#productmanagement — Public Fediverse posts

Pondering invisible bottlenecks and delays

Looking for a job is different to working in an organisations with many teams.

For one thing looking for a job is more boring than having a job. For another thing, my side of the job matching is visible, while the “hiring person” side is basically invisible.

I can see if I look at a role and then postpone applying and even though a day might slide by, I can see that the bottle next was me. I cannot see whether someone read my resume or discussed it internally or much of anything else.

This impacted my “quick cycle times lead to quick learning” because I can fire off a resume and hear nothing back, or I can hear back a couple of weeks later. So I improve my online profile and the information I submit for jobs with no feedback from real customers.

That is the opposite of continuous discovery and good product management but is also a factor of the constraints of my current market. Of course calling it a market when I have one product (me or the services I can offer) and a few potential hirers is probably a stretch, but the principles of product management apply.

I could innovate with the product by offering to work as an interim executive, a volunteer, a part-timer, a contract or permanent. I could add features to my offering by doing a course in governance or learning to be a hit man. I could do a lot to tweak what is on offer and I should prioritise that.

Interim executive is a good idea, governance training will not likely improve my time to the next role but it would add value in the long run. Hit-man qualifications are probably not going to enhance the value of a delivery manager, unless I change my target market to “illegal mafia like organisations and international espionage”.

What I was thinking though – is that nearly everything on my side is visible to me and that is a real luxury compared to working as part of a system.

Even the invisible bits can be tackled. If I want to improve my resume I can fire it off to many organisations and hear nothing, or I could share it with friends who work in the industry and get real feedback from these “proxies for the customer”, because they, like I, have hired for similar roles that I would apply for.

But in a real job, you often wait for feedback, wait for permission, wait for work to get done and generally depend on others. In this situation you can get really busy, but the activity will not move things along if you are waiting for some input or decision.

So you get good at managing dependencies, stakeholders and change. It is a real skill that comes with experience but can almost be invisible.

Of course you can “get good” at managing dependencies by harassing people mercilessly, but that strategy is both less fun and less effective than it sounds.

On TV, leaders are often very demanding and they get all the resources they need because they demand them. They also tackle drama head on and keep moving at a pace that keeps the audience engaged.

In the real world, asking for people to swarm onto your dependency is less likely to succeed when there are many more dependencies for them to manage, with their own dependencies to manage and many more annoying stakeholders asking for urgent attention.

So a real skill is to know when focusing on something will increase the speed and quality and when it is better to let things take their course.

I guess that seems like a vague statement, but they say you don’t appreciate something until you miss it or lose it. In this case I think I don’t see the challenge until I am in a simpler environment with few dependencies and a product I build myself without needing much negotiation with others.

So sitting hear next to the trees in my backyard, I can operate at my own sustainable pace. But the real thing I realise is that in a complex working environment, not everything is visible and the lack of that visibility has a real impact.

It is quite easy to “see” a bottleneck and focus on it. But generally you are seeing the symptom of a bottleneck and not the real cause, or sometimes not even the real bottleneck.

For example, at my last company, code reviews took a long time. So doing faster code reviews with AI would seem like a good idea. But in fact it was not the time spent on the code review that was an issue, it was the scarcity of experienced people who understood the code.

So getting more people experienced with the code seems like a good idea and it probably is.

But there are a couple of problems with that. Firstly the desire to use a tool or process to fix things – “Use AI to make it faster” – can actually be a distraction. Getting people to pull experienced technical people into meetings to decide whether to make AI code reviews better makes them more scarce on the ground. Doing that might cause them to take longer to go to a code review and make it harder to find time to mentor others to increase the people who can review code.

Of course, using AI to improve code reviews is the way to go and things like telemetry and automated checking will improve both quality and time spent on manual reviews. But it won’t in itself make senior staff more available.

So what might make the crew more available is better delegation, better decisions rights or more focused discussions when the crew do get together.

But then, what if the problem is not just scarcity of gurus? And what if the improvement or decay in our cycle time is the result of a change we made a month or two ago.

Guru scarcity will increase if we have a lot of turn-over but should reduce over time if we have stable teams and good senior staff who spend time coaching others. Even having people fix bugs will add to their understanding and reduce the overhead of code reviews. So we could see improvements now that are the result of one good tech lead creating stability in the team because people enjoy learning from her.

That change occurred a while ago, then people realised and then their skill increased and then we saw the improvement. Similarly the use of AI in code reviews might initially not have a visible impact and then a few weeks later it might be having a massive impact, but people are not mentioning it because it is obvious.

Or the code reviews are taking a long time because in fact the engineer was making “dumb assumptions” because they did not understand the way the customer used the product and the guru sent them back to ask questions of the PO and then make changes. If this happens then the delay in code reviews happens twice every story (or code package). And it might be the result of poor information coming into the team, which is fixed by neither the use of AI or the increased availability of coding gurus.

So I guess the conclusion is that, while AI is revolutionising the way we build products, there are skills and attitudes that still matter a lot. And the more complex the environment the more they matter.

The problem is though, the same as it was a hundred years ago. People fix and improve what they see and when they see a change they often assume that they understand the cause, when they are only looking at a symptom.

I don’t think I can put “take time to notice things and has a knack for finding invisible stuff” on my resume.

But I can say that it is a requirement of many senior product and engineering roles. It is part of the joy of the work and also part of the invisible value that motivated and experienced people can add.

So if it is not a resume thing, what is it?

Some of it is boring process and habit management. Making things visible, being transparent and looking at the system are all ways that uncover previously invisible delays and bottlenecks.

Part of it is also curiousity. There is an Irish saying, or at least a saying I was told is Irish – “If you fall over, don’t look where you landed, look back at what caused you to trip”.

We could say that it is moving upstream or shifting left, but it is really as simple as remaining curious. Instead of acting on the symptom you see, you pause and ask how it came about.

More often than not, the slow moving tech lead who is not getting to code reviews is not actually a slow moving tech lead not bothering to look at code. It is a person in a system with many forces creating the existing situation.

And the better you get at finding where the hidden things are the better. It is often the time to make a decision that delays the implementation. It is often the distraction and not the work itself that is a delay. And it is often a lack of access to the right information or resources that causes the delay and the rework.

That kind of sounds obvious when you sit back next to a tree and ponder. But the art comes in paying attention and creating space to notice and ponder things when being pulled in 20 different decisions while multiple important things compete for attention.

`

Engineering is about people, not just tech. A great manager isn't just a boss; they’re an enabler who boosts team output more than extra headcount. Check out Ryan Murphy’s insights on identifying true leadership and avoiding "accidental" managers: https://newsletter.eng-leadership.com/p/how-to-tell-if-your-manager-is-actually #Leadership #ProductManagement

Why Most AI POCs Stall and How to Build an AI Operating Model That Works

https://bhavingandhi.com/2026/08/26/why-most-ai-pocs-stall-and-how-to-build-an-ai-operating-model-that-works/

What is a North Star Metric (NSM)? ⭐

It’s the key metric a company uses to measure the core value its product delivers to customers. A strong North Star Metric can help startup teams align priorities, track meaningful progress, and focus on sustainable growth.

Read the beginner-friendly guide: https://startupglossary.blogspot.com/2026/08/what-is-north-star-metric.html

#NorthStarMetric #Startup #ProductManagement #Growth #SaaS

Decision-making is the real bottleneck in product today. As building becomes "cheaper", choosing what to build is where we struggle. David Pereira explores how to categorise impact vs reversibility to pick the right method. Stop defaulting to slow consensus: https://dpereira.substack.com/p/the-art-of-simplifying-decisions #ProductManagement #Agile

Here’s the thing that’s really changed though: a PRD assumes you already know the hill you’re planning to climb. You’ve picked the target, and now you’re aligning everyone to storm it together.

Exploration is the exact opposite. You don’t know the target yet, you’re trying to find it, running several bets at once with a million ideas in your head. Write the full doc now and you’ve set in stone a decision you haven’t earned yet.

https://marieclairedean.substack.com/p/got-no-time-for-prd

#ProductManagement

How do micro‑cameras and computer vision hardware change the product roadmap for earbuds and wearables? This article explores the technical constraints, UX opportunities, privacy implications, and market potential as audio devices move toward vision-enabled AI. Read more: https://wix.to/RkXAvW6

#AI
#WearableTech
#ComputerVision
#HardwareDesign
#ProductManagement

Is your team building "reasoning silos"? As PMs use AI to synthesise data, we risk losing the "caveat debt"—that vital context only humans hold. We must ensure local AI experimentation becomes a shared organisational capability. Dive into the full analysis: https://www.productfocus.com/are-your-pms-building-reasoning-silos/ #ProductManagement #Agile

Why You Should Reach Out Before You Need Something

https://tpgblog.com/2026/08/24/why-you-should-reach-out-before-you-need-something/

B2B procurement is shifting. As Maersk shows, AI agents now negotiate deals at machine speed, rendering per-seat pricing and relationship-based sales obsolete. We must architect for machine-readable surfaces and outcome-based models to stay visible.

Read the full analysis: https://www.oreilly.com/radar/when-your-buyer-is-an-ai-agent/

#AI #ProductManagement

Leadership is a team sport, yet many managers distance themselves when things get tough. I believe psychological safety starts with being on the same side of the table as your team, especially during high-stakes meetings. Build trust by closing ranks: https://staysaasy.com/the-same-side-of-the-table/ #Leadership #ProductManagement

Preparing to talk to a security vendor? Cyber Company Profiles gives you an independent read of its market position and product strategy, built on the questions I kept asking as a buyer and product leader.

https://zeltser.com/cyber-company-profiles

#productmanagement #artificialintelligence

Teams are racking up "prompt debt" by fighting model weights instead of working with them. Drew Breunig’s framing of this as technical debt is spot on—relying on hacks locks you into old models and kills innovation. Stop "vibe shipping" and start building: https://www.oreilly.com/radar/prompt-debt-and-fighting-the-weights/ #AI #ProductManagement

As we delegate more to GenAI, we risk losing the critical thinking skills needed to spot its errors. Is speed worth the cost of deskilling? We must be deliberate about what we automate and what we keep sharp. Check out this guide on staying engaged: https://agilepainrelief.com/blog/skill-loss-and-gen-ai-make-it-a-choice-not-an-accident/ #ProductManagement #Agile

AI lets us perform tasks outside our expertise, but can we actually judge the quality of the output? Itamar Gilad explores the "Danger Zone" where AI-driven overconfidence meets the Dunning-Kruger effect. A must-read for product leaders: https://itamargilad.com/artificial-competence/ #ProductManagement #AI

Can LLMs really replace human users in A/B testing? Spotify's latest research suggests caution. While AI can predict outcomes, it often underestimates true impact. We must balance speed with statistical validity to avoid poor shipping decisions.

Read the full study: https://engineering.atspotify.com/2026/8/when-can-llms-replace-humans-in-a-b-tests

#ProductManagement #AI

Coding agents work for four reasons. Constrained environment. Fast unambiguous feedback. Cheap reversible mistakes. Judgment deferred to a human at review. Almost no other domain has all four, which is why most agent products are still demos.

#AI #AIAgents #SoftwareEngineering #ProductManagement #BuildInPublic

I launched Cyber Company Profiles to analyze the market position and product strategy of hundreds of cybersecurity companies. Same questions, same scales, so you can compare vendors and go deep, instead of collecting one-off AI answers.

https://zeltser.com/cyber-company-profiles

#productmanagement #artificialintelligence

Designing Flow Friction: The Core Capability of AI-Native Teams, by (not on Mastodon or Bluesky):

https://frontenddogma.com/posts/2026/designing-flow-friction/

#translations #ai #processes #productmanagement #engineeringmanagement

AI agents excel at narrow tasks but stumble on open-ended research. New study shows they lack the judgement and creativity needed for true self-improvement. It's a vital reminder that human oversight remains the bottleneck in AI progress. See the evidence: https://www.normaltech.ai/p/ai-agents-cant-yet-do-open-ended #AI #ProductManagement

What I’ve been reading (and watching) week ending 9 August https://jchyip.medium.com/what-ive-been-reading-and-watching-week-ending-9-august-1ab6b2dde859 #AI #DistributedSystems #science #engineering #SoftwareEngineering #cybersecurity #ProductManagement

A defense manufacturing startup just raised $1.37 billion to build automated factories for military parts. Investors from Andreessen Horowitz to Morgan Stanley are betting robotics and automation, not just software, is the next big tech wave. If you are deciding where to build or invest, physical automation is heating up fast. Save it.

The new automated factory startup still can't do what?

#startups #venturecapital #saas #productlaunch #entrepreneurship #technews #productmanagement #innovation

Zapier is raising the bar, moving from ad hoc AI use to a mandatory "AI fluency" for all hires. I love their focus on the "slope" of learning—it’s about continuous experimentation, not just knowing a tool. Check their V2 rubric to see the future of talent. https://zapier.com/blog/raising-ai-fluency-bar-in-hiring/ #AI #ProductManagement

Building AI prototypes is easy, but relying on natural language for engineering leads to prompt debt. Over-engineered prompts slow iteration and lock you into specific models. Move from prose to metrics to build scalable, portable products.

See how to fix it: https://www.oreilly.com/radar/the-problem-is-prompt-debt/

#AI #ProductManagement

I recently filed a support case with #Okta in which I made three suggestions to them for minor but noticeable improvements to their SSO browser extension.
In response, the support rep told me I should submit my suggestions to the "ideas" section in the Okta Help Center.
This is annoying and dumb.
#tech #software #programming #ProductManagement #SoftwareQuality #TechIsShitDispatch (1/3)

C4 diagrams for #ProductManagers https://jchyip.medium.com/c4-diagrams-for-product-managers-3a118e4f9874 #SoftwareArchitecture #ProductManagement

New analysis: Weaponized Compliance — an investigation into how extortionists pressured Apple to remove Telegram from the App Store, the methods involved, and Telegram’s response. Essential reading for product managers, compliance officers, and anyone tracking platform governance. Read more: https://wix.to/t30mRis

#TechPolicy
#Compliance
#CyberSecurity
#PlatformGovernance
#ProductManagement

Bold IT bets need fair measures. Innovation accounting shows which experiments deserve to scale and which should stop. #InnovationAccounting #ITLeadership #DigitalTransformation #ProductStrategy #InnovationMetrics #DataDriven #CloudStrategy #AIinIT #TechGovernance #EnterpriseIT #ProductManagement #InnovationLeadership
https://www.linkedin.com/pulse/innovation-accounting-when-experiments-start-paying-rent-mohindroo--moxac

As AI accelerates coding, the PM role is shifting into a 'barbell' shape. We must move away from managing the middle to mastering the fuzzy front end and go-to-market strategy. It's about outcomes over output. See why the bottleneck is moving: https://www.mironov.com/barbell/ #ProductManagement #AI

Wrote something new: what #PragmaticMarketing is, and why "your opinion, although interesting, is irrelevant." (Their words, not mine!)

Spent years watching teams argue about features in a conference room, all of them sure they knew what customers wanted. Pragmatic Marketing is the antidote to that. It's a framework built on a simple, yet revolutionary 😉 idea: go TALK to the market instead of trusting your gut:

https://itguyeric.com/blog/what-is-pragmatic-marketing/
#ProductMarketing #ProductManagement #Marketing #B2B