Latest / Best of LinkedIn: Digital Transformation & Tech / Best of LinkedIn: Digital Products & Services CW 04/ 05

Transcript

Best of LinkedIn: Digital Products & Services CW 04/ 05 · Best of LinkedIn: Digital Transformation & Tech

• 0:00
  Speaker 0: This episode is provided by Thomas Allgaier and Frennus based on the most relevant LinkedIn posts about digital products and services in calendar weeks O four and O five.
• 0:10
  Speaker 0: Frennus is a B to B market research company helping enterprises gain the market customer and competitive insights needed to drive customer centric and cost efficient product development.
• 0:20
  Speaker 1: Welcome back to the deep dive.
• 0:22
  Speaker 1: We've got a A really interesting stack of sources today, covering late January and early February, twenty twenty six.
• 0:29
  Speaker 0: Yeah.
• 0:30
  Speaker 0: And looking through them, I feel like we're, I don't know, maybe turning a corner.
• 0:33
  Speaker 1: That's exactly it.
• 0:34
  Speaker 1: It feels like the conversation is finally shifting.
• 0:36
  Speaker 0: Shifting how?
• 0:37
  Speaker 0: Because for what, the last two years, it's just been this endless cycle of, look at this cool AI
• 0:42
  Speaker 1: demo.
• 0:42
  Speaker 1: Right.
• 0:43
  Speaker 1: Check out this magic trick.
• 0:44
  Speaker 1: And the sources from these past couple weeks, they suggest that the shiny new toy hangover is finally clearing up.
• 0:51
  Speaker 0: Okay, so we're moving from the magic to the mechanics.
• 0:53
  Speaker 1: The mechanics.
• 0:54
  Speaker 1: That's the perfect word for it.
• 0:55
  Speaker 1: It's less about, can AI do this?
• 0:57
  Speaker 1: And much, much more about, okay, how do we actually build a sustainable operational business with this?
• 1:03
  Speaker 0: Which I have to say is a relief.
• 1:05
  Speaker 0: But it also sounds like the hard part.
• 1:07
  Speaker 0: The plumbing.
• 1:08
  Speaker 1: It is the plumbing.
• 1:09
  Speaker 1: Yeah.
• 1:09
  Speaker 1: But that's where the value is.
• 1:11
  Speaker 1: We've got insights here on everything from, you know, vibe coding to the real nitty gritty of product ops.
• 1:17
  Speaker 0: And the common thread seems to be rigor.
• 1:20
  Speaker 1: Rigor.
• 1:21
  Speaker 1: Whether it's AI, leadership, or strategy, it seems like the free ride of just winging.
• 1:26
  Speaker 1: it is over.
• 1:27
  Speaker 0: Okay, so let's start with that term then.
• 1:29
  Speaker 0: Vibe coding.
• 1:30
  Speaker 0: I've seen it all over LinkedIn, all over Twitter.
• 1:33
  Speaker 0: It sounds like something you'd do in a hammock, not in a serious engineering department.
• 1:37
  Speaker 1: It does sound super casual, doesn't it?
• 1:39
  Speaker 1: I mean, in the general sense, vibe coding is this idea of using AI to generate code based on a loose idea of vibe.
• 1:47
  Speaker 0: You just say, make me a retro landing page and poof, it exists.
• 1:50
  Speaker 1: Exactly.
• 1:51
  Speaker 1: And that usually terrifies traditional engineers because the code underneath.
• 1:54
  Speaker 1: Towards
• 1:54
  Speaker 0: a mist.
• 1:55
  Speaker 1: It's a complete mess, right?
• 1:56
  Speaker 1: It might work, but you do not want to look into the hood.
• 1:59
  Speaker 0: It's unmaintainable.
• 2:00
  Speaker 1: Precisely.
• 2:01
  Speaker 1: But this is where we get our first really big insight.
• 2:03
  Speaker 1: It's from Lenny Rochetsky's newsletter.
• 2:05
  Speaker 1: And he was focusing on a workflow by Zivi Arnovitz.
• 2:08
  Speaker 0: And Arnovitz completely flips this idea on its head.
• 2:11
  Speaker 1: He does.
• 2:12
  Speaker 1: He argues that effective AI coding isn't vibey at all.
• 2:15
  Speaker 1: It's actually rooted in extreme structure.
• 2:19
  Speaker 1: And this guy, Zivi, he's a non-technical PM, but he's shipping code like a senior dev.
• 2:25
  Speaker 0: I was reading about his process.
• 2:27
  Speaker 0: He uses these things he calls slash commands.
• 2:29
  Speaker 0: It kind of reminded me of old school slack
• 2:31
  Speaker 1: commands.
• 2:32
  Speaker 1: It's exactly that.
• 2:33
  Speaker 1: So instead of treating the LLM like a chat buddy, he treats it like a junior developer who needs very, very specific management.
• 2:39
  Speaker 1: He has these preset commands like create issue or plan or even peer review.
• 2:44
  Speaker 1: Let's
• 2:44
  Speaker 0: pause on that because I think that's a huge shift for anyone listening.
• 2:47
  Speaker 0: when he types create issue what's actually happening behind the scenes.
• 2:50
  Speaker 1: So he's not just asking for code.
• 2:52
  Speaker 1: That command triggers this really rigid template in the AI's instructions.
• 2:57
  Speaker 1: It forces the AI to state the problem, identify the files, propose a solution before it writes a single line of code.
• 3:05
  Speaker 0: It stops it from just making something up.
• 3:06
  Speaker 1: It stops the hallucination.
• 3:08
  Speaker 1: But the one that really got me was peer review.
• 3:11
  Speaker 0: That's the killer feature.
• 3:12
  Speaker 1: Absolutely.
• 3:13
  Speaker 1: Because when you type peer review, you are forcing the AI to switch personas.
• 3:18
  Speaker 1: It stops being the builder.
• 3:20
  Speaker 1: who just wants to get it done.
• 3:21
  Speaker 1: And it
• 3:22
  Speaker 0: becomes the critic.
• 3:23
  Speaker 1: Exactly.
• 3:24
  Speaker 1: Which is such a smart workaround?
• 3:25
  Speaker 1: because these models, they're designed to be agreeable, you know, they want to please you.
• 3:30
  Speaker 1: They rarely say, wait, that's a terrible idea.
• 3:33
  Speaker 0: So you're basically encoding best practices right into the workflow, giving it a framework to critique its own work.
• 3:40
  Speaker 1: And here's the best part.
• 3:42
  Speaker 1: If the AI messes up.
• 3:44
  Speaker 1: He doesn't just hit regenerate and hope for a better dice roll.
• 3:46
  Speaker 0: No, he debugs the process.
• 3:48
  Speaker 1: Yes.
• 3:49
  Speaker 1: He asks the AI, what in my instructions caused you to make this mistake?
• 3:54
  Speaker 1: And then he updates the slash command so it doesn't happen again.
• 3:57
  Speaker 1: That
• 3:57
  Speaker 0: ties right into this whole slot problem.
• 3:59
  Speaker 0: There's so much buggy generic AI code out there.
• 4:03
  Speaker 1: And the inside is the slop isn't an AI problem.
• 4:06
  Speaker 1: It's a people problem.
• 4:07
  Speaker 1: Lazy prompts get lazy results.
• 4:09
  Speaker 0: Zevi actually calls his approach exposure therapy.
• 4:12
  Speaker 1: Exposure therapy?
• 4:13
  Speaker 1: Yeah.
• 4:14
  Speaker 1: He says to start with just a project context doc in chat GPT before you even touch a real coding tool.
• 4:20
  Speaker 1: You have to train yourself to give the right context first.
• 4:23
  Speaker 0: It's fascinating.
• 4:24
  Speaker 0: But you know, on one hand, we have this super structured text heavy workflow that's empowering builders.
• 4:29
  Speaker 0: On the other hand.
• 4:30
  Speaker 0: We have a source here, Manuel Silva, who's basically sounding the alarm.
• 4:34
  Speaker 1: This is such a necessary counterpoint.
• 4:36
  Speaker 0: He says we're seeing a forty-year regression in design.
• 4:39
  Speaker 1: Think about it, we spent decades moving away from the command line, you know, MS-DOS typing obscure codes to graphical interfaces, buttons, sliders, visual cues.
• 4:49
  Speaker 0: And now you open a so-called modern AI app.
• 4:53
  Speaker 0: And what is it, a blinking cursor in a text box?
• 4:55
  Speaker 0: Tell me what you want.
• 4:56
  Speaker 1: Right.
• 4:57
  Speaker 1: And Silva's point is that we're calling this futuristic, but for an average user, it's a huge cognitive load.
• 5:03
  Speaker 1: We're making our parents learn prompt engineering just to edit a photo.
• 5:07
  Speaker 1: It's kind
• 5:07
  Speaker 0: of lazy design, isn't it?
• 5:08
  Speaker 0: You're just delegating the work to the user instead of designing a clear button that says remove background.
• 5:13
  Speaker 1: It is the command line all over again, just with, you know, better autocorrect.
• 5:18
  Speaker 1: A great interface should show you what's possible.
• 5:21
  Speaker 1: An empty textbox just assumes you already know.
• 5:23
  Speaker 0: But wait, if we follow this thread to its logical end, maybe we get to a place where the interface doesn't matter at all because humans aren't even using it.
• 5:32
  Speaker 1: The no UI future.
• 5:33
  Speaker 0: Yeah, Akash Gipta had this prediction about agent to agent negotiation.
• 5:37
  Speaker 0: that kind of blew my mind.
• 5:39
  Speaker 1: The scenario he describes is so compelling.
• 5:41
  Speaker 1: Yeah.
• 5:41
  Speaker 1: Imagine you want a better interest rate on your savings account.
• 5:44
  Speaker 0: Ugh.
• 5:45
  Speaker 0: The worst.
• 5:45
  Speaker 0: I have to log in, navigate, twelve menus, maybe get on the phone.
• 5:49
  Speaker 1: In the agent to agent future, you just tell your personal AI, get me a better rate.
• 5:53
  Speaker 1: And your agent pings the bank's agent through an API.
• 5:56
  Speaker 0: So it's machine to machine.
• 5:58
  Speaker 0: No humans.
• 5:59
  Speaker 1: Strictly machine to machine.
• 6:01
  Speaker 1: They negotiate terms, compare offers, and just report back to you.
• 6:04
  Speaker 1: I got you four point five percent.
• 6:06
  Speaker 1: You never touched a screen.
• 6:07
  Speaker 1: There was no user experience, not visually anyway.
• 6:10
  Speaker 0: That have massive implications for product managers.
• 6:13
  Speaker 0: If I'm a PM at that bank, my whole world is button placement and color schemes in this future.
• 6:19
  Speaker 1: You're invisible.
• 6:20
  Speaker 1: If your product is just a pretty wrapper on a generic service, you're done.
• 6:25
  Speaker 1: PMs need to start writing PRD's product requirement documents for APIs.
• 6:31
  Speaker 0: How does my agent talk to your agent?
• 6:33
  Speaker 0: What are the negotiation parameters?
• 6:34
  Speaker 1: It shifts the whole discipline from like psychology of design to economics of APIs.
• 6:40
  Speaker 0: That makes me wonder about brand loyalty.
• 6:41
  Speaker 0: If my agent is just getting me the best math, how do you build a brand?
• 6:44
  Speaker 0: You can't charm an algorithm with a logo.
• 6:46
  Speaker 1: You can, which is a huge shift.
• 6:49
  Speaker 1: And it brings us to our second big theme.
• 6:51
  Speaker 1: the human element.
• 6:52
  Speaker 1: Because if AI is doing the coding and negotiating, what's left for us?
• 6:56
  Speaker 0: Judgment, taste.
• 6:58
  Speaker 1: That's the frontier.
• 6:59
  Speaker 1: Elaine Barsoom had a brilliant post on this.
• 7:01
  Speaker 1: She says, scaling with AI is a training problem, but not the way we usually think.
• 7:05
  Speaker 1: Right,
• 7:05
  Speaker 0: we're not talking about training the model on GPUs.
• 7:07
  Speaker 1: No, we have to train the humans.
• 7:10
  Speaker 1: And she makes this crucial distinction.
• 7:12
  Speaker 1: You train AI on signals.
• 7:14
  Speaker 1: Clicks, scrolls, purchases, behavioral data.
• 7:18
  Speaker 1: But you have to train humans on taste.
• 7:20
  Speaker 1: which is the judgment of what actually matters.
• 7:22
  Speaker 0: I love the phrase she used.
• 7:24
  Speaker 0: Humans retain decision authority and AI keeps input authority.
• 7:28
  Speaker 1: That's the whole game right there.
• 7:30
  Speaker 1: The AI gives you the inputs.
• 7:32
  Speaker 1: Hey, forty percent of users click this, but if you just let the AI make the decision based on that,
• 7:37
  Speaker 0: you get product bloat.
• 7:38
  Speaker 0: You get
• 7:39
  Speaker 1: bloat.
• 7:39
  Speaker 1: AI doesn't know how to say no.
• 7:42
  Speaker 1: It just optimizes the metric.
• 7:43
  Speaker 1: The human has to be the editor.
• 7:45
  Speaker 1: We have to look at the data and say, yes, users are asking for this, but it ruins the product's simplicity, so we're not doing it.
• 7:51
  Speaker 0: But taste is such a fuzzy word.
• 7:53
  Speaker 0: It feels subjective.
• 7:54
  Speaker 0: How do you pin that down?
• 7:55
  Speaker 0: Well,
• 7:56
  Speaker 1: this is where Atira Abdul-Ralph had a great piece distinguishing between product sense and product taste.
• 8:01
  Speaker 0: OK, this is useful.
• 8:02
  Speaker 0: Break that down.
• 8:03
  Speaker 1: So he defines product sense as knowing what to build, solving the right problem.
• 8:08
  Speaker 1: His example was Slack.
• 8:09
  Speaker 0: During the pandemic,
• 8:10
  Speaker 1: right?
• 8:10
  Speaker 1: Exactly.
• 8:11
  Speaker 1: Users were feeling isolated.
• 8:12
  Speaker 1: A purely functional PM or an AI would say, okay, people want to see each other.
• 8:17
  Speaker 1: Let's add a start zoom button everywhere.
• 8:19
  Speaker 1: Which
• 8:19
  Speaker 0: is just more meetings.
• 8:23
  Speaker 0: The last thing anyone wanted.
• 8:25
  Speaker 1: Right.
• 8:26
  Speaker 1: But someone with product sense realizes the problem isn't a lack of meetings.
• 8:30
  Speaker 1: It's a lack of a hallway.
• 8:32
  Speaker 1: That spontaneous low friction connection.
• 8:35
  Speaker 1: And so they build huddles.
• 8:37
  Speaker 0: That's product sense, so what's product taste?
• 8:39
  Speaker 1: Taste is about how it feels, the polish, the intuition.
• 8:43
  Speaker 1: Roth's example here was loom.
• 8:46
  Speaker 1: Loom noticed users were messing up their recordings and getting frustrated.
• 8:49
  Speaker 0: A standard solution would be just add a delay button.
• 8:53
  Speaker 1: Functional, but clunky, kills your flow.
• 8:56
  Speaker 1: But product taste is realizing the user has performance anxiety.
• 8:59
  Speaker 1: They're vulnerable.
• 9:01
  Speaker 1: So Loom adds a quick restart button right next to the stop button.
• 9:03
  Speaker 1: One click, wipes the take, restarts the countdown.
• 9:06
  Speaker 1: It
• 9:06
  Speaker 0: feels like the tool is on your side.
• 9:07
  Speaker 1: That's haste.
• 9:08
  Speaker 1: And you can't really prompt an AI to have that kind of empathy.
• 9:11
  Speaker 1: Not yet, anyway.
• 9:12
  Speaker 0: Which is why leadership has to change, too.
• 9:15
  Speaker 0: Roshan Gupta had a post about moving from a product manager to a product leader.
• 9:19
  Speaker 1: It's a tough transition.
• 9:20
  Speaker 1: As a PM, your whole value is being the best problem solver in the room.
• 9:24
  Speaker 0: But as a leader, you have to stop solving the problems yourself.
• 9:27
  Speaker 1: You have to become a builder of problem solvers, which is terrifying.
• 9:32
  Speaker 1: You have to let go.
• 9:33
  Speaker 1: You own the narrative, not the roadmap.
• 9:36
  Speaker 0: And speaking of narratives, Stephanie Liu had this painfully true post about the illusion of decisiveness.
• 9:42
  Speaker 0: Oh,
• 9:42
  Speaker 1: this one hits close to home for so many companies.
• 9:44
  Speaker 1: It's that moment in the boardroom, everyone agrees on the big bold vision, everyone nods.
• 9:48
  Speaker 1: The
• 9:48
  Speaker 0: slide deck looks amazing.
• 9:50
  Speaker 1: And then Monday morning comes.
• 9:51
  Speaker 1: And sales is chasing a deal that contradicts the vision.
• 9:55
  Speaker 1: Engineering is drowning in tech debt, and marketing is off on a tangent.
• 10:00
  Speaker 0: Everyone is aligned on the vision, but totally misaligned on execution.
• 10:04
  Speaker 0: And
• 10:04
  Speaker 1: Lou points out it's the gray dots that kill you.
• 10:06
  Speaker 1: The
• 10:07
  Speaker 0: gray dots.
• 10:08
  Speaker 1: Can we just add this request?
• 10:12
  Speaker 1: The favor for a big client.
• 10:14
  Speaker 1: The things that aren't on any roadmap, but they suck up forty percent of the team's energy.
• 10:18
  Speaker 1: Okay,
• 10:18
  Speaker 0: so how do we fix that?
• 10:19
  Speaker 0: This moves us into our third theme, right?
• 10:21
  Speaker 0: Operations, discovery, the nitty gritty.
• 10:24
  Speaker 0: The answer seems to be product ops.
• 10:27
  Speaker 1: It is.
• 10:27
  Speaker 1: And for a long time, product ops was seen as just, you know, bureaucracy.
• 10:31
  Speaker 0: More paperwork.
• 10:32
  Speaker 1: Right.
• 10:33
  Speaker 1: But Phil Evans makes this great comparison to DevOps.
• 10:36
  Speaker 1: We don't see DevOps as bureaucracy.
• 10:39
  Speaker 1: We see it as essential infrastructure for shipping code.
• 10:42
  Speaker 1: ProductOps is the infrastructure for shipping decisions.
• 10:45
  Speaker 0: It's a tooling and process that lets the PM focus on strategy instead of fighting fires.
• 10:50
  Speaker 1: And Ed Biden added some really practical advice here.
• 10:52
  Speaker 1: He says most leaders have no idea if a team is off track until it's too late.
• 10:56
  Speaker 0: The big launch day surprise.
• 10:58
  Speaker 0: The
• 10:58
  Speaker 1: worst surprise.
• 11:00
  Speaker 1: So he suggests a four step review rhythm, kickoff, solution review, launch readiness and impact review.
• 11:07
  Speaker 1: But the one most teams skip is the solution review.
• 11:10
  Speaker 0: And why is that one so critical?
• 11:12
  Speaker 1: because it happens before a line of code is written.
• 11:14
  Speaker 1: It's when the solution is just a sketch, a prototype, that it's the cheapest possible time to fix a misunderstanding.
• 11:20
  Speaker 1: Before
• 11:21
  Speaker 0: you have three weeks of engineering sunk cost and emotional attachment to the code.
• 11:25
  Speaker 1: Exactly.
• 11:26
  Speaker 1: But all this process can't come at the expense of the customer.
• 11:29
  Speaker 1: Omar Salem had this great quote, a product idea without validation is just a well-packaged guess.
• 11:35
  Speaker 0: I love that.
• 11:35
  Speaker 0: Validation isn't a phase you check off, it's a principle.
• 11:38
  Speaker 1: You have to challenge your assumptions.
• 11:40
  Speaker 1: not just ask users what they
• 11:42
  Speaker 0: want.
• 11:42
  Speaker 0: And speaking of challenging assumptions, Michael Goitin shared the strategy test called the toggle method.
• 11:47
  Speaker 0: I thought this was a brilliant mental model.
• 11:50
  Speaker 1: It's a tough one.
• 11:51
  Speaker 1: Yeah.
• 11:52
  Speaker 1: So every strategy have a where to play your customers and how to win your differentiator.
• 11:57
  Speaker 0: Okay.
• 11:58
  Speaker 1: Goitin says to take your how to win and toggle your where to play.
• 12:03
  Speaker 1: to a competitor's customer base.
• 12:05
  Speaker 0: Let's try it.
• 12:06
  Speaker 0: Say my where to play is creative agencies and my how to win is a simple intuitive interface.
• 12:13
  Speaker 1: Okay, now toggle the customer.
• 12:14
  Speaker 1: Let's say accountants.
• 12:16
  Speaker 1: Does simple intuitive interface work as a differentiator for accountants?
• 12:20
  Speaker 1: Well,
• 12:21
  Speaker 0: yes.
• 12:22
  Speaker 0: Everyone wants a simple interface.
• 12:23
  Speaker 1: Exactly.
• 12:24
  Speaker 1: Your strategy failed the test.
• 12:26
  Speaker 1: If your differentiator works just as well for a totally different segment, it's not a differentiator, it's just table sticks.
• 12:32
  Speaker 0: So a real strategy would be something like, we visualize timelines in a way that only makes sense for video production workflows.
• 12:38
  Speaker 1: Yes.
• 12:39
  Speaker 1: Because if you showed that to an accountant, they would hate it.
• 12:42
  Speaker 1: That means you've made a real choice.
• 12:43
  Speaker 1: You've sacrificed one market to win another.
• 12:46
  Speaker 0: I feel like a lot of PowerPoint strategies would fail that test.
• 12:49
  Speaker 0: Most
• 12:49
  Speaker 1: would.
• 12:50
  Speaker 0: Okay.
• 12:50
  Speaker 0: Before we wrap up, I want to get to that technical nugget you found.
• 12:53
  Speaker 0: This is for the e-commerce folks listening.
• 12:54
  Speaker 0: It's from John Roberts about how LLMs like ChatGPT actually see products on Shopify.
• 13:00
  Speaker 1: Right.
• 13:01
  Speaker 1: This is a perfect example of mechanics over magic.
• 13:04
  Speaker 1: We just assume AI knows things.
• 13:06
  Speaker 1: But Roberts explains that these LLMs, they often prioritize reading the body thermal of a product page.
• 13:13
  Speaker 0: And they do not necessarily read the SEO meta descriptions we've all spent years optimizing.
• 13:18
  Speaker 0: That's
• 13:18
  Speaker 1: huge, right?
• 13:19
  Speaker 1: The meta description is often just marketing fluff.
• 13:21
  Speaker 1: The body terminal is where the rich details are, materials, the use cases, the siding.
• 13:25
  Speaker 1: That is the data the LLM needs to answer a real question.
• 13:28
  Speaker 1: So
• 13:28
  Speaker 0: the advice is, stop writing for a search engine spider from twenty fifteen.
• 13:33
  Speaker 0: Write detailed, rich descriptions in the main body text.
• 13:37
  Speaker 1: If you want an AI to recommend your shirt when someone asks, what's a good shirt for a hot climate?
• 13:43
  Speaker 1: The answer, one hundred percent organic, cotton, breathable weave, needs to be in that body text.
• 13:50
  Speaker 1: If it's buried in a metatag, the AI might just miss it.
• 13:53
  Speaker 0: It all comes full circle back to the beginning.
• 13:55
  Speaker 0: Whether it's ZV's slash commands or writing product descriptions, you have to understand how the system actually works, not just how you wish it did.
• 14:04
  Speaker 1: The era of vibes is ending.
• 14:06
  Speaker 1: The era of understanding the system is here.
• 14:08
  Speaker 0: So if you're listening, I guess the question to ask yourself is, Are you building for the shiny toy version of the world?
• 14:14
  Speaker 0: Or are you building for the reality of agents, rigor, and structured workflows?
• 14:18
  Speaker 1: And maybe one final thought to leave with.
• 14:19
  Speaker 1: We talked about taste being this uniquely human skill.
• 14:22
  Speaker 1: But as we get better at encoding our preferences into these systems, into these slash commands, are we just teaching the AI to have taste?
• 14:29
  Speaker 1: At what point does the human element just become another dataset we hand over?
• 14:33
  Speaker 0: That is a slightly terrifying thought to end on.
• 14:36
  Speaker 0: but a necessary one.
• 14:38
  Speaker 0: If you enjoyed this episode, new episodes drop every two weeks.
• 14:41
  Speaker 0: Also check out our other editions on ICT and tech, artificial intelligence, cloud, sustainability, and green ICT, defense tech, and health
• 14:49
  Speaker 1: tech.
• 14:50
  Speaker 1: Thanks for listening.
• 14:51
  Speaker 0: Don't forget to subscribe for more insights from the next calendar week.
• 14:53
  Speaker 0: See you then.