Best of LinkedIn: Digital Products & Services CW 36/ 37

00:00:00: This episode is provided by Thomas Allgaier and Frennus, based on the most relevant LinkedIn posts about digital products and services from CW-thirty six and thirty seven.

00:00:09: Frenness is a B to B market research company that supports enterprise product teams with building feature by future competitive intelligence that shows exactly how their product stacks up against competition.

00:00:19: you can find more info in description

00:00:22: If you are leading digital transformation in your organization right now, or building software products.

00:00:28: You can basically consider this your custom tailored briefing.

00:00:33: We've synthesized a massive stack of recent industry insights to pull out the actual ground level trends that reshaping digital product and services.

00:00:43: I mean, imagine generating a flawless fully functional piece of software in like one afternoon only to realize you just built the exact wrong thing but faster than anyone in history.

00:00:52: Yeah!

00:00:52: The ultimate nightmare scenario

00:00:54: exactly so that is the mission for this deep dive.

00:00:57: we're cutting through the noise to figure out how to avoid that scenario because as we were going through the material there was this one underlying thread they kept hitting us over-the head.

00:01:07: AI has permanently changed the speed building products.

00:01:12: as a direct result of that, human judgment and strategic alignment have suddenly become like the rarest most valuable resources in the tech industry.

00:01:20: Oh totally!

00:01:21: The historical bottleneck has completely inverted.

00:01:24: I mean for decades the hardest part of software development was literally just writing the code right provisioning the infrastructure.

00:01:30: but today...the hardest part is figuring out if that code should even exist in the first place.

00:01:36: Yeah, which is wild to think about.

00:01:38: and I want to start right there with how this dynamic is completely redefining the product manager role.

00:01:43: Elena Linova shared this insight that i think frames it perfectly.

00:01:46: she compares the current obsession with mastering AI mechanics so you know prompt engineering or autonomous agents.

00:01:52: She pairs that to the agile software craze from fifteen years ago.

00:01:55: Oh, the parallel holds up so well.

00:01:57: I mean think back to like twenty ten.

00:02:00: oh

00:02:00: boy

00:02:01: entire organizations were just obsessing over scrum ceremonies story points certifications post-it

00:02:07: notes everywhere

00:02:08: exactly.

00:02:10: people honestly thought that standing in a circle for fifteen minutes every morning magically made them product visionaries.

00:02:16: yeah and today we're seeing the exact same behavior Just with AI tools

00:02:20: right?

00:02:21: So professionals are learning how to automate user research or how to synthesize feedback in seconds, or generating a product requirements document ten times faster.

00:02:30: Which is great.

00:02:31: but speed and writing at PRD does not matter if the feature you're building actively destroys your unit economics?

00:02:37: Totally.

00:02:37: Understanding a company's revenue model its customer acquisition cost it's lifetime value.

00:02:43: that matters infinitely more than knowing the perfect sequence of prompts to generate wireframe.

00:02:47: Okay so let's unpack this bit because what does the actual job look like?

00:02:55: Steven Wunker wrote this piece in Forbes and he describes modern product management as a barbell roll.

00:03:01: A Barbell Roll.

00:03:02: Yeah, He argues that the PM is now carrying two distinct really heavy loads at opposite ends of the spectrum.

00:03:08: so on one end basically like developers.

00:03:13: They're using AI to mine data, build prototypes even write initial code without having to wait for an engineering handoff which

00:03:21: is incredibly empowering right.

00:03:23: but then on the other end they are burdened with these massive stewardship duties guarding output quality ensuring data security maintaining the strategic standard

00:03:35: creates just immense tension because the PM is no longer orchestrating the work of others, they are actively generating technical output while simultaneously acting as a final checkpoint for compliance and strategic fit.

00:03:48: Right!

00:03:48: And let me play devil's advocate here... Because if AI is basically one thousand horsepower engine it feels like everyone is obsessed with pushing the pedal to floor but nobody is installing steering

00:03:59: wheel?

00:03:59: That's great way

00:04:00: Right.

00:04:01: Like, are we actually expecting one human being this new barbell PM to operate the engine build a chassis and steer the car all at the same time?

00:04:10: I mean that sounds like a recipe for a massive crash not a sustainable job description.

00:04:15: Well yeah And That is precisely The systemic risk We're facing.

00:04:18: Josh Hellman recently issued This really stark warning about that exact dynamic.

00:04:24: He focused on the rise of AI slop in product development.

00:04:28: Ah, AI slop.

00:04:29: Like the content slop we see online?

00:04:31: Exactly!

00:04:32: You know those soulless AI-generated articles that use a ton of words to say absolutely nothing...

00:04:36: Oh yeah, We've all read them

00:04:38: Right.

00:04:38: Well Elmond points out because the cost building software has effectively dropped near zero.

00:04:43: The temptation is just cram every possible feature into product.

00:04:48: Let's vibe code it in an afternoon and see if users like approach

00:04:52: Which exactly how a product loses its narrative.

00:04:56: Because product management is fundamentally about storytelling and judgment.

00:05:01: It's about curating an experience.

00:05:03: when you accelerate the output without accelerating your judgement, don't actually get a better product You just get feature bloat faster.

00:05:10: that makes a lot of sense

00:05:12: Yeah And Amit Raul Actually mapped out how this shifts The day-to-day skill stack especially during the discovery phase.

00:05:19: So the traditional question was always what should we build?

00:05:22: Right.

00:05:22: But the AI-enabled PM has to ask entirely different questions, like should this workflow even involve AI in first place?

00:05:29: Right!

00:05:30: Do we need

00:05:30: it?!

00:05:31: Exactly!!

00:05:32: Can we audit data that is feeding

00:05:33: it??

00:05:34: And crucially what happens when the AI inevitably hallucinates or makes a wrong decision?

00:05:41: Yeah and that question What Happens When It's Wrong?

00:05:44: Is The One That Frankly Nobody Wants To Ask.

00:05:46: When They Are Mesmerized By Some Shiny New Prototype On Their Screen Nobody

00:05:49: Ways To Ruin Magic

00:05:51: Exactly.

00:05:51: But if judgment and curation are the new modes, how does that change the way we plan?

00:05:57: I mean, If take this thousand horsepower speed to apply it strategy.

00:06:01: The traditional frameworks start break down pretty fast.

00:06:04: Wendy Sturges made a super bold claim on this front.

00:06:08: She argues that the annual sauce roadmap is entirely dead, completely

00:06:12: dead

00:06:12: right and then it actually serves as a liability in a fast-moving market.

00:06:16: So she advocates for what?

00:06:17: She calls A rolling horizon model built around roughly five month cycle which

00:06:23: honestly makes mechanical sense.

00:06:24: I mean if you commit to a rigid twelve month plan And then a foundational AI capability drops in Month two That changes your entire industry's baseline

00:06:33: Which happens like every week now.

00:06:35: Exactly if you stick to that original roadmap it means your spending the next ten months actively defending decisions that are objectively obsolete.

00:06:42: right but in real quick.

00:06:43: just as an aside, If You're listening and finding this deep dive helpful for navigating all these shifting frameworks.

00:06:50: um take a second hit subscribe.

00:06:53: so don't miss our future additions.

00:06:55: we really want keep bringing this actionable intelligence straight.

00:07:00: But okay, let's hold on a second and look at the reality of B to be enterprise software.

00:07:05: I understand the theory of a five month rolling horizon?

00:07:09: I really do.

00:07:10: but Enterprise clients demand annual or even multi year commitments before they sign a massive contract.

00:07:16: Oh Of course you

00:07:17: literally cannot walk into a fortune-five hundred procurement office And tell them yeah we don't have a roadmap We're just running on a rolling five month vibe.

00:07:24: no They would laugh you out of the room

00:07:26: exactly.

00:07:27: so how does that actually work in practice?

00:07:29: Well, it requires a major shift in how you communicate those commitments.

00:07:33: You have to stop committing to specific rigid features on specific dates and start committing to solving specific business problems over that horizon.

00:07:41: but do your point when timelines compress like that?

00:07:45: Your ability to rapidly filter signal from noise becomes your only real survival mechanism.

00:07:50: Yeah!

00:07:50: That makes sense.

00:07:51: Alex Bass shared a case study for his time as Chief Product Officer at Typeform that illustrates this perfectly.

00:07:57: He walked into an organization that was sitting on a hundred and seventeen scattered highly debated feature ideas.

00:08:03: Oh, wow A hundred and Seventeen

00:08:05: yeah just a massive backlog.

00:08:06: And instead of spending months debating opinions in trying to please all the stakeholders They ran a max diff study layered with willingness to pay data.

00:08:14: Okay

00:08:15: Let's pause on that mechanism for a second because I'm Max diff study is very specific tool?

00:08:19: And i think it's important.

00:08:20: if you Just ask your customer hey do You want this new reporting dashboard?

00:08:23: they will always say yes

00:08:24: Always free features.

00:08:26: Yes Please

00:08:27: Right.

00:08:28: But in MaxDiffForce's trade-offs, it asks the user do you want the reporting dashboard or Do You Want The CRM integration to run twice as fast because you can only pick one?

00:08:38: Exactly

00:08:39: and mathematically corners the user into revealing what they actually value.

00:08:42: right.

00:08:43: And that force prioritization is where the absolute truth lives.

00:08:47: Yeah.

00:08:47: So by layering that tradeoff data with actual willingness To pay metrics yeah.

00:08:52: They turned those a hundred and seventeen subjective ideas Into forty evidence backed priorities almost immediately.

00:08:58: That is incredible!

00:08:59: Yeah, and they discovered that nearly eighty percent of customer needs overlap completely across their different segments.

00:09:04: so the takeaway here is profound.

00:09:06: modern product teams do not have a building problem.

00:09:10: They Have A Filtering Problem.

00:09:11: It

00:09:11: all comes down to filtering.

00:09:13: But when you filter that aggressively, how you document the surviving decisions becomes the next critical point of failure.

00:09:19: Oh

00:09:19: absolutely.

00:09:19: Which brings up this massive debate happening right now in the industry?

00:09:22: If we are moving at agentic speed rolling our horizons every five months Generating high fidelity prototypes and an afternoon Why write specs at all?

00:09:32: Write the anti-documentation movement

00:09:34: Exactly.

00:09:35: A lot of teams are arguing That traditional product requirements document The PRD is obsolete.

00:09:41: But to me, relying solely on a high-fidelity prototype is like handing the client beautiful three D rendering of house.

00:09:49: All this analogy?

00:09:50: Right!

00:09:50: The client loves it and lighting looks amazing.

00:09:54: but that rendering does absolutely nothing for construction crew trying figure out where load bearing walls go or where plumbing needs to be routed.

00:10:02: Analogy cuts to the very core of what Ed Biden and Rishav Gupta are arguing against right now, because there is a very dangerous narrative out there that vibe coding can just replace written specifications entirely.

00:10:15: Yeah!

00:10:16: But Rishabh points out that a prototype in PRD execute two fundamentally different jobs for business.

00:10:22: A prototype communicates intent forward.

00:10:25: It shows stakeholders' vision on what user will eventually experience.

00:10:29: but a written spec serves an entirely different master

00:10:31: Exactly.

00:10:33: A written spec is an accountability artifact that looks backward.

00:10:37: The value of a prd doesn't come due on launch day its value comes due six or twelve months later On your absolute worst day as a company.

00:10:45: like when something goes wrong.

00:10:46: Yes, When there is a critical security breach Or a massive spiking customer churn?

00:10:51: Or an audit from a regulatory body.

00:10:54: in That crisis room nobody cares what the figma file looked Like.

00:10:58: they're asking Who approved a specific API endpoint?

00:11:01: What data privacy trade-offs were made and based on what evidence.

00:11:05: Yeah, A Dateless Figma file won't save you in an SOC to audit

00:11:08: it absolutely will not.

00:11:10: And AI context windows do not inherently retrieve human decision history or operational state.

00:11:15: So if you skip the documentation just because building is fast You aren't actually moving faster.

00:11:19: your just crashing without a black box recorder.

00:11:21: Wow,

00:11:22: okay.

00:11:23: so the PM role is stretching roadmaps are shrinking to rolling horizons and vital documentation has been bypassed.

00:11:29: but this friction isn't just happening at the individual level.

00:11:33: it's breaking down boundaries of entire teams.

00:11:36: let us look at operational chaos inside.

00:11:41: Douglas Ferguson recently interviewed the chief product and technology officer at Miro, And they detailed how traditional lanes of product development are just blurring entirely.

00:11:51: Designers are submitting code PMs delivering high fidelity interactive prototypes Engineers orchestrating autonomous agents

00:11:59: Everyone is in everyone else's lane Right!

00:12:01: The output of individual contributors has gone up ten X But the actual bottleneck has shifted to cross-functional decision making.

00:12:09: And that friction exists because the new tools are not evenly distributed across the disciplines.

00:12:15: Patty Hannon made a really sharp observation about this imbalance, he points out that AI coding agents—the tool driving this massive acceleration —are currently hyper-optimized for one specific persona... ...the software

00:12:28: engineer.".

00:12:28: Okay why is that?

00:12:29: Well…because code is deterministic!

00:12:31: It either compiles or it doesn't which makes it perfect playground for large language models to test and correct themselves.

00:12:37: Ah right but product strategy and user experience design, those are totally subjective.

00:12:42: They require human consensus.

00:12:44: Exactly!

00:12:44: Therefore because the agent's heavily favored developers right now... ...the Product & Design functions are left stranded in basically pre-agentic workflows.

00:12:54: So they're still doing things the old way?

00:12:56: Yes.

00:12:57: They are still trying to coordinate strategy and notion pinging each other endlessly in Slack And managing states into traditional static documents.

00:13:05: There is no shared operating system for a cross-functional team to move together at this new velocity.

00:13:10: That

00:13:11: sounds incredibly frustrating!

00:13:12: It IS.

00:13:13: Patty's actually attempting to solve it by building the tool called CHI, which sits on top of a coding harness and that goal there is create a shared canvas.

00:13:23: So while the engineer is generating code The PM & designer remain anchored into exact same customer intent in real time rather than constantly trying play catch up with the repo.

00:13:33: That makes total sense.

00:13:34: But let's zoom out even further for a second.

00:13:36: If the product team is struggling to keep their internal boundaries intact, how does the rest of the company function?

00:13:42: You know because you can't have a product teams sprinting on rolling five month horizons while all other enterprise operates in traditional fiscal calendar...

00:13:50: You really cant!

00:13:52: And Stephanie Liu offered harsh reality check on this exact misalignment.

00:13:57: Companies love to announce they are transitioning into new product operating model to keep up with AI, like it's a magic wand.

00:14:04: Right

00:14:04: big reorg announcement

00:14:06: but those transformations almost always plateau.

00:14:09: and the reason is that departments entirely outside of the product teams control specifically sales.

00:14:15: in finance.

00:14:16: they absolutely refuse

00:14:19: Because sales is out there closing massive enterprise deals by promising rigid a twelve month delivery dates for specific features, they need that certainty to close the deal.

00:14:30: Exactly!

00:14:31: While finance is demanding predictable capital expenditure models which heavily conflicts with variable token costs of running AI infrastructure you simply cannot fix operational speed of an entire company solely from inside product department.

00:14:44: it doesn't work.

00:14:45: And if the internal structure is buckling like that, The external business models of digital services are facing an existential threat too.

00:14:51: A seamalic broke down a dynamic That's frankly terrifying If you run an agency or consulting firm right now.

00:14:59: Right Because historically the pricing model for Digital Services has relied on very simple equation Headcount multiplied by unit rate multiplied by duration.

00:15:09: Right,

00:15:09: the classic time and materials math?

00:15:11: Exactly!

00:15:12: But AI is actively collapsing all three of those variables simultaneously.

00:15:16: a project that used to require a pot of twelve people now requires a pot-of-three.

00:15:21: The duration shrinks from six month engagement To A Three Week Sprint And the blended rate plummets because expensive human labor hours are being replaced By incredibly cheap LLM token costs

00:15:34: headcount, rate and duration all compressed at the exact same time.

00:15:37: The price tag of building digital services doesn't just experience a slight margin squeeze —the entire revenue model just

00:15:44: evaporates.".

00:15:44: It's gone!

00:15:45: Yeah software development is transitioning from an artisanal bespoke craft to industrial scale production

00:15:51: which means if you survive the build phase...you face a market that has flooded with industrial-scale software.

00:15:57: so how do you actually take a product to market when everyone else is moving just as fast?

00:16:02: The go-to market playbook has completely rewritten itself.

00:16:05: Ryan Carruthers observed that.

00:16:06: the traditional biannual big push, you know?

00:16:09: The massive launch event anchored by a glossy cinematic demo video.

00:16:13: That is effectively

00:16:14: dead.

00:16:15: Oh totally dead.

00:16:16: By the time your marketing team finishes polishing that glossy video ten competitors have already launched similar capabilities.

00:16:23: Yeah and your announcement is just drowned out by the sheer volume of noise in the market

00:16:28: Right.

00:16:28: So companies like Notion, Rippling and Profound they're proving that the new go-to market strategy relies on frequent highly interactive releases.

00:16:36: you don't ask the user to do them mental heavy lifting of imagining how a feature works based on video.

00:16:42: You embed interactive walkthroughs directly into launch announcement allowing buyer click through immediately

00:16:49: Because tactile proof beats theoretical marketing every single time but strategically about what actually deserves that kind of go-to market energy.

00:17:00: A tier at Dual Roof shared an invaluable rule for product marketing.

00:17:03: in this accelerated era, your launch tiers must track commercial leverage not engineering effort.

00:17:09: Ah, that is a trap so many technical founders fall into.

00:17:12: It happens constantly.

00:17:14: the team spends hundreds of engineering hours untangling complex back-end plumbing or like migrating a massive database and because it was incredibly difficult to build they demand at tier one high noise marketing launch.

00:17:26: look how hard we worked

00:17:27: exactly.

00:17:28: but if That backend migration doesn't move The needle commercially If the buyer doesn't explicitly care about it?

00:17:36: market windows, like enterprise budget seasons or compliance deadlines.

00:17:39: Those are immovable right?

00:17:41: Your market window has to dictate your code freeze.

00:17:43: the code freeze cannot dictate The Market Window.

00:17:46: Yeah you have to deeply understand the buyer's actual operational reality And speaking of which, there is an incredibly entertaining story from Guillermo Flor that perfectly encapsulates this idea of finding the real customer problem versus just falling in love with technology.

00:18:02: Oh

00:18:03: I know what you mean!

00:18:03: Yeah

00:18:04: There was a viral post circulating recently about entrepreneur who supposedly bought walked into a local coffee shop, and charged the owner twenty-one hundred dollars to install an offline private AI agent plus an ongoing monthly retainer to maintain it.

00:18:21: And the internet just loves The Ultimate Tech Hustle story right now?

00:18:23: Oh

00:18:24: they ate up.

00:18:24: everyone was dissecting trying to figure out how to replicate business model.

00:18:28: but Guillermo cuts through this hype on this one.

00:18:30: he points that pitching offline large language model to a local cafe is almost certainly fake or at the very least, terrible business.

00:18:40: I mean, a coffee shop has zero need for an air-gapped LLM.

00:18:44: Zero?

00:18:45: But if you extract the underlying mechanism of that hustle and point it out in the correct market segment... The strategic insight…is actually brilliant!

00:18:53: Yes.

00:18:53: Guillermo points out that the real target market for an offline Mac mini agent isn't a coffee shops—it's regulated small businesses.

00:19:01: Think about like a local law firm managing M&A documents, an accounting firm or private medical practice.

00:19:08: These businesses desperately want the efficiency of AI document assistance and intake tools but they literally legally cannot send unredacted client data to open AIs cloud servers.

00:19:20: it violates attorney-client privilege IAPIA data residency laws.

00:19:24: They have massive compliance anxiety And they do not have dedicated IT departments.

00:19:29: because The entire SAAS industry has moved cloud over the last ten years, there are almost no modern software products serving these highly regulated on-premise needs anymore.

00:19:38: Exactly!

00:19:39: So The Real Business Model isn't selling a flashy AI wrapper...the real value is being the local trusted technician who shows up plugs in physical hardware configures a localized agent that never touches internet and charges premium retainer to guarantee compliance.

00:19:53: It's the ultimate proof that mastering technology is secondary understanding customers actual constraints which echoes a recent warning from Ferrari's CEO.

00:20:03: If you present cutting-edge piece of technology to customers and their response is, so what?

00:20:08: What does this actually do for

00:20:10: me?"

00:20:10: You cannot answer them clearly – the product is wrong!

00:20:12: Yes…

00:20:13: The Technology Is Merely A Lever.

00:20:16: The Business Problem Is The Fulcrum.

00:20:19: Let us bring it all together.

00:20:20: We have covered a massive amount of ground in this session.

00:20:23: The central current running through all these insights is that AI has permanently commoditized the active building software, it's cheap and unimaginably fast!

00:20:33: Which means for you listening to this your career modes are no longer tied into raw output or ability to write PRD quickly.

00:20:40: Your professional mode your ability to mathematically filter the noise, and you're deep nuanced understanding of your customer's actual operational reality-like knowing why a law firm needs an air gap to Mac Mini.

00:20:53: Mastering the mechanics of AI is necessary yes but it entirely secondary to mastering business fundamentals behind

00:21:00: Absolutely.

00:21:01: And if we accept this premise that software has entered an era of industrial scale production where features are just cheap and abundant, I want to leave you with one final thought to mull over...

00:21:12: All right lay it on us!

00:21:14: We have spent the entire deep dive discussing how AI accelerates what human users and human builders do but we're rapidly approaching a threshold.

00:21:22: were humans might not be in loop at all Wait..

00:21:25: Where does that lead the product?

00:21:27: Exactly If we read to point where autonomous AI agents become the primary users of saw's products, meaning an agent is the entity navigating the UI making decisions executing workflows and pulling data via API.

00:21:41: What does a software product actually become if no human being ever opens your beautifully designed reporting dashboard again?

00:21:49: what Is Your Products actual value?

00:21:50: Wow

00:21:50: The thousand horsepower engine is running at top speed but there's literally No Human in the driver seat.

00:21:55: That is a wild paradigm shift to consider.

00:21:57: Thank you for joining us on this deep dive!

00:21:59: Keep questioning the consensus and keep filtering.