About
I help companies map the pathways that allow them to disrupt and grow, successfully…
Articles by Jon
-
The problem with Large Language Models
The problem with Large Language Models
What’s all the noise? One defining factor of humanity is our ability to communicate in highly complex ways. Language…
7 Comments -
Communication and the metaverse
Communication and the metaverse
The noise of a digital world In the digital world we exchange information in so many different ways. We have a global…
1 Comment
Activity
6K followers
-
Jon Ayre posted thisPossibly the biggest catch 22 right now is businesses undertaking an AI transformation in the vain hope it'll give them the tools to complete all those other transformations they failed to complete.
-
Jon Ayre posted thisDid you "quietly" ask any questions today? Did you "read that again" or pause to "let that sink in" because someone suggested you did? Did you wake up to find that "everything changed"? Nope? Me neither. The problem with LLMs is not that they're writing all the words. It's simply that we trust words too easily. (See what I did there?)
-
Jon Ayre posted thisBinance and Polymarket hacked. Billions in cryptocurrency stolen from predominantly US "investors". This time it isn't just the apes that have disappeared. When it comes to finances, "decentralisation" often just means someone sold you a bridge they didn't own.
-
Jon Ayre posted thisOne of the biggest mistakes the IT industry made was when it drew "data" and "applications" as two separate boxes on the architecture diagram. There is no application without the data, and the data is nothing without the application. They are one and the same thing. The result is organisational dysfunction. Why does it matter? Because once those two boxes were drawn, so were the organisational boundaries. CEOs decided having a CIO wasn't enough. Now they also needed a CDO. Suddenly there are two executives with a remit to control their respective domains. Unfortunately those domains overlap, and not just a bit. Everything they try to touch overlaps. Now, set each of those exec's different objectives, and you have a great way to ensure they spend all their time fighting for control and very little of it fighting for the success of your company. Their teams are no better off either. The data people can do nothing with the data unless they can dictate what the applications can and can't do with that data. Meanwhile, the applications teams can't create their applications unless they can dictate what does and doesn't happen to the data. To succeed they have to somehow collaborate while their bosses compete. A hard enough task if they weren't also following completely different roadmaps. Data and applications are one thing, not two. Trying to get anything done in IT while they're separated is a fools errand. When it comes to software product development multidisciplinary is the way and getting your application and data people together is just one thing you need to do to make that happen. It's only one piece of the puzzle, but it's a start.
-
Jon Ayre posted thisMicrosoft: Announces record profits Microsoft: Announces 60,000 redundancies Apple: Becomes most valuable company in the world and is on track to report record profits. Apple: Lays off a couple of hundred people with a further planned 5000 sackings "on hold" for now. Watch this space. Two more reasons to use Linux on someone else's hardware.
-
Jon Ayre shared thisHey Claude. What is a tsunami? https://lnkd.in/e-J4hq7m
-
Jon Ayre posted thisThere's a very simple technique you can use to avoid absorbing misleading or just plain wrong information. If you don't have time to check it, ignore it. You'll usually be right. Under no circumstances act on it or form opinions based on it.
-
Jon Ayre reposted thisJon Ayre reposted thisWe’re really looking forward to this one. On Wednesday 11 November at 17:40, our colleague Jon Dickinson takes the stage at Gartner Barcelona with Adam Linford from AXA UK: ➡️ “Buying AI Coding Tools Is Easy. The Return Isn’t. AXA’s Route to Value.” It’s twenty minutes on what the move from AI-augmented to AI-native delivery takes inside a large, regulated organisation. Working through this with your team? Christian Singer, our Growth Director in Germany, is in Barcelona all week. Book 30 minutes with him. The link is in the comments. #GartnerSYM
-
Jon Ayre shared thisLeaders. There's more than one way to scale an organisation, and how you do it depends on what you're scaling for. Are you scaling for brains or brawn? Either can make sense but you need to decide, because how you scale will dictate what you get. Scaling for brawn means scaling production. If the service or product you provide is repeatable and predictable, then your reason for scaling is to increase output. The scaling method best suited to you, is top down hierarchy. Someone needs to analyse, decide and guide. Everyone else needs to deliver the service in a consistent way. Independent thinking might result in better outcomes in specific situations, but overall, quality will be negatively impacted. Each layer in the hierarchy needs a leader, and that leader needs a small group of subordinates who propagate instructions, feed information back, and advise where necessary. Each of those subordinates will need a similar structure, as the number of people they're leading increases. Hence we get the ubiquitous tiered hierarchy. Many people believe this structure is inevitable, but it's not the only way. There's also scaling for brains. If your organisation's primary product is human intelligence, hierarchy is not the way to go. A goal of hierarchy is to limit independent thinking below the topmost roles. Although some thought is required by a leader's immediate reports, the scope reduces significantly as you descend the hierarchy. When scaling for intelligence, this is the opposite of what you want. Adding more brains to this structure doesn't increase the IQ of the organisation. A hierarchy with 100 people has the same IQ as one with 100,000. To scale intelligence, you scale by team. A team has an optimum size, and when that size is reached, it's time to add another. Each team has the autonomy to deliver to the best of their abilities based on the situation they find themselves in. Each member of the team is free to exercise the full breadth of their experience and intelligence. For it to be an organisation, it needs a set of goals, which each team needs to respect and pursue. There has to be some form of accountability, but to avoid slipping back into the hierarchical model, that accountability also needs to be at team level and shared across all teams. This model is extremely effective, but also very uncomfortable. Many founders struggle with the lack of control this gives them over their organisation. It requires a high degree of trust and respect between all participants, and skillful hiring to ensure the core values of the organisation are maintained. It results in repetition of approach and it's a lot harder for learnings to propagate through this loose collection of teams. But these compromises are more than offset by the increase in the ability of the organisation to do great things. So, if you're scaling for brains, ditch the hierarchy, no matter how unfamiliar or uncomfortable that might be. The rewards will be worth it.
Jon Ayre commented on a post
3h
Sam merely demonstrates that when it comes to humanity and philosophy, he is an ignorant fool without the slightest recognition of how little he knows. And as for "a trillion IQ" that doesn't even mean anything.
Jon Ayre replied to a comment
1d
Uldis Barbans Unreliable isn't a single thing. Humans are unreliable in a predictable and resolvable way. They also learn in real time and experience consequences personally. They adapt and grow so the effort you put in is rewarded with improvement and reduced dependency. LLMs are consistently unreliable regardless of repetition, and experience nothing that might improve that unreliability. The one line human comparison is a lazy one that doesn't stand up to even the most cursory of examination.
-
Jon Ayre liked thisI read a lot of stuff like this 👇. Somebody lamenting the "chaos" at the end of the Sprint and then advocating waterfall development with a fully fleshed-out upfront spec as the solution. It isn't. That's a massive failure to grasp the concept. Instead, gather enough information to start, then start. Release frequently (every few hours—that's why they call it _continuous_ integration/deployment) and adjust based on feedback. Unresolved issues are probably unresolved because nobody knows the answer upfront—it takes having something in your hands to make the question, much less the answer, clear. So, you collect answers by deploying and having a conversation. No time for that within your arbitrary two-week deadline? Then work on something smaller. Reverting to waterfall in order to maximize output by shoehorning too much work into an arbitrary deadline is not viable. Working with agility is. P.S.: Hardcore waterfall (typically big-corporate) folks will push back on this, but to me, waterfall is _never_ the solution. Some problems require well-defined algorithms up front and adherence to some regulation or another (or equivalent), but well-defined, typically nonfunctional, requirements are _not_ the same as a full, inflexible specification. Spec out what you have to. Approach the rest with agility.Jon Ayre liked thisMost teams treat the last two days of a sprint as a delivery problem. It’s usually a refinement problem. I understand why. When the board is red and the deadline is tomorrow, delivery feels like the only thing to fix. Add more standups. Push harder. Cut scope. But the chaos at the end didn’t start at the end. The developer who never got a clear acceptance criteria. The story that went into the sprint with three open questions. The dependency that was “probably fine.” You only notice those things when you’re close to the work before it starts. Sprint planning that’s rushed, refinement that’s skipped, and suddenly the last two days look like a delivery failure when it was always a preparation failure. My view: if your sprint ends in fire every time, don’t change how you close. Change how you open. I know many teams will push back on this — refinement takes time and time is always the first thing cut. Where does your sprint usually break down — at the start or the end? ↓ #ScrumMaster #Scrum #Agile
-
Jon Ayre liked thisJon Ayre liked thisWe're very much at the "Ah, but what if it *didn't* have to be reliable?" stage of this AI journey.
Experience & Education
Recommendations received
4 people have recommended Jon
Join now to viewOther similar profiles
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More