Best of LinkedIn: Digital Products & Services CW 10/ 11

00:00:00: This episode is provided by Thomas Allgaier and Frennus, based on the most relevant LinkedIn posts about digital products and services in calendar weeks ten and eleven.

00:00:09: Frenness 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.

00:00:19: Right, so welcome to our deep dive.

00:00:21: Today we're digging into the top digital products and services trends that have been popping up all over LinkedIn for the past two weeks.

00:00:28: Yeah And For you listening whether your in the middle of sprint planning or trying to rework your whole product org The mission here is really about practical adoption.

00:00:37: Exactly I mean the conversation has decisively shifted.

00:00:40: We are moving way.

00:00:41: pass that initial AI experimentation height.

00:00:44: Oh one hundred percent.

00:00:45: yeah The magic trick phase is over right?

00:00:48: operating discipline now.

00:00:50: Like, how is AI actually rewiring daily workflows?

00:00:54: And honestly even the definition of a product manager...

00:00:57: It's changing everything!

00:00:58: ...it really is.

00:00:59: I saw this great example from Nico Noel that just nailed this.

00:01:02: he used an AI agent Claude Code to actually ship changes to a landing page.

00:01:09: Okay so actually writing in code Yeah

00:01:11: but he didn't stop there or you know hand it off to a dev.

00:01:14: He stayed right in the interface used productanalyst.ai and immediately wired up a Slack alert for funnel monitoring.

00:01:22: He did the coding, tracking or automated reporting all in like ten minutes.

00:01:26: Ten minutes?

00:01:27: I mean that kind of turnaround is just staggering.

00:01:29: but you know it's also massive trap.

00:01:32: How so?

00:01:32: Well,

00:01:33: Akash Gupta broke this down brilliantly.

00:01:35: When building becomes that frictionless the danger of unguided-building just skyrockets He pointed out.

00:01:41: we're seeing two distinct types of PMs emerge right now.

00:01:45: Okay what are they?

00:01:46: So the first group he calls The Builders.

00:01:48: These are ones who spin up a quick prototype show it off and get green light.

00:01:52: because you know working software always looks better than slide deck.

00:01:55: Right.

00:01:55: but their moving fast probably aren't checking underlying metrics

00:01:59: Exactly!

00:01:59: They don't check metrics or crucially, API costs.

00:02:04: Generative AI isn't a flat rate server cost.

00:02:06: you're paying compute per token.

00:02:08: Oh, right.

00:02:09: So every unoptimized search just drains the budget.

00:02:13: but then Gupta says that second group is The Evaluators.

00:02:16: I'm guessing they operate a bit differently Completely

00:02:19: different and evaluator will prototype something like fifty times And deliberately kill eighty percent of those iterations.

00:02:26: Ah so they filter it.

00:02:27: Yeah

00:02:27: gupta calls having taste at speed It's ability to refine through rapid iteration.

00:02:33: That makes total sense.

00:02:34: But yeah i have push back here.

00:02:36: Go for It's just there.

00:02:37: is this really seductive trend right now called vibe coding, where a PM can basically generate an app in the afternoon.

00:02:44: Oh yeah!

00:02:45: Vibecoding is everywhere

00:02:46: Right.

00:02:46: So isn't that necessary first step for raptid discovery?

00:02:49: I mean why shouldn't a PM own the whole stack if AI allows it?

00:02:54: Its like giving everyone on team a nail gun when they only know how to use a hammer.

00:02:58: Ok thats great analogy.

00:02:59: Sure you build your house way faster but Without structural knowledge, the whole thing might just collapse in a breeze.

00:03:06: But still...the speed!

00:03:08: The Speed is exactly why Tanya Maldonado and Marlon Davis were raising massive red flags about this

00:03:14: Really?

00:03:15: What did they say?

00:03:16: They pointed out that PMs absolutely should not be replacing developers on enterprise production grade systems.

00:03:23: Scalability and security are not, you know side quests.

00:03:26: Yeah

00:03:27: if your building fifty times faster with your nail gun You can ship disasters fifty time's faster Which means launch readiness has to become the new bottleneck.

00:03:35: Launch Readiness Okay.

00:03:37: so how do we actually manage without losing speed?

00:03:40: Because Shardle Nyak actually shared a twenty-three item AI product launch checklist on this.

00:03:46: Twenty three items?

00:03:47: That's rigorous!

00:03:48: Very, and he highlighted risks that traditional saws just doesn't face like prompt injection defenses.

00:03:54: Oh,

00:03:54: prompt injection is terrifying.

00:03:56: users can just trick the system into ignoring its own security guardrails

00:04:00: Exactly.

00:04:01: And he also mandates strict budget alerts, like setting a warning at eighty percent of budget but literally killing the feature if it hits two hundred percent.

00:04:10: Because those runaway API costs we mentioned

00:04:12: right?

00:04:12: He says gradual rollouts are absolute necessities.

00:04:14: now you go five percent to twenty-five to fifty to one hundred launching to a hundred percent on day One is just asking to lose user trust.

00:04:23: one hundred percent agree And it's a systemic risk too.

00:04:26: Dr Bart Jaworski had this fascinating advice about setting up.

00:04:34: Its only job is strictly to rate and evaluate the answers of the main AI product.

00:04:39: Oh, wow!

00:04:39: So it's an AI babysitting your

00:04:41: A.I.,

00:04:41: to block bad outputs?

00:04:42: Exactly...and it ties right into what Patricia Bertini was saying about compliance

00:04:46: Like The AI Act in GDPR.

00:04:48: Yes,

00:04:48: Ann Diara She pointed out that Compliance Is Shifting Entirely Upstream.

00:04:52: You cannot wait until the end To check these things anymore.

00:04:55: you don't want Find Out Your Non-compliant Via Massive Fine Or An Audit.

00:04:59: Okay But Wait If We Are Slowing Down To Check a twenty-three item list, and running secondary LLM evaluations.

00:05:08: Doesn't that completely neutralize the incredible speed AI gave us in first place?

00:05:12: It

00:05:13: feels like it right!

00:05:14: Yeah

00:05:14: I mean, how do we balance safety with the intense pressure from leadership to ship?

00:05:19: Well.

00:05:20: The hard truth is you can't balance it without changing the underlying structure of the organization.

00:05:25: dreaming You can just add AI to a broken system and expect it to work.

00:05:29: You have to fix the product operating

00:05:32: model POM.

00:05:33: people talk about that all the time but It's usually so abstract

00:05:36: It is, but Ralph Joachim brought some brutal reality to it with the time-poverty paradox.

00:05:41: Oh

00:05:42: I saw those stats.

00:05:42: they were wild!

00:05:43: Yeah, ninety two percent of product leaders own revenue outcomes But they spend sixty six percent of their week on manual coordination work

00:05:51: Which leaves them zero time for actual strategic analysis.

00:05:54: Exactly

00:05:55: Zero Time.

00:05:55: That's insane.

00:05:56: its like being hired as a ships navigator... ...but you're forced to spent ninety percent your time below deck shoveling coal into engine.

00:06:03: How are ya supposed steer?

00:06:04: You aren't!

00:06:05: And that's why Catherine Shepard King argues a real product operating model isn't just renaming your project managers.

00:06:12: Or copying Spotify's squad template from ten years ago?

00:06:15: Right, A Real POM is about the mechanics of how work actually gets funded and decisions are made.

00:06:22: How do we fix those mechanics?

00:06:24: Igor Voth shared a really battle-tested five layer model that I

00:06:28: loved.

00:06:28: Okay, bring it down for me!

00:06:29: It goes company priority which leads to root cause then to strategic bet...down to opportunities and finally experiments.

00:06:38: So every experiment has to tie all the way back up to a company priority?

00:06:42: Exactly if doesn't trace back to a priority you are just shipping features.

00:06:46: You aren't driving outcomes.

00:06:47: Ok.

00:06:48: so If we fix the operating model and we actually give our leaders the time to navigate from the deck of a ship, they still need to be looking at the right maps.

00:06:56: Right?

00:06:56: Absolutely!

00:06:57: Which leads us into the evolution of product intelligence.

00:07:00: Because Mauricio Cardin has pointed out that this problem isn't lack-of-beta

00:07:06: Oh...we have too much data.

00:07:07: Exactly The problem is interpretation.

00:07:10: He says AI shouldn't just use to build chat features.

00:07:14: It should an intelligence layer That connects fragmented telemetry And user feedback

00:07:19: to spot the patterns that humans just miss.

00:07:21: Yes, and Nigs or Voodies highlighted how Netflix does this by treating data as a product

00:07:26: meaning it has clear ownership in life cycle management.

00:07:29: right but there's a massive warning sign here from boost Raku skooner about optimizing for the wrong metrics.

00:07:36: oh this is such a common trap.

00:07:37: what were her examples?

00:07:38: so one was a mahjong app.

00:07:40: they made the game easier to boost retention And sure it did temporarily, but it completely killed the challenge of game which was actual value.

00:07:47: Oh!

00:07:48: So they alienated their core users?

00:07:50: Completely.

00:07:51: and her other example is a puzzle game called Screwdom.

00:07:53: They basically forced monetization and just lost user immediately.

00:07:57: Optimizing dashboard metric without understanding user values destroys product.

00:08:02: It really does.

00:08:03: But

00:08:03: let's be real for second.

00:08:04: If your PM and retention metrics revenue graphs are pointing up into right aren't we doing our jobs?

00:08:11: It looks like it on paper.

00:08:12: Right,

00:08:13: so how does a PM justify ignoring a green dashboard to tell the CFO trust me?

00:08:19: The users actually hate it.

00:08:21: Well

00:08:21: that right there is the definition of true product discipline...it means looking beyond that immediate dopamine hit of vanity metric.

00:08:29: If you spike revenue this month but churn triples next month..the product is dying.

00:08:34: Yeah!

00:08:35: That makes sense.

00:08:35: its about sustainable outcomes

00:08:37: Exactly.

00:08:39: Well, as we wrap up the steep dive We want to leave you with a final thought inspired by Pavel Fabrikantov.

00:08:43: Oh

00:08:44: I love his take on this.

00:08:45: Yeah he reminded us that buying AI tools and writing strategy docs is The easy part.

00:08:50: anybody can do That.

00:08:51: but true competitive advantage comes from Process knowledge

00:08:55: process.

00:08:56: Knowledge yes

00:08:57: that tacit know-how That only accumulates when a team ships fails and adjusts together over time.

00:09:03: So ask yourself are you just lawyering your roadmap with new tools or?

00:09:06: Are you actually engineering the deep process knowledge, but your competitors simply can't copy?

00:09:12: that is The real question to ask your team this week.

00:09:14: it really

00:09:14: does

00:09:15: well.

00:09:15: if you enjoyed This episode new episodes drop every two weeks.

00:09:18: also check out our other editions on ICT in tech artificial intelligence cloud sustainability in green ICT, defense tech and health tech.