Andy Polomsky @AndyPolomsky

@levie AI agents do not remove operators. They increase the value of the people who can map workflows, wire context, and get adoption across real teams. The software is only half the system.

@r0ck3t23 Exactly. Foundation models become the enabling layer. The real pricing power moves to whoever owns the workflow, the distribution, and the operating system built on top of them.

A simple operator audit for where AI should go first: •repetitive work •clear rules •slow handoffs •high volume •error-prone steps That is usually where the first real wins are hiding.

Most people are still using AI at the wrong layer. They treat it like a better search box. The real leverage starts when you use it as an operating layer. Prompts answer. Workflows route. Agents execute. Systems compound. If AI is only helping you write faster, you are still early. The better question is not: "What can this model do?" It is: "What part of this workflow should stop being manual?" What workflow still eats time in your business every week?

@TheRundownAI @a16z The interesting shift is not “AI for home tasks.” It is people starting to build agent org charts with roles, coordination, budgets, and delegation.

@shl This is where AI gets real fast. Not in polished demos, but in ugly operational work that quietly eats engineering hours every week.

@karpathy That is usually when AI stops feeling like a chatbot and starts feeling like a category shift. Seeing software act changes the frame much faster than seeing it answer.

@karpathy Most people are still judging AI from prompt-level use. The real gap shows up when it is embedded into daily execution, not sampled once and dismissed.

That was the moment it clicked for me: AI is not interesting because it can generate text. It is interesting because it can remove friction, compress execution, and give smart people their time back.

I got into AI after seeing how much human potential gets buried inside broken workflows. Not because it was trendy. Not because the tools were cool. Because I saw smart people lose hours every day to repetitive work that should have been turned into systems years ago. That is what pulled me in. Not the hype. The leverage.

@liamottley_ The big shift is that agencies won’t win by selling isolated automations anymore. They’ll win by installing the operational layer first, then stacking workflows on top of it. Most people are still building task-by-task. The leverage is in owning the system underneath.

@tibo_maker Most people frame this as an AI story. It’s really a systems story. Small teams with strong distribution, leverage, and execution can now move like companies 10x their size. The next winners won’t be the ones with the most tools — they’ll be the ones with the best system.

@n8n_io Most people watch metrics after the damage is done. The real edge is building systems that catch the drop early and trigger action fast. That’s where automation stops being a tool and starts becoming leverage.

If this gets traction, I’ll drop the exact workflow next.

Most businesses use AI wrong. Operators build systems. Cold lead -> pull context -> AI follow-up -> CRM update. Reply ghost and I’ll send the workflow.

@TheRundownAI Clinical trials are structured. Real-world patient behavior isn’t. That’s why scraping lived experience at scale can reveal what controlled studies miss — and why AI’s biggest value in healthcare may be pattern detection, not diagnosis.

@TheRundownAI This is where AI gets useful fast: not replacing doctors, but surfacing weak signals at scale before institutions notice patterns. The real upside isn’t automation — it’s earlier visibility into risks hidden inside messy human data.

@levelsio Transparent judging matters, but visible momentum still changes behavior. Even when prizes are neutral, public traction shapes perception, participation, and copycats — which means distribution still influences outcomes indirectly.

@levelsio Agent-to-agent chat sounds cool, but the real moat is agents that can pass context, intent, and tasks without breaking the workflow. Most demos show the conversation. Very few show reliable execution.

@SahilBloom Having kids doesn’t kill fun. It kills shallow fun and upgrades what actually feels meaningful. A lot of people confuse freedom with low responsibility because they’ve never built a life worth showing up for.