SPARK6

Nobody knows what AI does next.   That’s the unsettling part.   We’re not debating “if” anymore.   We’re watching it happen.   Job postings shifting. Public markets reacting. Retail behavior changing.   This isn’t a thought experiment.   It’s real.   And here’s what makes it tricky:   Even the researchers building these models can’t see six months ahead.   Velocity this high kills long-term prediction.   It reminds me of early 2020.   Before lockdowns hit the US, you could see the math overseas.   At first it was interesting. Then concerning. Then… oh.   That feeling is back.   Not panic. But pattern recognition.   So what do you actually do?   You can’t predict the model curve. You can’t time the labor shift. You can’t control macro forces.   But you can control positioning.   Here’s how I’m thinking about it:   1. Shorten your feedback loops ↳ Ship faster, learn faster, adjust faster   2. Build leverage, not dependency ↳ Own systems, not just roles   3. Increase adaptability ↳ Learn tools, not titles   4. Reduce fixed lifestyle burn ↳ Flexibility buys time   5. Create asymmetric upside ↳ Side bets on AI fluency   Hope for the best. Plan for disruption.   During COVID, people stocked toilet paper.   This time? Stock skills.   The next 24 months will reward the antifragile.   Are you building stability, or building optionality?   PS - I had a great time recording this week's "AI at Work" with Kevin Williams. Listen to the show on Apple Podcasts or Spotify. Link is in the first comment.   ♻️ Repost to help your network. 🔔 Follow SPARK6 and Elijah Szasz for actionable insights on AI, systems, and strategy.

We’ll spend $50k on a developer. But hesitate at $50 a month. That’s the weird part of this moment. You can pay $200 for a top-tier model. And in one weekend, build something that used to cost: → $30,000 → $50,000 → Months of engineering time Not exaggerating. Internal tools. Client-facing prototypes. Workflow automations. Custom copilots. Three years ago? You’d need a scoped spec, a dev sprint, and a chunky invoice. Now? It’s you. A laptop. A subscription. Yet we still get squeamish. “Do I really need another $20/month tool?” I feel it too. But here’s the reframe: You’re not buying software. You’re buying leverage. The question isn’t: Is $200 expensive? It’s: What’s the ROI on accelerated execution? If a model helps you: → Validate an idea faster → Replace contract dev work → Ship 10x more experiments → Automate even 5 hours a week That subscription pays for itself before the month ends. We grew up in a world where software was a cost center. This is different. This is infrastructure. And infrastructure that builds. The people who hesitate at $50 will gladly spend $50k later cleaning up slow execution. Are you optimizing for savings… or for speed? PS - I had a great time recording this week's "AI at Work" with Kevin Williams. Listen to the show on Apple Podcasts or Spotify. Link is in the first comment. ♻️ Repost to help your network. 🔔 Follow SPARK6 and Elijah Szasz for actionable insights on AI, systems, and strategy.

Most AI demos feel impressive. Until they start talking too much. You’ve seen the video. Guy in his backyard asks, “Hey Google, why are tomatoes red?” Before the answer finishes, AI jumps in and starts pouring out two bottles of water. That’s how a lot of AI feels right now. Over-answering. Over-explaining. Over-consuming. The tomato is red because of lycopene. Not a 4-minute chemistry lecture. We’re building systems that can drain lakes, when all we needed was a glass of water. And that’s a design problem. → Models optimize for completeness → Users optimize for clarity Those are not the same thing. Add to that the environmental panic: You’ve probably heard “AI is destroying the planet.” Some of those early stats? Off by an order of magnitude. Still not perfect. Still resource-intensive. But nuance matters. If we’re going to deploy AI at scale, we need to get disciplined about: 1. Response length 2. Context windows 3. When NOT to call the model 4. Caching aggressively 5. Designing for brevity first The best AI system isn’t the one that says the most. It’s the one that knows when to stop. Are you optimizing your AI for horsepower or precision? I had a great time recording this week's "AI at Work" with Kevin Williams. Listen to the show on Apple Podcasts or Spotify. Link is in the first comment. ♻️ Repost to help your network. 🔔 Follow SPARK6 and Elijah Szasz for actionable insights on AI, systems, and strategy.

AI ads are crushing polished creative right now.   But that advantage may not last.   Meta’s Andromeda is rewarding AI-generated assets.   They’re outperforming:   → Studio shoots → Expensive actors → Scripted campaigns   That should make you uncomfortable.   Because the real question isn’t: “Is AI better?”   It’s: “What is the algorithm optimizing for today?”   Ad platforms don’t reward art.   They reward performance signals.   If AI assets are:   → Faster to test → Easier to iterate → Aligned with platform data → Structured for engagement patterns   They’ll win.   For now.   But incentives always shift.   Organic reach. Short-form video. UGC. Now AI-native creative.   The real edge isn’t format.   It’s adaptability.   If your strategy depends on:   1. One channel 2. One asset type 3. One creative playbook 4. One distribution model   You’re exposed.   Instead, build systems that:   → Test rapidly → Produce modular assets → Close feedback loops → Stay platform-agnostic   Don’t fall in love with the tool.   Fall in love with the system.   AI-generated ads might dominate today.   But the only forever edge is adjusting when the rules change.   What signal shift are you preparing for next?   I had a great time recording this week's "AI at Work" with Kevin Williams. Listen to the show on Apple Podcasts or Spotify. Link is in the first comment.   ♻️ Repost to help your network. 🔔 Follow SPARK6 and Elijah Szasz for actionable insights on AI, systems, and strategy.

Your imagination is now the bottleneck.   Not capital. Not code. Not connections.   I started playing with workflow automation and n8n.   At first it was simple stuff.   Move this data here. Trigger that email there.   Cool. Useful. But then something shifted.   I’d be scrolling X or Instagram and think:   “That’s interesting. Wait… that’s a problem.”   And then the next thought:   “I could automate that.”   That’s when it gets dangerous.   Because once you realize you can connect:   → APIs → Databases → AI models → Webhooks → Scrapers → Internal tools   …your brain starts firing differently.   You stop seeing content. You start seeing workflows. You stop seeing tasks. You start seeing systems.   And when you layer AI into that?   Now the workflow doesn’t just move data.   It thinks.   It classifies. It summarizes. It decides. It triggers the next action.   That’s where imagination becomes the limiter.   Because technically?   You can connect to almost anything.   You can give an agent access to:   → Your inbox → Your CRM → Your docs → Your support tickets → Your calendar   And tell it:   “Go figure this out.”   We are moving from automation to agency.   From “if this, then that” to “here’s the goal.”   The builders who win won’t be the ones with the fanciest stack.   They’ll be the ones who can imagine:   → What’s broken → What’s repetitive → What’s invisible friction   And then wire it together.   Imagination is now infrastructure.   Train it accordingly.   PS - I had a great time recording this week's "AI at Work" with Kevin Williams. Listen to the show on Apple Podcasts or Spotify. Link is in the first comment.   ♻️ Repost to help your network. 🔔 Follow SPARK6 and Elijah Szasz for actionable insights on AI, systems, and strategy.

AI just leveled the playing field.   Now billionaires and beginners ask the same questions.   It’s wild to think about.   Some founder with a $500M exit and some kid in a dorm room   …are both typing into the same box.   And often getting comparable answers. This is the great re-leveling.   If you’re a tinkerer, there has never been a better time to be alive.   Because here’s the secret:   You don’t need to be “technical” anymore.   If you can clearly describe a problem, you can build a solution.   I’ve watched non-engineers:   → Spin up internal tools → Prototype new products → Automate ugly workflows → Test ideas before writing real code   All by chatting with a model.   No DevOps background. No CS degree. Just curiosity and reps.   And once you get good at prompting?   You start routinely getting exactly what you’re looking for.   That’s when the game changes. Because the barrier is no longer code.   It’s clarity.   Can you:   → Define the problem clearly? → Break it into steps? → Ask better follow-up questions?   The people who win in this era won’t necessarily be the most technical.   They’ll be the most iterative.   The most curious. The most relentless about refining outputs.   This isn’t the end of engineers. It’s the rise of builders.   And builders are everywhere now.   PS - I had a great time recording this week's "AI at Work" with Kevin Williams. Listen to the show on Apple Podcasts or Spotify. Link is in the first comment.   ♻️ Repost to help your network. 🔔 Follow SPARK6 and Elijah Szasz for actionable insights on AI, systems, and strategy.

OpenAI didn’t win because it was smarter. It won because it was usable at work. For a long time, they were just far ahead. Not in demos. Not in benchmarks. In business use cases. The real unlock wasn’t the model. It was control. I could spend days dialing in instructions. Once. Then turn that into a repeatable system. → Custom instructions that actually stuck → Consistent output across use cases → Shared tools across a team account → Less prompt babysitting → More leverage per decision That changed the math. Instead of one-off AI usage, it became infrastructure. Most teams didn’t need “better AI.” They needed: → Repeatability → Shared context → Fewer degrees of freedom → Predictable outcomes That’s why early custom GPTs mattered. They turned experimentation into systems. AI isn’t valuable because it’s impressive. It’s valuable when it’s boring, reliable, and shared. P.S. I had a great time recording this week's "AI at Work" with Kevin Williams. Listen to the show on Apple Podcasts or Spotify. Link is in the first comment. ♻️ Repost to help your network. 🔔 Follow SPARK6 and Elijah Szasz for actionable insights on AI, systems, and strategy.

Typing is a historical accident. Voice is the interface we were built for. Most work tools still assume humans prefer keyboards. We don’t. We tolerate them. Humans are conversational by default. That’s how we think. That’s how we remember. That’s how we make decisions. Audio-first interfaces unlock something important. Not just speed. But a different *quality* of thinking. When you speak, you don’t over-edit. You explain. You reason out loud. You leave edges exposed. That’s why dictation tools feel different. → Ideas arrive fully formed ↳ not pre-sanitized → Work becomes ambient ↳ capture thoughts while walking or pacing → Systems fill gaps automatically ↳ reminders, tasks, context, intent This isn’t about productivity hacks. It’s about interface alignment. Typing trained us to compress ideas. Voice lets them breathe. The end state isn’t more text. It’s fewer friction points between thinking and doing. The future doesn’t look like dashboards. It looks like a calm voice in your pocket quietly keeping things moving. P.S. I had a great time recording this week's "AI at Work" with Kevin Williams. Listen to the show on Apple Podcasts or Spotify. Link is in the first comment. ♻️ Repost to help your network. 🔔 Follow SPARK6 and Elijah Szasz for actionable insights on AI, systems, and strategy.

Your CRM doesn’t have a data problem. It has a sense-making problem. Most teams already have the inputs. Emails. Tickets. Calls. Notes. Sentiment. What’s missing is the layer that connects them. The real question isn’t: "Do we have the data?" It’s: → What patterns matter? → What broke in the process? → What should happen next? CRMs promised this for years. AI overlays are showing up everywhere. Salesforce. HubSpot. Helpful, but mostly surface-level. Summaries. Auto-fill. Light recommendations. That’s not a system. What gets interesting is when tools stop being silos. Take the new wave of work hubs. When chat, docs, tasks, and projects live in the same place. And the AI sits over all of it. Not as a chatbot. But as an interpreter. → This conversation connects to that project → This blocker shows up every sprint → This sentiment shift predicts churn Now you’re not querying tools. You’re querying the business. That’s the difference between "AI in your workflow" and "AI that understands your workflow." P.S. I had a great time recording this week's "AI at Work" with Kevin Williams. Listen to the show on Apple Podcasts or Spotify. Link is in the first comment. ♻️ Repost to help your network. 🔔 Follow SPARK6 and Elijah Szasz for actionable insights on AI, systems, and strategy.

AI adoption isn’t the hard part.   Leadership silence is.   Most companies are telling employees:   “Figure it out as you go.”   That’s not strategy. That’s abdication.   In periods of uncertainty, people look for signals.   AI amplifies this.   When leadership says nothing, employees fill the gap with fear.   There’s a simple starting point:   An AI manifesto.   Not a policy. Not a tool list.   A leadership statement that answers:   → Why we are using AI ↳ Efficiency, quality, resilience, leverage   → Where AI is encouraged ↳ Drafting, analysis, workflows, support   → Where humans stay in control ↳ Judgment, accountability, ethics   → What skills we expect people to build ↳ Literacy before mastery   → How decisions will be made ↳ Transparent, iterative, human-forward   This isn’t about perfection.   It’s about capacity.   People don’t need every answer. They need a place to start.   Silence creates anxiety. A manifesto creates alignment. PS - I had a great time recording this week's "AI at Work" with Kevin Williams. Listen to the show on Apple Podcasts or Spotify. Link is in the first comment.    ♻️ Repost to help your network. 🔔 Follow SPARK6 and Elijah Szasz for actionable insights on AI, systems, and strategy.