Kyle Krum

I’m becoming more convinced the future isn’t PMs, designers, or engineers. It’s product people. High-agency, high-taste individuals who uncover problems, create concepts, and ship solutions. Fast. The disciplines we’ve relied on? They were just constructs. Useful scaffolding—until they weren’t. AI collapses the scaffolding. The builders are what matter now. A product person isn’t a job title. It’s the engineer embedding with a customer. The designer prototyping ideas before anyone asks. The CS rep turning feedback into a feature. AI makes it possible for anyone to build. If you’ve ever blurred the lines and just made it happen—you’re probably a product person.

1,533

170 Comments

Spent a few hours with Anthropic's Claude 4. The jump isn't just incremental - this feels like a different category of tool entirely. These are qualities I think more teams should care about: 1. Handles Long Contexts Without Losing the Thread Claude can work with 200K+ tokens. That’s more than just scale — it changes how you think about model interactions. You can feed in entire legal docs, policy drafts, even books, and it still keeps its bearings. Great for continuity across large, messy inputs. 2. It Knows When to Pause One thing I appreciate: Claude is more comfortable saying “I’m not sure” or identifying ambiguity than most models. It doesn’t rush to fill in blanks. That kind of epistemic humility goes a long way in research and policy conversations. 3. Built-in Value Alignment Claude’s responses often reflect its “constitutional” training — it tends to explain why it's taking a stance and does so in a transparent way. If you’re working on systems where safety, fairness, or ethics matter, this foundation makes it a strong candidate. 4. Less Performative, More Human While GPT can sometimes default to polished, confident output, Claude feels... quieter. More thoughtful. Especially in sensitive settings: cross-cultural dialogue, mental health, collaborative ideation, etc. This tone can make all the difference. 5. Surprisingly Good at Reading the Room I've not yet seen Claude overstep. It seems to wait for cues. In multi-turn conversations, I’ve seen it defer, ask for clarification, or slow down when things get fuzzy. That’s a different kind of intelligence — one tuned to collaboration, not control. Best thing I appreciated was that it didn't feel like a model trying to “wow” me. One that’s trying to understand. That distinction might define the next era of AI systems. #AI #Future #LLM #Dev #Claude

592

31 Comments

I’ve just published the next installment in my "Theory of Durable Execution" series. Across frameworks we hear terms like workflows, activities, virtual objects, handlers, and functions. Underneath the terminology, they’re all describing a small number of execution patterns. In this new post, I break durable functions down into three forms: 🔹 Stateless functions for one-shot deterministic logic 🔹 Sessions for long-running, interactive orchestration 🔹 Actors for persistent, message-driven state To get there, I walk through: 🔹 The behavior–state continuum 🔹 How identity, lifetime, communication, and concurrency shape execution semantics 🔹 How frameworks like Temporal, Restate, DBOS, and Resonate map onto these forms This builds on the previous two posts in the series: 🔹 Demystifying Determinism in Durable Execution 🔹 The Durable Function Tree (part 1 and 2) This was the last post in this series, distilling the various frameworks down to a simpler logical model. Full post: The Three Durable Function Forms https://lnkd.in/eJHpY9qC

77

7 Comments

A year ago, if you had asked me whether our engineering team at Owner.com was truly adopting AI, not just dabbling, but actually integrating it into how we work, I would’ve said no. We had a few early adopters playing with tools. A few GitHub Copilot installs. Some LLMs used here and there. But nothing close to an org-wide shift. That’s all changed. And while we’re still evolving, I’ve seen what it takes to start moving the whole team forward. First: hire people who are curious. We’ve put deliberate effort into bringing on engineers who are open to trying new things, who see AI as an accelerant, not a threat to quality or process. It’s subtle, but it matters. You can’t force this mindset into a team. You build it from the start. Second: find your champions. Every org has a few folks who tinker in their off hours, who show up to standups with something cool they just built. Identify them. Support them. Make them visible. They’re your internal force multipliers. Then: show up yourself. I’ve made a point of using AI in my own day-to-day work. Whether it’s brainstorming in ChatGPT, automating routine writing tasks, or even building simple prototypes, when the team sees leadership leaning in, it shifts the culture. Boldness is contagious. We’ve also created real space for experimentation. Our internal AI hackathons have surfaced some of the most creative, practical tools we now use: automated code generation for straightforward tasks (more on this exciting stuff in a future post), auto-generated tests, and internal bots. Not all of them made it to production, but they created momentum. And that momentum changes everything. On measurement - we’re still figuring it out. Traditional metrics like velocity and story points fall short. We’ve started thinking about time saved, friction reduced, and AI-assisted workflows adopted, but I’d be lying if I said we’ve nailed it. It’s an active conversation, and we’re learning as we go. Throughout it all, we’ve kept our engineering guardrails. Code reviews still matter. Design quality still matters. AI is there to raise the ceiling, not lower the bar. This isn’t about chasing hype. It’s about giving your team better tools, faster feedback, and more creative freedom. The orgs that figure this out now are going to be miles ahead six months from now. There’s no playbook yet. But that’s the opportunity: you get to write your own. At Owner.com, we’re not waiting for someone else to figure it out. We’re leading from the front and taking the future into our own hands.

144

10 Comments

New rule, every time you see "Ai" written in a product or design spec, you are required to do jazz hands. "Ai driven" has become synonymous with "I didn't take the time to map out how it should work".

25

2 Comments

I began building workflows I didn’t even know were possible. Three months ago, the term “MCP protocol” didn’t mean anything to me. At Ideas‑Realized, we specialize in turning chaotic systems into seamless automations. One morning, as my coffee slowly warmed in the mug, I flipped through the MCP guide, presented in a comic‑book style that was both unexpected and brilliant. That’s when it clicked: what if a single universal protocol could make all our tools truly speak to each other? Within the hour, I had spun up a lightweight MCP client using Cursor and dropped it into an Supabase flow. Suddenly, my app and backend were exchanging data effortlessly, as if they had been best pals all along, with zero extra code. By week two, I ported that same MCP-powered file into an old‑school service desk system used by a local plumbing firm. Data flowed flawlessly. Week three, and I was mining off‑market real estate leads, prioritizing hot prospects, and feeding them directly into our CRM, thanks to just one additional MCP endpoint. Jump to week five: I layered on real‑time alerts so that when a customer responded, MCP triggered , which summoned Claude to draft a reply, and all I had to do was review and send. Now, every one of our workflows begins with MCP. Whether it’s AI integrations, custom dev, or bringing legacy systems up to speed, everything snaps together faster, ships cleaner, and stays rock‑solid. No more late‑night patchwork. Just smart, modular pieces that click. If you build real‑world automations, spend your weekend with the MCP guide. Open your favorite editor and discover how one protocol can simplify your entire stack. Credit to Daily Dose of Data Science for this brilliant guide.

58

22 Comments