shot. A very bad shot. But, at least Dave tried - so credit to the CMO for giving him the opportunity to mess up. Just maybe let Dave know that this one was a miss, and maybe check in on the next one.

How should startup equity be split? Everyone wants a 50/50 split because it feels "nice" and avoids an awkward chat. But "fairness" in a startup is a total trap. Founding equity should be tied to risk, not fairness. When we started Phrasee, I took 55% and the other two founders split the rest. At the time, everyone agreed it was fair. At the end, perhaps not. But... ... I was the one who went all-in with no safety net from day zero, while they still had their day jobs. And this risk is what startup equity should reflect. A lot of people get this mixed up. They think equity should reflect your contribution in year five. It shouldn’t. Equity is the price of the risk you take to get the engine running on day zero. There are other incentives for when the company grows, for example salary and bonuses and options. But not Ords. This did lead to some weird times later on. As the business grew, it's natural for people to think they deserved more. But the time for that discussion was at the beginning when the valuation was 0, not on day 1,000 when the valuation was tens of millions. You can’t renegotiate the past just because the future looks bright. If you’re starting out, have the uncomfortable conversation now. That's what we did at Drumbeat - and we agreed we should each take 1/3, because we were all taking equal risk when it was worth zero. Let's see what it's worth on day 1,000. Be honest about who is actually putting their neck on the line. And if people want to re-litigate it when your business is booming - tough luck!