惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
美团技术团队
Last Week in AI
Last Week in AI
WordPress大学
WordPress大学
博客园 - 三生石上(FineUI控件)
博客园 - 聂微东
雷峰网
雷峰网
阮一峰的网络日志
阮一峰的网络日志
博客园 - 叶小钗
IT之家
IT之家
Google DeepMind News
Google DeepMind News
D
Docker
J
Java Code Geeks
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
Apple Machine Learning Research
Apple Machine Learning Research
博客园 - 【当耐特】
V
V2EX
Hugging Face - Blog
Hugging Face - Blog
博客园 - Franky
月光博客
月光博客
宝玉的分享
宝玉的分享
酷 壳 – CoolShell
酷 壳 – CoolShell
aimingoo的专栏
aimingoo的专栏
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More

Hacker News - Newest: "AI"

AI can't read an investor deck AI as an attorney? Student uses ChatGPT, Gemini to sue UW over alleged racial discrimination Hacking MCP Servers in AI Systems – The Rug Pull: Tool Changes After Approval GitHub - MeepCastana/KubeezCut: Free Web based video editor Can AI judge journalism? A Thiel-backed startup says yes, even if it risks chilling whistleblowers Coming soon: 10 Things That Matter in AI Right Now DARPA built an AI to fact-check enemy weapons claims What explains heterogeneity in AI adoption? When AI Meets Muscle: Context-Aware Electrical Stimulation Promises a New Way to Guide Human Movements - Department of Computer Science AI Changed How We Build. It Did Not Change What Matters. Linux rules on using AI-generated code - Copilot is OK, but humans must take 'full responsibility for the… Meta spins up AI version of Mark Zuckerberg to engage with employees Code Mode: Let Your AI Write Programs, Not Just Call Tools | TanStack Blog GitHub - Delavalom/graft: Go framework for building AI agents. Type-safe tools, multi-provider (OpenAI, Anthropic, Gemini, Bedrock), zero vendor SDKs. India's TCS tops estimates, says new AI models did not dent services demand Gen Z's fading AI hype Strong feeling: we are in a folded AI reality GitHub - machinarii/total-recall-catalog: A reference catalog of latest knowledge retrieval, memory & RAG systems GitHub - mensfeld/code-on-incus: Give each AI agent its own isolated machine with root, Docker, and systemd. Active defense detects and stops threats automatically.. Quantization, LoRA, and the 8% Problem: Benchmarking Local LLMs for Production AI Iran war: We spoke to the man making Lego-style AI videos that experts say are powerful propaganda Powell, Bessent discussed Anthropic's Mythos AI cyber threat with major U.S. banks GitHub - immartian/bellamem: Persistent belief-graph memory for AI agents. Retrieves decisive context by importance — not recency, not RAG, not /compact. recursive-mode: The Repo-Native Operating System for AI Engineering After the attack on Sam Altman's home, will AI CEO's go on the offensive? The biggest advance in AI since the LLM Opus 4.6 vs GPT 5.4 One Prompt Unity World Generation Test “AI polls” are fake polls Client Challenge Can AI be a 'child of God'? Inside Anthropic's meeting with Christian leaders
I Gave My Company to AI for 6 Months — Here's What Surviv...
laurentlof · 2026-06-21 · via Hacker News - Newest: "AI"

 The Experiment That Could Have Killed My Business

If you're interested, I also have a really good article here

Six months ago, I made a decision that sounded insane to everyone around me.

I started handing over pieces of my company to artificial intelligence.

Not just the boring stuff. Not just the email drafts and meeting notes. I mean the real stuff. The systems. The processes. The infrastructure that keeps a 10-person video production company running without collapsing into chaos.

And I didn't stop there.

I moved almost every computer in my office to Linux. I killed Slack. I killed Notion. I replaced them with open-source alternatives running on my own servers. I built custom software. I connected everything to an AI agent that now runs the show.

After six months, I can say with confidence:

We are now an AI-powered company.

But here's the twist. The term everyone uses for this is "AI-first." I hate that term. It's marketing fluff, like "cloud-based" or "blockchain" back in the day.

I prefer something else.

We are Human-First, AI-Powered.

And today, I'm going to show you exactly what that means.

The Problem: Creative People Are Chaos Machines

My company makes YouTube videos. Lots of them.

Every video goes through the same process:

  1. An idea

  2. Research

  3. A script

  4. Filming (this video, right now)

  5. Product shots

  6. Editing

  7. Review

  8. Thumbnail + title

  9. Publishing

Simple, right?

Wrong.

Behind the scenes, there's purchasing, inventory management, finance, project management, sales, IT, and a thousand tiny decisions that keep everything from falling apart. And here's the thing about creative people — we're not exactly known for our organizational skills.

We miss deadlines. We get inspired at 2 AM and change everything. We lose track of gear. We forget what we said in a video three years ago.

For years, I tried to fix this with traditional software development tools.

I implemented Agile. I used Scrum, Jira, Kanban, standup meetings — everything I learned from my years as a programmer. And it helped. A little.

But I always dreamed of something more.

I dreamed of a custom application. One program that controlled everything — tasks, calendars, budgets, scripts, footage, inventory. All in one place. All perfectly organized.

The problem? Building that kind of software costs a fortune. You need a developer for six months, then more to maintain it. Most small companies can't afford that.

I couldn't afford that.

But I could afford something else.

An AI.

How I Built the Impossible

First, I did a full financial analysis of my company. Every expense, every subscription, everything we were wasting money on.

I fed it all into an LLM and asked it to categorize every single expense. It cross-referenced emails, databases, and company names. It figured out what we were actually paying for and whether it was worth it.

That's when I discovered we were spending a ridiculous amount on apps like Slack and Notion.

So I killed them.

I replaced Slack with Element (open source). I replaced Notion with Outline. I built custom integrations. I moved everything to open-source, self-hosted alternatives running on my own hardware.

Then I built Orion.

Orion started as a simple benchmark database. I'd test computers, run performance tests, and store the results. It became an inventory system. Then a task manager. Then a project tracker. Then a full operating system for the entire company.

Today, Orion handles everything:

  • Video pipelines

  • Team tasks

  • Google Calendar sync

  • Sales data

  • Inventory tracking

  • Everything

But here's the thing — I built all of this with AI.

I tell the AI what I need. It writes the code. I review it. We iterate. What used to take me months now takes days.

10x faster. No joke.

The Brain in the Basement

Remember that server I mentioned?

It's a monster. Threadripper CPU. 96 cores. 96GB of VRAM on an RTX 6000 Pro. The most powerful graphics card you can put in a PC before you need a datacenter.

Inside that server runs something called Janus.

Janus is my AI assistant. But "assistant" is an understatement. Janus is the brain of the entire company.

Here's what Janus can do:

  • Access every video we've ever made

  • Search transcriptions and pull clips instantly

  • Know which products we've covered

  • Answer any question about company operations

  • Transcribe audio

  • Send WhatsApp messages

  • Alert the team when something is confirmed

  • Find things from old videos

  • Remember decisions so we don't repeat mistakes

And every Monday morning, Janus runs a routine:

"Here are the videos that should have been published last week. Here are the tasks that didn't get done. Is it realistic to hit this week's deadlines? If not, let's reschedule. And by the way, I checked your calendar — you have a dentist appointment on Thursday. Don't forget."

Janus doesn't just track tasks. It understands context. It knows when I'm traveling. It knows when we're behind. It knows when the team needs breathing room.

And I am the only one who talks to it.

The AI Editor That Saves Hours

Filming this video took me 1 hour and 20 minutes.

That's 80 minutes of me talking, repeating myself, messing up, going off-track, and generally being a human.

My editor used to spend hours cutting this down to a clean 30-minute video.

Now we have an AI tool that:

  1. Watches the entire raw footage

  2. Transcribes every word

  3. Detects repetitions and mistakes

  4. Makes a first cut

  5. Imports product clips and b-roll automatically

  6. Generates a DaVinci Resolve project file

The editor still does a second pass. He still makes creative decisions. But the AI saves him hours of boring, repetitive work.

He can focus on making it good, not making it exist.

The Digital Clone

Here's where it gets crazy.

We have a database that contains everything we've ever said in a video. Every review, every opinion, every benchmark, every off-hand comment.

When I ask Janus about a video I made three years ago, it finds it instantly.

When the team needs b-roll of a specific product, it's already indexed and ready.

When sales needs to know if we've covered a certain brand, they ask Janus.

It's like having a digital clone of the entire company.

And this is where the "Human-First" part comes in.

The Line I Refuse to Cross

People ask me all the time: "Do you use AI to write your scripts?"

No. Absolutely not.

I tried. The AI wrote something generic, predictable, and soulless. It had no personality. No edge. No me.

And honestly? The AI thought my ideas were terrible.

I once made a video about a MacBook Pro and didn't mention the performance at all. I spent the whole video talking about the engineering, the curvature of the chassis, the angle of the edges — and how it represented a return to Apple's PowerPC era.

The AI told me it was a bad idea.

That video got millions of views.

AI can't predict what makes humans connect. It can't replicate your personality, your sense of humor, your weird obsessions, your off-beat takes. That's what makes us creators unique.

So I drew a line.

The creative part is for humans. The boring part is for AI.

Scripts? Human. Concepts? Human. Art direction? Human.

Editing, organization, scheduling, research, data management? AI.

The Team's AI Nudge

Creative people are chaotic. We don't check our tasks. We don't move our cards. We don't update our status.

It drives me crazy.

So I built Fama.

Fama is an AI that follows up with the team. It sends them messages:

"Hey, how's that edit going? Any blockers? Let me know and I'll update your tasks."

They respond with a voice note. Fama transcribes it. It updates the task board. It tells me what's happening.

I don't have to be a manager with a whip. The AI does it for me.

And they don't even mind. It's a bot. They can ignore it if they want. But it keeps things moving.

What This Means for the Future

Look, I'm not saying AI is going to replace everyone.

But I am saying that a single person with technical skills can now do what used to require a whole team.

Before AI, a small company like mine couldn't afford a full-time developer. Building custom software was a luxury. Now? One person with the right knowledge can automate massive parts of the business in weeks, not months.

The new job isn't "programmer." It's "AI implementer."

Someone who understands the business, understands the systems, and knows how to make AI solve real problems.

And that's exactly what I want to teach.

The Big Question

After six months, I've learned something important:

AI will change work. It already is.

People are losing jobs. That's real. But new opportunities are opening too. People who understand how to integrate AI into real businesses — not just prompt engineering, but actual systems — are going to be incredibly valuable.

I built a system that my company couldn't have dreamed of a year ago. I did it with almost no budget. I did it with AI.

And I'm just getting started.

So here's the question I keep asking myself:

Can anyone do this?

Do you need to be a programmer? Do you need a computer science degree? Do you need to understand networking and cybersecurity and databases?

Or can someone with curiosity, determination, and a few good prompts learn enough to build something remarkable?

I think the answer is yes.

And that's what I'm going to explore next.

The Takeaway

Six months ago, I was skeptical. I thought AI might be all hype. I thought it might be like crypto — a flashy trend that fades.

I was wrong.

AI is real. It's here. And it's changing how small businesses can operate.

But it's not replacing humans. Not yet. Maybe not ever.

It's making humans more powerful.

My company is Human-First, AI-Powered.

We do the creative work. We make the art. We have the weird ideas that nobody else has.

And the AI does everything else.

That's the future I want to live in.

Your Turn

Are you using AI in your business? Or are you still waiting to see if it's worth it?

Drop a comment below — I reply to everyone.

And if you want to learn how to build systems like this, I'm working on something. Stay tuned.

The Verdict: AI doesn't replace humans. It replaces boring tasks. And that's a future worth building.