Miki Makhlevich

dream sequence is the trend now. You gotta follow the hype otherwise what are we savages?

"...never sleeps" means no dream sequence???

@ToriSeidenstein @pk_iv @irvinebroque @zeeg @ianlivingstone @vtahowe First takeaway really reminds me "why information grows" by Cesar hidalgo. way harder to move know how over matter (although here the matter is bits so easy 😅)

Went to a Building Internal AI panel last week w/ @pk_iv (Browserbase), @irvinebroque (Cloudflare), @zeeg (Sentry), @ianlivingstone (Keycard), moderated by @vtahowe Usually I talk tech with these folks. This time it was behavioral: what AI adoption actually looks like inside their orgs. 3 takeaways 🧵

@techwith_ram now that harnesses are a thing, some of the gov needs to drift there. out-of-band file validation enables deeper checks --> lower fp/tn while not getting the agent slower. and fix one for a multi-agent.

Good AI governance isn’t about doing everything fast. It’s about understanding what matters before taking action. Most companies rush into AI governance by adding random rules, checklists, and risk assessments. But the smarter approach is to pause first and ask: - What AI systems are we actually using? - Which laws and regulations apply to us? - What risks do we truly care about? - Who are our clients and stakeholders? Because if you skip this clarity step, you’ll keep rebuilding everything later. Read the blog attached to learn more about AI governance layers.

@samgrosss made me feel so warm inside 🤩

The main visual direction for my next project, specifically designed for designers and creatives.

@urieli17 שירן היא מהאנשים המבריקים שיצא לי לעבוד איתם! ❤️ שילוב מדהים בין היכולת להסתכל במאה אלף רגל ובו זמנית לדאוג שתכלס נתקדם מחר בבוקר במשהו מדיד. בטוח פרק מעולה :)

הקלטתי היום פרק מיוחד עם שירן בראלי, VP Research בחברת Cyera והיה כל כך כיף. למדתי הרבה והבנתי איך הם מצליחים לצמוח בקצב כזה מסחרר. המסר העיקרי: לגרום ל-AI לעשות משימה זה נחמד, אבל לגרום לו לעשות את זה בסקייל שלהם זה כבר אומנות. במיוחד אם אתה לא רוצה לסבסד את חברות ה-AI...

@ToriSeidenstein @bkuan001 customer spying > customer obsession

Classic SF moment @bkuan001 posted this about tadata earlier in the week. last night we ran into each other irl at an event Best parts: - candid product feedback on a rooftop - he'd already referred a friend doing startup GTM to try it Made my week

@AI_x_il מעניין מאיזה חברה הם למדו את זה 👀 מתחרז עם רמלה לוד

ממשיכים עם ההייפ אבל לא משחררים את המודל...

@ataiiam @CopilotKit one would claim that self-optimizing UI converges to the average user. If only personalized generative UI were possible... 🙃

@makhlevich @CopilotKit 🚀 it's exciting

🌟 Introducing @CopilotKit's Enterprise Intelligence Platform Our self-hosted solution for teams taking agentic applications to production on any platform, with confidence. Everyone starts from our open-source. That's not changing. Enterprise Intelligence goes on top of it. The #1 thing we hear from companies deploying fullstack, interactive agents is that they need optionality and self-hosting. With our platform, you get persistent threads, analytics, and self-learning that captures the full interaction surface between agents and users, which is framework agnostic via AG-UI. This is the DNA of CopilotKit. We don't just save chat history, we record the interaction: - what your users did (asked, edited, answered, ran a tool...) - how your agent responded and more... Every agent-user interaction becomes a feedback signal. Your agents improve just by being used. Sign-up here: go.copilotkit.ai/enterprise-int…

@ToriSeidenstein Oh wait there is a link already 😅

@ToriSeidenstein ACTUALLY

How do the best salespeople *actually* do it? "Claude Code for GTM" is having a moment, rightly so, but AI without sales finesse just gets you faster slop. So when we built our sales agent library, we didn't write the playbook. We went to the people who already had. 🧵

@NGrossman1999 md is dead, read all about it at markdown.rip

@makhlevich This post gives me md vibes

@shannholmberg Love the review gates wording! Had somewhat related take on the connection between form factor and moving between stages x.com/makhlevich/sta…

Thariq opened my eyes to HTML vs Markdown His thesis is that HTML is more effective than MD: > HTML gives agents a richer canvas than Markdown: diagrams, styling, layout, tables, SVGs, images, and interactive elements. > It lets humans stay in the loop with agents: instead of reviewing walls of text, you can inspect, tweak, annotate, move sliders, copy changes back into prompts, and use the artifact as a working interface I´ve found that this framework works best for me: > Markdown for agents only > HTML for agents and humans practically, that looks like this: I´m having a content system bring ideas to a written draft, but reviewing ideas, drafts and the whole workspace end-to-end in Markdown is terrible UX HTML solves that, so I´m using it for review gates, overview of agentic systems and to enable humans to be in the loop

Tangentially: I also wrote the actual serious version of this take --> x.com/makhlevich/sta…

Everyone thinks the markdown → HTML push is about more output tokens. That's marginal money. The real money is making every dev rebuild their markdown viewer pet project as an HTML viewer. Game on, Anthropic. I shipped mine: markdown.rip

@JayaGup10 try tadata! it's Claude agents running not locally and more :)

:(

the bottle neck sets the pace and it's often not the model capabilities

@audiencon tadata.com

drop your project i’m boosting builders tonight 👇