Amit | Frogomo | AI 🐸

"all white-collar work automated in 18 months" really? microsoft's AI chief mustafa suleyman just told the financial times that lawyers, accountants, marketers, and project managers will be "fully automated" by late 2027. i've been tracking AI automation closely. here's what the actual data says: the prediction: → "human-level performance on most, if not all, professional tasks" → "most tasks that involve sitting down at a computer will be fully automated" → timeline: 12-18 months the reality: 1. 80% of workers are refusing AI adoption fortune reported last month that 54% of workers bypassed company AI tools in the past 30 days and did the work manually instead. another 33% haven't used AI at all. combined: 8 in 10 enterprise workers are either avoiding or actively rejecting the technology. 2. only 29% of companies see significant ROI writer's 2026 enterprise AI survey: 97% of executives say they benefit from AI personally. but only 29% report significant organisational ROI. individual productivity gains aren't translating to business outcomes. 3. 95% of AI pilots fail to produce measurable impact MIT's NANDA initiative found that 95% of generative AI pilot programs fail to deliver measurable financial results. the failures stem from poor workflow integration and misaligned organisational incentives — not model quality. 4. AI actually made experienced developers slower METR's randomised controlled trial (february-june 2025): experienced open-source developers using AI tools took 19% longer to complete tasks. before the study, these same developers predicted AI would make them 24% faster. 5. only 8.6% have AI agents in production recon analytics surveyed 120,000+ enterprise respondents: only 8.6% have AI agents deployed in production. 63.7% report no formalised AI initiative at all. deloitte's tech trends 2026: only 11% have agents in production. 42% are still developing their strategy roadmap. 6. gartner predicts 60% of AI projects will be abandoned the 2025 gartner survey on data management: organisations will abandon 60% of AI projects through 2026 due to lack of AI-ready data. 7. the trust gap is massive walkme's state of digital adoption report: → 61% of executives trust AI for complex decisions → only 9% of workers do that's a 52-point trust chasm. here's my take: suleyman isn't wrong about AI capability. the models can do impressive things. but "can do" and "will be deployed at scale" are completely different problems. automation requires: → clean, structured data (most companies don't have it) → workflow integration (most pilots fail here) → employee adoption (80% are refusing) → organisational change (takes years, not months) → trust (9% of workers trust AI for complex decisions) the bottleneck was never the model. it's everything around the model. 18 months to automate white-collar work? maybe 18 months to automate a handful of narrow tasks in a handful of companies with exceptional data infrastructure and change management. but lawyers, accountants, marketers, project managers "fully automated"? the data says otherwise. sources: → fortune (suleyman interview, worker rebellion data) → METR (developer productivity study) → MIT NANDA (pilot failure rates) → writer/workplace intelligence (enterprise AI survey) → walkme (digital adoption report) → deloitte tech trends 2026 → gartner data management survey → recon analytics enterprise survey

nick saraev just dropped a claude code tutorial that's different from everything else out there. most content is surface-level intros that leave you more confused. this is zero to shipped. full build. real judgment. when to let claude run. when to intervene. how to go from demo to production. compress months of trial and error into one sitting. bookmark this.

@owoldestiny I get it, Owol

@frog_omo still cursor mostly, but code is growing on me for certain things

Are you still using the cursor, or have you switched to Claude Code? and for workflows — n8n, make, or something else?

@shubh19 I do the same, man!

@frog_omo i use both cursor and claude code because they serve different roles in my dev workflow

@reeder1865 Have you ever shifted to Claude?

@frog_omo Codex currently. App is stable and fast. Just does what I want it to do without a lot of prompts/ steering.

@simplydt Oh, that's great...Why n8n?

@frog_omo still on cursor for most, n8n for automations

@bkarishma360 That's true, the real advantage comes when you train AI on your taste, decision-making, writing patterns, and standards. That’s when it stops sounding artificial and starts becoming an actual extension of your workflow.

@frog_omo Most people think prompting is the skill But the real unlock is teaching AI how you think, write, and filter ideas. Without a style guide, AI sounds like the internet. With one, it starts sounding like you

most people give up on AI writing too early. they get generic output and assume AI isn't good enough. but the model is doing exactly what it was trained to do: produce safe, average writing. the fix: an AI style guide. here's what goes in one: 1. voice and tone don't say "smart and conversational" — too vague. get specific: → how formal? → how much emotional temperature? → what tone feels wrong? 2. structure AI infers structure poorly. make it explicit: → how do you open? → how quickly to the point? → how do you end? 3. sentence-level preferences → short vs long sentences → concrete vs abstract language → punctuation preferences "$400/month replacing $400k/year" — not "cost-effective" 4. anti-patterns (most valuable section) build a blacklist: → hedges: "actually," "maybe" — delete → "not X, but Y" — rewrite → meandering intro — start with friction → saggy conclusion — end by extending 5. examples show good: "I used to be physically unable to open my email." show bad: "at the end of the day, it's still just a tool." 6. revision checklist → stakes clear by paragraph one? → does this sound like a real person? → ending extends (not summarises)? how to build one: don't write from scratch. let AI interview you. → give it 3-5 examples of your writing → ask it to interview you about tone, structure, rhythm → react to examples: "too corporate," "too polished" → turn interview into draft guide → test for a week, revise based on mistakes without guidance, AI converges toward the mean. with a style guide, it converges toward you.

@opentestudox People underestimate how much invisible engineering goes into making web apps feel “instant.” Good browser software hides an incredible amount of complexity underneath.

Building high performance software for the browser can be harder than building native apps not because the web is weak but because browsers operate inside extremely constrained environments RAM management becomes a constant battle you are constantly fighting with memory pressure, GC pauses, browser crashes tab limits storage bottlenecks and cross-browser behavior differences the hardest part is not making it work on your machine It is making it work reliably across low memory devices and multiple browsers Modern browser engineering is much closer to systems programming than most people realize

@divyavani_ai Interesting how quickly AI products move from: “Can we build this?” to “Should we legally and ethically build this?” Voice cloning especially feels like one of the most complicated spaces right now.

This week I started working on voice to voice(V2V). Tried free tools first — yt-dlp, FFmpeg, pyannote-audio, Resemblyzer. They're fine for clean 2-speaker audio. Throw in background music, 4+ speakers, variable quality? They fall apart. Adobe Podcast Enhance was the real find. De-noising and music removal that actually work. Worth the subscription. Built an end-to-end pipeline. Then I read about voice rights in India. Actors own their voice. Studios own the work. ElevenLabs ToS bans unauthorized cloning. Bombay HC just ruled against AI voice clones. Licensing? Lakhs at minimum. Solo founder, no funding. Not the path. Switched to ElevenLabs Library — pre-licensed voices, ships in days. The build was the fun part. Stopping was the hard part. Voice in beta this week. #BuildInPublic #AI

@arshi_mohib One thing I respect here is that you didn’t kill the project just because growth was slower than expected. A lot of good products quietly die during that phase

A Small decision about Screenforge. When I started building Screenforge, I thought it would be freemium. Core features free, advanced stuff paid. Then I spent a month building, improving UX, shipping updates, trying to market it, posting on X and Reddit with my tiny audience, and watching growth move... very slowly. At the same time, every week I saw another screen recording tool launch. Monthly subscriptions. One-time payments. New competitors everywhere. Honestly, I almost stopped. But I had already put too much into Screenforge to let it die. More importantly, people were using it. So I made a decision: Screenforge is now free forever. No "free for now." No "early access pricing." Free. I'll keep improving it. I'll keep shipping. I'll keep competing with products that lock things behind paywalls. How will I sustain it? Simple. I won't treat Screenforge like a business I need to force revenue out of immediately. I'll work on other ideas too and hopefully fund its development from there. I also added a support option inside the app. If Screenforge saves you time and you want to help keep it alive, you can support it. I built it because not everyone records demos every day, and paying recurring subscriptions for occasional use never felt right. I hope this decision ends up being beneficial for Screenforge. You can try Screenforge from below 👇

@anomsiiwa One thing I’ve noticed is that developers value “trustworthy workflow behavior” way more than flashy demos once these tools become daily drivers.

Codex is a banger. One underrated thing I love about codex is when you hit the usage limit mid-task, it doesn’t break. It keeps executing until the task is finished properly. Claude stops instantly the moment you hit the limit. Then you wait, restart the task from scratch, and risk hitting the limit again. That difference matters. Shoutout to the @OpenAI @OpenAIDevs team. Codex is the best.

@souvenger Love this hiring approach, honestly. Real projects usually say way more than resumes now, especially in AI engineering. Feels like RAG + agents + MCP workflows are quickly becoming the new “portfolio projects” for AI devs.

My team is hiring backend and AI engineer with 3+ yrs of exp . Just share one of your best work in the comments. It will be grt if that uses some kind of RAG pipelines , Agents , mcps etc

@HyveMindx1 What stands out to me is that you’re optimising for openness and flexibility instead of controlling the full ecosystem. That usually creates much more interesting developer experimentation long term.

I have recently shipped QuantaLLM & the most common question I've been getting is: "How is this different from Google AI Edge Gallery?" Totally fair question. I get it. Both projects are about running AI locally on Android, but honestly, the why behind them feels pretty different to me. Google AI Edge Gallery reads like a polished showcase. It's built around Google's own ecosystem, their stack, their models (Gemma), and it's great at what it does. QuantaLLM is something I’ve been building more like a general-purpose local inference runtime for Android. Less of a showcase, more of an infrastructure layer, where open-source model families (currently quantized GGUF-based models) can be pushed onto the device and run locally. The direction I'm pushing toward: - Broader model and runtime support (GGUF now, ONNX on the way) - Real control over inference behavior, not just preset demos - Runtime tuning for the actual constraints of handheld hardware - AIDL-based IPC, so other Android apps can plug into QuantaLLM directly, almost like calling a local AI service The core idea behind all of this is something I keep coming back to, Modern phones are genuinely capable machines. NPUs like Qualcomm Hexagon are doing real work, we just don’t treat them that way yet. I want to explore how far local inference can realistically go on a handheld device, not in theory or lab setups, but in practical day-to-day usage. It’s still early. Very experimental. But the response so far has genuinely been beyond what I expected! ✨ quanta-web-pi.vercel.app/?v=8/ #EdgeML #GoogleAIEdge #BuildInPublic

@1001binary Honestly, one underrated side effect of AI coding tools is that people now stay glued to screens even longer because the feedback loop never stops. Feels like intentional breaks are becoming a real productivity skill now.

Hey indies, most of you don’t have a productivity problem. You have a “forgetting to step away from the screen” problem. You sit 8–12 hours at your laptop “just fixing one thing” and suddenly it’s night, your neck is broken, and you’ve eaten something questionable. Classic. Yesterday, @RoyInProgress asked something that stuck: while Claude Code is running… what are you doing next? And I realized… a lot of us never actually step away anymore. So I started forcing 2-minute breaks during the day using MicroMoment - an Apple Watch app I designed. Just: stand up. breathe. reset. don’t become furniture. Turns out I ship better when I’m not slowly merging with my chair. Who knew. Try it if you want: apps.apple.com/app/micromomen… #indiehackersbuildinpublic #buildinpublicjourney #buildinpublicsaaS #indiedeveloperlife #solopreneurjourney #indieappdevelopment #applewatchapps #watchosapps #productivitytoolsforcreators #deepworkhabits #focusandflowstate #digitalwellbeingapps #screenlifebalance #burnoutpreventiontools #creatorproductivity #startupgrindlife #founderjourneydaily #microhabitsforlife #habitbuildingapps #modernworkculture #remoteworkproductivity #saasfounderjourney #bootstrappedstartup #minimalproductivitytools #attentionmanagement #worklifebalancejourney

@zjpeng94 A lot of breakthrough people are not experts in one narrow thing forever. They’re extremely good at learning unfamiliar systems quickly and connecting dots others miss.

There are many ways to demonstrate intelligence. And recommendation is an important one.

@rcelanodev revops-calculator.frogomo.com what do you think of this, Roberto

Good morning devs! 💻 What are you shipping today? Drop your project below — let's connect 👇

@priya_mukta @OpenAI Honestly, I feel this is one of the coolest things about AI: how it’s letting people explore fields they never imagined themselves entering before. The barrier to experimentation dropped massively.

WHATTT NO WAYYY🫠 Me in my 7 lives would’ve never believed that one day I’d write code. Law school really had me thinking my future involved courtrooms. Nobody mentioned I’d be out here vibe coding with Codex AI is committing some serious academic fraud on my behalf #VibeCode

@imabhijeet_s Nothing much, Abhijeet. What about you?

Good morning X ☀️ Hope today brings: good ideas, productive coding, fewer bugs, and a little progress toward your goals 🚀 What are you working on today?