Four conversations across enterprise sales, 1920s advertising, games engineering and agentic marketing land on the same uncomfortable finding: producing competent work now costs nothing, which means the only edge left is the unglamorous hours nobody sees you spend - and that is precisely the line item every automation programme cuts first. Jen Abel walks through the fifteen-step motion behind a £100k enterprise close, Kipp Bodnar puts an agentic marketing team against a real company's positioning, Casey Muratori explains why the flood arrived a decade before the robots, and Claude Hopkins turns out to have written the whole argument down in 1923.

Featured in this round-up:

🕒 Estimated Reading Time: 18 mins. Time saved: 284 mins! 🔥

The sharpest disagreement is about whether quality can still be found. Casey Muratori's account of the games industry is unsentimental: once Unity and Unreal removed engine risk, releases went from hundreds to tens of thousands a year, and "there is no way that your game will be organically noticed anymore pretty much period." He is careful to date this. It happened pre-AI. Kipp Bodnar's live experiment points the other way, because his read on the agentic marketing repo is that it lifts the floor rather than flattening the ceiling, which leaves room for the exceptional operator to go further.

The second split is older than both of them. Hopkins, as read by David Senra, is emphatic that once an ad works you stop touching it, because you are advertising to a moving parade rather than a standing army. Jen Abel's enterprise motion is the exact inverse: "I do not have a scripted pitch or frame for any call because it feels artificial."

The third is whether craft is a commercial argument at all. Muratori declines to make one, comparing handwritten code to knitting a hat.

The big tensions to watch:

💡 Top tip: if you have five minutes rather than eighteen, read section three and section eight. One tells you where the machine already beats your team, the other tells you what that does to the economics of doing the work yourself.

Not your topic this week? Try these five instead:

———————————————————————————

Casey Muratori has spent a career on performance engineering and games, and his account of what happened to that industry is the most useful diagnostic in this week's material, because it describes a completed cycle rather than a predicted one. Engine risk used to be real. If you wanted to build a competitor to Warcraft, you built the pathfinder, the renderer and the level editing tools yourself, which is why Blizzard rolling one game's technology into the next was such an advantage. Then Unity and Unreal removed that risk for everybody, and the market behaved exactly as you would expect.

The first-order effect was good. People who could never have shipped a game shipped one, and some of those were excellent. The second-order effect was not. "It's so massive that there is no way that your game will be organically noticed anymore pretty much period," Muratori explains on The Pragmatic Engineer. Steam absorbs tens of thousands of releases a year. Word of mouth, which used to be a sufficient distribution strategy for a genuinely fun game, stopped clearing.

He dates it precisely, and this is the part worth carrying: "again, pre AI, it's just because now the barrier to entry is very low."

Where they split: Bodnar's live test of an agentic marketing system reaches a more optimistic conclusion, that these tools lift the floor and let strong operators climb. Both readings are internally consistent, but Muratori's has a decade of completed evidence behind it and Bodnar's has three weeks. The games market is the control group. Content marketing is about to run the same experiment with a much shorter cycle time.

Key Takeaways:

David Senra's episode on Claude Hopkins is nominally about writing advertisements. It is actually the clearest available statement of where competitive advantage sits when the output itself is easy to copy. Hopkins wrote Scientific Advertising around 1923, and Albert Lasker locked the manuscript in a safe for twenty years because publishing it would have given away the firm's edge. Ten million copies later, the central claim is still doing work.

"Genius is the art of taking pains," Senra quotes. "The advertising man who spares the midnight oil will never get very far." Hopkins was describing a specific asymmetry: the finished ad is simple because it must be simple to appeal, but "back of that ad may lie reams of data, volumes of information, and months of research."

The canonical example is Ogilvy reading through hundreds of pages of Rolls Royce material and pulling one line out of an obscure engineering document that was never meant for public consumption. At sixty miles an hour, the loudest noise in the new Rolls Royce comes from the electric clock. The line took a week to find and thirty years to exhaust.

Muratori and Abel are both describing the same asymmetry in different trades. Abel's fifteen-step enterprise motion produces a demo that looks effortless to the buyer precisely because fourteen steps of intel gathering preceded it. Muratori's performance work is invisible until you use the competitor.

The implication for anyone running a team is uncomfortable, because the input Hopkins is describing has no line item. Nobody bills for the fourteenth engineering document.

Key Takeaways:

Kipp Bodnar did the thing everyone talks about and almost nobody publishes. He installed an agentic marketing repo, stripped out the brand context file so it ran cold, pointed it at a fast-growing AI sales company he had no internal data on, and asked it to audit the website, rewrite copy, redo positioning and work the AI search channel.

His verdict is precise enough to plan against. "I think what we're continuing to show here is AI is better than an average marketer. And a great marketer knows how to take this and turn it into something," Bodnar says on Marketing Against The Grain. On the overall output: "I would rate this good, not very good." On the generated email sequence, less comfortably: "It's better than most emails I get. I actually kind of I'm worried about how good this is."

The failure mode was the headline. The system proposed "AI sales agents that qualify demo and close while your team sleeps," which Bodnar rejects on the grounds that every other AI SDR in a crowded category would say something nearly identical. The success was strategic reframing: the system spotted that buyers tell an AI the truth they withhold from human reps, and that this makes the company a sales intelligence product rather than a cheaper rep. Bodnar calls that better than what an average product marketer would return.

Where they split: Hopkins puts the money in exactly the place the machine is weakest. "It's not uncommon for a change in headlines to multiply returns from five to ten times over," and Ogilvy later reported differences of nineteen to one. The machine is strong on research, synthesis and positioning hypotheses, and weak on the single highest-variance decision in the entire artefact. That is a workable division of labour, but only if you know which half is which.

Key Takeaways:

Jen Abel's enterprise motion is built on a refusal that will sound like heresy to anyone who has ever bought sales enablement software. She does not use a script. "I do not have a scripted pitch or frame for any call because it feels artificial. They feel it, it feels like you're putting them into a square box," she explains on Lenny's Podcast. The alternative is harder to hire for and harder to teach: listen actively, then build the pitch live around what the buyer just told you.

Her observation about who is good at this is the part worth arguing with your head of sales about. "The most successful salespeople are not trained salespeople. That's why founders are so good at this," Abel says. Founders are poor at following process and excellent at pulling on strings, extracting information and getting people excited by a vision.

Where they split: Hopkins says the opposite, and says it with a hundred years of direct response data behind him. Once an ad works, stop changing it. Senra relays Ogilvy's summary of why: "you're not advertising to Standing Army, you're advertising to moving parade." The audience refreshes faster than the creative tires, and the person bored of the ad is the person who wrote it.

The reconciliation is audience size, and it is worth being explicit about it because it determines what you should automate. Hopkins is optimising one artefact against millions of strangers, where variance is expensive and repetition compounds. Abel is optimising one conversation against five named people, where repetition reads as contempt. Automate the moving parade; never automate the room of five. Most companies currently have this backwards, running heavily templated enterprise outreach while nervously rewriting a homepage that was converting fine.

Key Takeaways:

Hopkins built his method by standing in town squares under torchlight, watching pitchmen and, in Senra's phrase, outright fakers work a crowd, noting which appeals lit people's eyes and which got rejected. His instruction was to "study salesmen, canvassers and fakers," on the grounds that no argument in the world compares with one dramatic demonstration. The point was never the snake oil. It was proximity to the moment of decision, observed live rather than reported.

Abel's process is the modern industrial version of the same instinct, and it is almost entirely composed of pre-work. Before any group demo, she is running individual calls, catching each new participant up on what she learned from the last one and asking what they would add. "Do you see how you're just collecting more and more Intel along the way? So now you're building the business case for them. Now they feel like they're buying, not being sold to."

The corollary is a hard rule that cuts against short-term pipeline pressure. When an enterprise organisation invites you to present to four or five people as part of a due diligence sweep, decline. "You're either a checkbox and you don't want, you want to remove yourself from being a checkbox. Or two, you're about to demo your product with no competitive Intel. Like you just lost the deal at that point."

Hopkins has the compressed version of why the pre-work matters more than the performance: "you can't save souls in an empty church." Abel's version is that the deal is won when it "feels like it was built specifically for me. That's the whole game."

Key Takeaways:

The oldest rule in Hopkins' book is also the one generated copy breaks most reliably. "I have seen many an ad killed by a single unfortunate phrase, usually a selfish phrase," Senra reads. Insist on this brand. Avoid imitations. Look out for substitutes. Hopkins' instruction is unambiguous: "Forget yourself entirely. Seek in every word to increase your good impression. Say only what you think a good salesman should say, if that prospect stood before him. Do not boast. Boasting is repulsive."

Set that against Bodnar's generated headline. "AI sales agents that qualify demo and close while your team sleeps" is a sentence entirely about the seller's capability, which is why he rejected it as indistinguishable from every competitor in the category. The system defaulted to the failure mode Hopkins named in 1923, not because it is stupid but because the statistical centre of all marketing copy ever written is a company describing itself.

Abel's framing is the same test applied to a conversation rather than a page. She is unbothered by AI in her workflow, but specific about where it fails: "so many people will create like AI reports that they can send to their clients or like AI imagery... it's like, anyone can do that. It just feels salesy." Her closing instruction is almost tactile - "Emails are living and breathing. I swear to God that like people can read you saying the email out loud."

The generalisable test is cheap to run and slightly humiliating. Read your headline, your email and your first demo slide, and count the sentences that describe you rather than the buyer's situation. Hopkins' alternative, which is the harder move, is to earn the claim: "Tell the pains that you take to excel."

Key Takeaways:

Muratori's most commercially interesting observation is that the decade of complaining about slow software appears to have worked. "I'm seeing a lot of new emphasis on performance, people talking about performance, people posting benchmarks on things," he says, and more usefully, he names the pattern: new entrants attacking established product categories with performance as the entire pitch. FilePilot. The Blick video editor. "Like things like this that have been coming out lately, where it's like, oh, really performance software to try to take on incumbents in a space. And they've been getting traction."

This is quality functioning as positioning rather than as an engineering virtue, which is a different argument from the one performance advocates usually make. It is also Hopkins' argument, transposed. "The maker is too close to his product. He sees in his methods only the ordinary. He does not realize that the world at large might marvel at those methods." The instruction that follows is the one most engineering-led companies never act on: tell the pains that you take to excel.

Where they split: Muratori immediately limits his own optimism, and the limit matters. Performance "can't sell a product on its own into a monopoly space." If you are an app someone chooses to download, speed is a viable wedge. If your category runs through platforms people are required to use, the buyer's experience is not the buyer's choice, and the wedge disappears. His read is that the discretionary-download slice of software is shrinking while the mandated-platform slice grows. That maps almost exactly onto the enterprise procurement world Abel operates in, where the person suffering the software and the person signing for it are rarely the same human.

Key Takeaways:

Muratori is asked directly about AI coding agents and refuses the commercial frame entirely, which is the most honest moment in any of this week's material. He is not interested in arguing that handwritten code is faster or better. "I would like to be part of whatever the set of people are who are going to keep this traditional craft alive just because that's something humans do, not because we're trying to say that's the right business case."

He extends it into the knitting analogy, and it lands harder than it sounds. Anyone can buy a wool hat. Someone is knitting one right now. Some people buy the yarn, some raise the sheep, and there is always somebody further along the chain than you. Handmade is a preference, exercised at whatever depth the person chooses, and it does not require an economic justification.

Abel is in the same place on capability and the opposite place on economics. Asked whether AI will take enterprise sales, she is untroubled: "I'm not worried about AI replaced because there's so much about it that's just human, right? Like, do you vibe with this person?" She is not defending craft as a human tradition. She is billing for it, at £100k minimum deal sizes, on the grounds that enterprise buyers are more sensitive to the process you take them through than to the product demonstration itself.

Where they split: this is the week's real disagreement, and Bodnar's data adjudicates it. Craft pays only where the market can still perceive the gap above the automated baseline, and that baseline just moved to "good, not very good." Muratori's honesty is the right default position for individuals and the wrong one for teams. If you are choosing how to spend your evenings, no justification is required. If you are defending headcount, you need Abel's version, which means being able to name the specific perceivable difference your craft produces for a buyer who is comparing you to a competent machine.

Key Takeaways:

Games solved a version of this problem twenty years ago, under financial pressure, and the answer generalises further than the industry realises. Muratori describes the shift away from building an engine and hoping the gameplay emerged. Projects were too expensive to discover at the end whether the core loop was fun, and the famous cautionary examples are games where the gameplay only came together right at the finish, leaving no time to polish.

The fix was to prove the hardest uncertainty first and as cheaply as possible. "What we're going to do is we're going to focus the entire studio on building one vertical gameplay slice as fast as we can as hacky as we can, whatever we have to do, do that, prove that that is engaging to play. And then start building out everything else." Only once the slice is running do you build the spreadsheets and asset schedules, because now you believe in the thing.

Abel runs the identical sequence with people instead of prototypes. The expensive artefact is the group demo, so the uncertainty gets resolved before it: two or three individuals already excited, each of whom contributed intel that shaped the pitch. By the time the room assembles, the risky part is already proven.

Bodnar's method is the same discipline applied to tooling. He deleted the brand context file and ran the system cold, because a clean test of the capability was worth more than a flattering one.

Key Takeaways:

———————————————————————————

Three unresolved questions from this week's brief worth carrying into your team:

Want more of the same? Try these five:

That’s a wrap.

As always, the journey doesn't end here!

Please share and let us know what you liked or want changing! 🚀👋

Alastair 🍽️.