Best of LinkedIn: Digital Products & Services CW 40/ 41

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

00:00:10: Frenness is a BDB market research company that supports enterprise product teams with building feature by future competitive intelligence That shows exactly how their product stacks up against the competition.

00:00:21: You can find more info into description

00:00:23: And you know we are so glad your joining us for this Today were doing deep dive to the absolute top Digital products and services trends that are just circulating all over LinkedIn right now.

00:00:34: Yeah, we've got a really great stack of insights to get through.

00:00:37: We're gonna be focusing on how AI is well radically changing How products are actually built?

00:00:42: Right

00:00:43: And will get into why strategy not engineering is officially the true bottleneck Now exactly

00:00:49: Plus we're going talk about how the traditional SaaS market Is facing this really existential threat from users just building their own tools, this whole DIY vibe coding thing.

00:01:00: Oh that part is wild!

00:01:02: I can't wait to unpack that for you.

00:01:03: But let's start with what feels like the most glaring contradiction in tech industry today

00:01:08: The shipping paradox.

00:01:10: Yeah...the shipping paradox.

00:01:12: Look we've literally never been faster at building and shipping software right?

00:01:15: Totally.

00:01:16: but people actually using these products as customers aren't feeling any impact of speed.

00:01:23: Akash Gupta and Antonia Landy actually highlighted this massive finding from an Atlassian survey recently.

00:01:28: Right, that Atlassians data?

00:01:29: Yeah it was

00:01:29: eye-opening.

00:01:30: That really was

00:01:32: So.

00:01:32: eighty percent of product teams say they are shipping code faster now thanks to AI

00:01:36: Which sounds amazing.

00:01:38: On paper.

00:01:38: yeah But the survey shows customers aren't seeing value any sooner.

00:01:43: It's like Well its'like.

00:01:44: we've given everyone a much faster car but nobody bothered.

00:01:47: check if we Actually have

00:01:48: map.

00:01:49: That is exactly issue.

00:01:52: The underlying mechanism there is all about where the bottleneck moved, right?

00:01:57: Because I mean the coding got faster Generating.

00:02:00: The documentation got faster but the human decision-making process that didn't speed up one bit.

00:02:06: now You can really automate the deciding

00:02:08: part exactly.

00:02:09: the same survey noted That sixty nine percent of people say decision making Is just as slow as it was before generative AI.

00:02:19: Wow and we're at a point where AI adoption virtually universal, like ninety-nine percent say it's fully integrated into their workflows.

00:02:27: But because these tools blur the lines between disciplines so much, sixty seven percent of people can no longer clearly define their own job

00:02:35: responsibilities.".

00:02:35: That

00:02:35: is wild!

00:02:36: So you have designers doing front end engineering work and engineers are suddenly drafting product specs?

00:02:42: It just turns in to a disorganized free for all if you aren't careful.

00:02:45: It really does... And Richard Alexander Bautista proposed a really critical rule for this exact scenario, he said.

00:02:52: teams have to stop measuring AI features and they need to start measuring workflow changes.

00:02:57: Because the feature isn't an outcome?

00:02:58: Exactly!

00:02:59: Just slapping in AI-powered label on textbox that doesn't mean you've solved anything

00:03:06: Right.

00:03:06: Like, if a customer support team takes twenty minutes to handle the standard ticket and you build an AI draft generator for them...

00:03:13: You think you've innovated?

00:03:14: Yeah!

00:03:15: ...you're a hero.

00:03:15: but If the agent still has to spend eighteen minutes reviewing fact-checking And rewriting that draft because it hallucinated policy details....

00:03:23: You haven't changed at work flow at all

00:03:25: No..You just hit an arbitrary engineering milestone.

00:03:29: It only becomes actual business outcome if entire ticket process drops from twenty minutes down to say seven minutes.

00:03:35: And when you focus on those workflow changes, it forces a complete restructuring of how product teams operate.

00:03:42: Like Markiewicz-Giedus brought up this structural shift happening at Coinbase...

00:03:45: Oh yeah the pods!

00:03:46: Right they reorganized around what they call AI native pods and some of these pods are literally just one human being

00:03:53: One person?

00:03:54: One person acting as the central orchestrator.

00:03:57: They handle the engineering The design The product management All by using various AI agents to execute the

00:04:02: tasks.

00:04:04: I mean Having one person orchestrate the entire life cycle sounds incredibly efficient on the surface.

00:04:10: But

00:04:11: but I have to ask if one person can now do the job of five using AI Doesn't that inherently make the product really fragile?

00:04:20: How so well

00:04:22: a solitary builder might lack The scar tissue right like the past industry experience.

00:04:27: To know why is specific compliance field needs to be on a checkout screen.

00:04:32: aren't we Just risking building really polished mistakes at high speed?

00:04:36: That is the core vulnerability, hands down.

00:04:38: And Gaidus actually issued a warning about this exact dynamic.

00:04:41: He did Yeah he noted that AI dramatically compresses exploration speed.

00:04:46: You know you can ask an agent to generate ten different versions of a UI in seconds But AI does not compress judgment.

00:04:54: Ah The model doesn't know about the regulatory audit.

00:04:56: your company failed three years ago, which is literally the only reason that specific annoying compliance field exists.

00:05:03: It's the difference between doing something fast and doing the right thing?

00:05:08: Exactly!

00:05:09: Adi Thacker framed this beautifully using a mathematical concept.

00:05:13: he said product management is a vector not a scalar.

00:05:16: Okay unpacked it

00:05:17: So A Scalar Is Just A Measurement Of Magnitude Right Raw Speed But a vector is magnitude plus direction.

00:05:24: Oh, I like that!

00:05:25: Yeah Patrick Mack echoed this too.

00:05:27: he pointed out because building isn't incredibly cheap.

00:05:30: now deciding what to build has become incredibly expensive.

00:05:33: Right Because marching in the wrong direction at light speed just means you're wasting resources faster than ever.

00:05:39: Exactly

00:05:41: So look if Direction and Judgment are our real bottlenecks.

00:05:45: Now then figuring Out What To Build Like The Whole Product Discovery Phase.

00:05:51: That has to be the most critical part of operation.

00:05:53: Unquestionably!

00:05:54: But ironically, teams are trying to automate that too?

00:05:57: Oh yeah.

00:05:58: Aiden Zayapur shared a stat from a listener survey that really highlights how far this is gone.

00:06:03: Currently, ninety-four percent of product pros use AI in their research workflows.

00:06:07: Okay...that makes sense.

00:06:08: but The alarming part Twenty percent of them actually rely on AI overdoing their own original research.

00:06:15: Wait

00:06:15: Really?!

00:06:15: Yes They're having models summarize old data or simulate user personas instead you know, actually talking to human beings.

00:06:24: So they're just feeding like a hundred old user interview transcripts into an LLM and asking for the top five themes?

00:06:31: Essentially yeah!

00:06:32: And the danger there is how these large language models are mathematically designed.

00:06:37: They are fundamentally probability engines.

00:06:40: Their whole function is predict most likely average answer based on patterns.

00:06:45: so by design smooth over contradictions wash out anomalous edge cases.

00:06:50: But those edge cases are where the gold is.

00:06:53: I mean, if one user out of a hundred's using your software in completely weird unorthodox way that anomaly is usually when big innovation is hiding

00:07:02: Exactly Which perfectly aligns with what Christian Idiotti and Teresa Torres were arguing.

00:07:07: What do they say?

00:07:09: They noted that.

00:07:10: look AI is fantastic for preparing for interviews or synthesizing raw notes afterward.

00:07:16: Sure

00:07:16: admin stuff,

00:07:17: but if you let it replace the actual customer conversation You aren't doing better discovery.

00:07:22: You are just guessing faster.

00:07:25: guessing faster That's a great line

00:07:26: It is.

00:07:27: they point out that AI raises the floor for novices.

00:07:31: So a beginner can write a decent interview script now But it raises the ceiling for experts Because the true advantage isn't having a tool.

00:07:40: It's having judgment to know what good looks like and knowing when model is missing nuance.

00:07:45: That makes total sense.

00:07:47: And Ed Biden actually suggested really practical framework for managing this balance.

00:07:51: The discovery budget.

00:07:53: Yeah,

00:07:53: treating product discovery like financial budget sized entirely by risk.

00:07:58: Right I love that approach.

00:08:00: How does he structure it?

00:08:01: You allocate your research effort based on consequence of being wrong.

00:08:05: So, if you're just adding a new date filter to the search page.

00:08:09: Low risk

00:08:10: Super low-risk The risk of failure is basically zero so it gets tiny discovery budget.

00:08:15: maybe few days of cheap validation like running fake door test.

00:08:19: Where are we put up button for feature that doesn't exist yet?

00:08:22: Just see how many users try.

00:08:24: click it

00:08:24: Exactly.

00:08:25: See there's intent before build it.

00:08:27: Low risk and low budget But proposing an entirely freemium tier That changes whole revenue model

00:08:34: Oh, that's huge.

00:08:35: Right?

00:08:35: The risk of getting that wrong could literally be fatal to the business.

00:08:39: so... That requires a massive budget!

00:08:41: Weeks of deep qualitative human-led discovery.

00:08:45: That makes perfect sense.

00:08:46: And by the way if you are enjoying how we impact these frameworks make sure you subscribe So you catch our future deep dives.

00:08:52: We do this every two weeks.

00:08:53: Definitely subscribe

00:08:54: But okay let me play devil's advocate on this budget framework for a second.

00:08:58: Go

00:08:58: for it.

00:08:58: If I strictly budget my discovery based on what already feels risky Aren't I just guaranteeing that i'll only ever find safe incremental improvements?

00:09:08: Like, how do you ever stumble into a massive unexpected aha moment if You Only aggressively research the things.

00:09:15: You already know are dangerous?

00:09:16: That is A really sharp question and it points to a much deeper issue with How companies manage ideas once they're actually generated.

00:09:24: The

00:09:24: roadmap problem

00:09:25: exactly.

00:09:26: Steven Grenise and Helene Amster-Cople both touched on the systemic failure of roadmaps.

00:09:32: Grenise noted that roadmps generally fail through a process of addition without subtraction.

00:09:37: Oh, man!

00:09:38: The backlog only ever grows and never shrinks.

00:09:41: we've all been there

00:09:42: We ALL have.

00:09:43: An executive request comes down BAM Added to the queue A competitor launches something flashy BAM A reactionary feature gets added

00:09:51: And nothing get's removed.

00:09:52: Right...and six months later The engineering team has twelve different initiatives in progress, all sitting at eighty percent completion delivering absolutely zero value to the end user.

00:10:02: It's exhausting!

00:10:04: And Cople adds a really harsh truth for founders in this situation... They often assume they need to hire a product manager to organize that chaotic backlog

00:10:13: but they don't

00:10:14: know what they actually need.

00:10:15: as a Product Strategist who has the authority to look them in the eye and tell then their pet ideas simply isn't good?

00:10:22: Wow yeah Because the most expensive mistake a company can make isn't building the wrong feature.

00:10:28: It's building the right product with peak

00:10:30: efficiency.".

00:10:31: That

00:10:31: is so true, and while Discovery supposed to dictate what we build AI making how?

00:10:37: So radically accessible that traditional SaaS companies are suddenly waking up to find they're competing against their own customers.

00:10:43: The DIY software trend.

00:10:45: Yeah

00:10:45: this is crazy.

00:10:46: going back to that story from Ashish Patel his team was hit with a five X price hike from their time-tracking software vendor.

00:10:54: A five hundred percent increase!

00:10:56: Insane, right?

00:10:58: So instead of paying it or enduring this brutal migration process to a competitor they just built it themselves.

00:11:04: over

00:11:04: a single weekend... ...a weekend

00:11:07: They fed the AI that previous vendors API specs had generate the front end components to clone the UI and Just hooked up there own database.

00:11:15: It's incredible.

00:11:17: After a week of minor bug fixing, they processed an entire month's worth of invoicing on their own homegrown tool without a single issue.

00:11:23: And what is really fascinating in the catalyst driving this behavior?

00:11:27: Todd Orchard and Dan London pointed out that SaaS vendors are frequently there own worst enemies here.

00:11:32: How so

00:11:33: They push users away with restrictive API rate limits or pad free tiers for useless vanity features while locking core necessary functionality behind massive enterprise paywalls.

00:11:45: Yeah, the payable fatigue is real.

00:11:47: It is!

00:11:48: Users are exhausted from constantly adapting their daily workflows to fit the rigid constraints of a vendor's product.

00:11:54: so they turn to models like Claude and start vibe coding their own bespoke tools.

00:11:59: Okay I have to ask about this Vibe Coding.

00:12:03: Isn't it just the twenty-twenty six version of The Office Excel Macro?

00:12:07: What do you mean?

00:12:09: Well

00:12:09: You know, we all remember Kevin in accounting who built a brilliant automated spreadsheet and an afternoon saved everyone hours of work.

00:12:15: Oh yeah

00:12:16: But three years later kevin leaves the company The macro breaks And the entire accounting department is paralyzed because they're stuck with an unmaintainable mess.

00:12:24: Isn't weekend AIF just more complicated excel macro?

00:12:28: It's a very fair comparison, but it misses a crucial evolution in the technology.

00:12:33: The fundamental difference between Kevin's macro and todays' agentic systems is that modern AI can be prompted to entirely self-documenting and self testing.

00:12:43: When Kevin left his logic was mystery.

00:12:46: Today an AIA agent could read the custom application write comprehensive documentation explaining its own architecture, generate testing environments and seamlessly explain the logic to the next Kevin who takes over.

00:12:59: Okay so the knowledge transfer is automated too?

00:13:01: Exactly though you are absolutely right that the core organizational challenge like deciding who actually owns and maintains this shadow IT when it inevitably needs a security update... That remains unsolved!

00:13:16: Right And Steve Brayden actually connected this exact dynamic to the plummeting valuations we're seeing across the SaaS market.

00:13:22: Yes, The technical debt issue?

00:13:25: Yeah if you look at a mature legacy SaaS product it is usually just bogged down by A decade of technical debt Abandoned integrations Labyrinthian code paths.

00:13:36: It becomes so bloated

00:13:37: and buyers Look at a platform that takes a year To ship a meaningful update?

00:13:42: They refuse to pay premium for especially when they know an AI agent can replicate the tiny, highly efficient core of that software and just ignore all the legacy bloat.

00:13:52: Vic Gumbier highlighted the massive cognitive dissonance happening in the market regarding this too.

00:13:57: but what did he say?

00:13:59: He noted there are business leaders who fully believe AI is capable of automating entire sales department right

00:14:05: sure prospecting outreach booking meetings

00:14:07: exactly all autonomously.

00:14:09: Yet, those same leaders somehow doubt that AI is capable of writing the code for the SaaS product... ...that the sales team was selling.

00:14:17: That's a brilliant observation!

00:14:18: Why The Blind Spot?

00:14:20: It just lags in perception.

00:14:22: Meanwhile, the open-source community has already filling this gap.

00:14:25: So Hedthiab recently launched an Open Source tool called Product Mode….

00:14:29: …that hit over two hundred stars on GitHub almost instantly.

00:14:32: What does it do?

00:14:33: It literally embeds product management judgment directly into the coding agent.

00:14:38: Wait how

00:14:39: Before the AI is allowed to write a single line of code, this system forces it to challenge users' assumptions.

00:14:45: It actively cuts unnecessary scope and defines clear success metrics.

00:14:48: It's trying to automate judgment not just typing.

00:14:51: Wow!

00:14:52: Automating pushback?

00:14:54: That really smart!

00:14:56: Which brings us into the ultimate question for enterprise teams If anyone can spin up functional app in weekend.

00:15:03: The underlying tech is no longer a defensible moat.

00:15:06: So how do companies win in this new environment.

00:15:10: It seems like it all comes down to flawless execution, rapid activation and aggressively removing friction.

00:15:17: Absolutely!

00:15:18: Jalila Solomon highlighted a perfect example of this with the recent update on the Dashin Super App.

00:15:23: Oh

00:15:24: I saw that.

00:15:24: Yeah

00:15:25: instead forcing users through multi-step authentication process just make payment.

00:15:29: The app was updated so user can simply open it scan merchant code and pay immediately without even logging first.

00:15:36: That is smooth

00:15:37: It is.

00:15:38: Can we make the user do less, as a new competitive advantage?

00:15:42: Because failure points in software are rarely these massive dramatic crashes anymore.

00:15:48: As Atiku O'Rachaman noted User activation usually fails into tiny silent moments of friction.

00:15:53: Yes Like a dashboard that loads completely blank for a new user

00:15:57: Or mandatory setup form that asks three pieces of info instead just two

00:16:01: Exactly.

00:16:03: And Gouzi Chukua, Julie Uso-Chuku built on this.

00:16:07: She warned teams that their internal metrics are often deceiving them.

00:16:10: How so?

00:16:11: Well marketing teams will celebrate a sky high click through rate on new campaign.

00:16:15: right sure but if the product team isn't tracking the full conversion journey all the way down to actual meaningful account activation then Click means absolutely nothing.

00:16:26: Capturing attention is useless If The Product Fiction prevents you from monetizing it

00:16:30: Precisely.

00:16:31: And monetizing that attention requires a really deep understanding of user behavior.

00:16:36: Bushret Choskunar analyzed how Doodle one-to-one mapped their pricing models to the classic pirate metrics.

00:16:43: Remind me, That's the AAARR framework right?

00:16:46: Right!

00:16:47: Acquisition activation retention referral and revenue Got it.

00:16:51: So How does that apply to pricing?

00:16:53: Well When you offer a freemium model, You are fundamentally delaying the revenue stage because you aren't charging up front.

00:17:00: You put massive intense pressure specifically on retention.

00:17:04: Oh that makes sense.

00:17:05: Because they have to keep coming back

00:17:06: Exactly.

00:17:07: The product has to consistently prove its value drawing them back in day after day until They finally hit usage wall or realize they need premium features enough To pull out of credit card.

00:17:17: If your product doesn't naturally drive high retention A freemia model will bankrupt you.

00:17:22: Wow And getting users excited and engaged from day one is critical for that retention.

00:17:28: A tier Abdul Rauf analyzed over two hundred product launch posts on LinkedIn to identify what actually drives initial momentum,

00:17:35: What did the data show?

00:17:37: Two very clear practical rules.

00:17:39: first

00:17:40: give your new product or feature a specific name.

00:17:43: interesting yeah.

00:17:44: vaguely announcing an update performs terribly.

00:17:49: Naming it earns one point nine times more reposts because people want a concrete noun they can talk about and share.

00:17:55: Right, makes the tangible

00:17:56: exactly.

00:17:57: Second rule if you use of video The first three seconds dictate everything.

00:18:03: You have to skip the corporate throat clearing in this low logo fades And just immediately show the product solving problem.

00:18:10: That's great tactical advice.

00:18:12: But, you know as teams hyper focus on these micro interactions and distribute their work across smaller AI pods a major operational vulnerability emerges.

00:18:22: What's that?

00:18:23: Matthew Small pointed out that in the transition away from monolithic structures to this highly distributed autonomous teams organizations have lost horizontal visibility.

00:18:33: Ah!

00:18:33: The cross-team connection ensuring the entire product actually makes sense.

00:18:40: Small proposes that organizations need to introduce a service owner role, To sit directly alongside the traditional product owner.

00:18:48: Okay how do they differ?

00:18:50: While the Product Owner is focused on creating new value through localized changes The Service Owner is responsible for protecting That Value across the entire end-to-end customer journey.

00:18:59: They connect the dots when a user's workflow breaks because it has to cross Through six different AI pods

00:19:05: And Ravi Metta approached this from an org design perspective too right.

00:19:09: Yes He noted that while the formal title of product manager might disappear in some these highly automated companies, The actual work-of-product management remains absolutely mandatory.

00:19:20: The tasks don't go away?

00:19:21: Exactly understanding the customer's needs Setting a cohesive strategy connecting technical outputs to business outcomes.

00:19:28: You can distribute those tasks among different people but you cannot eliminate them.

00:19:32: But wait let me ask you something.

00:19:34: If we are fragmenting the PM role into these highly autonomous one-person AI pods to gain speed.

00:19:40: But then, we suddenly realize things are disjointed.

00:19:43: so we introduce service owners to sit above them and stitch the journey back together.

00:19:48: Haven't we just reinvented corporate middle management bureaucracy under a shiny new name?

00:19:53: You know it is classic pendulum swing of organizational design.

00:19:56: It really is.

00:19:57: We rapidly decentralized everything because wanted the sheer speed that AI agents offer.

00:20:02: Now the friction of a disjointed user experience is forcing us to realize we need to re-centralize slightly To maintain coherence.

00:20:09: Yeah, We have to find the middle ground.

00:20:11: The companies that ultimately dominate this next era software are gonna be the ones That figure out how to perfectly balance the raw Scalable speed of agentic systems with the cohesive contextual judgment Of human strategy.

00:20:24: Beautifully said you've got the fast cars now?

00:20:26: We just have to figure out How to draw them up

00:20:29: exactly.

00:20:30: well thanks for joining Us on This deep dive.

00:20:32: If you enjoy this episode, new episodes drop every two weeks.

00:20:35: Also check out our other editions on Cloud Insights and Sovereignty, ICT & Tech Insights, AI in Agents Systems, Sustainability in Green ICT, DefenseTech and HealthTech.

00:20:47: And we will leave with one final thought to mull over drawing some provocative ideas shared by Venkatesh Krishnan and Cameron Tonkinways.

00:20:55: Look, if AI systems can eventually synthesize the collective history of a product's entire life cycle cataloging every past decision Every abandoned prototype.

00:21:05: Every regulatory mistake and every strategic pivot into queryable wisdom Are we moving toward a future where accompanies most valuable asset isn't its source code or even it's active user base but its institutional memory?

00:21:20: Because everyone has access to the fastest car in the world The only thing that truly gives you an advantage is remembering exactly where you've already crashed.