Krishna 🚀 $0→$100k ARR

I'm building my first SaaS to $100k ARR. Nights & weekends. While working full-time. AI-powered CRM for WhatsApp & Instagram. $0 MRR. No investors. No co-founder. I'm sharing everything publicly revenue, mistakes, lessons. Follow along for the raw, unfiltered journey.

idea alone is not enough, and implementation alone is not enough either. I see so many websites and web apps that are vibe-coded, work end-to-end in a closed network, but have zero customers. They are facing a sales challenge. They are facing a trust and distribution challenge, where people still give weightage to brands that were built before AI and have earned trust and transparency over time and those aren’t even AI-native companies. So even if the idea is taken or already exists, if you implement it while covering all the gaps and pain points of the existing product, it will do wonders. And by the way to do well, you don’t need the complete market share.

Every idea feels taken. Every API already exists. Every SaaS has 12 competitors. So what do we even build now?

I can see all of the internet talking about one common topic layoffs. There was a time when people used to think companies like Meta, Google, Apple, Microsoft, LinkedIn, Oracle were safe careers. There was even a term "retirement companies" where all these orgs used to fit in well. But that has vanished since November 2022 when the market started crashing and uncertainty hit the ground. People were just trying to recover from the Covid crisis and getting back on track. Then suddenly AI entered and completely changed the whole landscape of "safe job" and "safe company." Now I think everyone feels the fear of losing their job. Feels like “safe job” was always a bit of a myth we’re just realising it now. The only thing that’s actually safe is staying relevant keeping up with real trends, building skills that create value, not chasing fancy stuff, and adapting fast.

The gap between "I have an idea" and "I shipped a product" got small. Agreed. But from what I'm seeing in my network even among solo part-time builders most people shipping products are completely clueless about distribution and end users. The pattern is the same everywhere. Buy a Claude subscription, search for micro-SaaS ideas, look for a problem within your company or just assume the problem exists, and start building. Nobody talks to customers first. Code is cheap. But building something meaningful is still hard. Distribution is still hard and it's gotten harder with thousands of vibe-coded apps that all look the same. HubSpot users are not going to sign up for a HubSpot alternative vibe-coded by 2 people with zero credibility. They just won't. Trust, transparency, reliability these still matter. And security has become the most important parameter for any newly created SaaS. Your customers are handing you their data. They don't care how fast you built it. They care if they can trust you with it.

the gap between "i have an idea" and "i shipped a product" just got so small it's basically not a gap anymore for anyone, anywhere

@gvanrossum LLM = CPU (data: tokens not bytes, dynamics: statistical and vague not deterministic and precise) Agent = operating system kernel

This is very sad. I can see AI anxiety in my close circle, on social media, everywhere. And it's heartbreaking how employers are treating people with 15+ years of experience like they're disposable. What makes it worse half the people on X are using this fear to sell courses, gain followers, or just farm engagement. Most of them have no idea what the end result of this panic looks like. It looks like this. A family destroyed. Big tech CEOs are pushing this narrative through podcasts, news articles, live shows not because they care about you, but because they're under investor pressure to sell subscriptions/gain users and justify their AI spending. It's a planned agenda dressed up as thought leadership. I'm not an expert. I don't know what the future holds. But I know this ur life is more precious than any job. You can bounce back from a layoff. You can recover from failure. You can rebuild after losses. But your family only has one you. Stay close to your people in tough times. Make a plan. Work on it. Good days will come. And if the EMIs are drowning you sell the flat, sell the plot, sell the car. Get rid of it. None of it matters more than you being alive. Don't let AI panic on social media decide the value of your life.

A tragic incident unfolded in Bengaluru when a techie, Bhanu Chander Reddy, was found dead in his apartment, having hanged himself. Shortly after he was found dead, his wife jumped from their building's 18th floor. Reddy's suicide note revealed his distress over losing a lucrative job in the US due to AI cuts, which led to anxiety and joblessness. Scroll the thread for details👇

Definitely not the vibe coder. This needs people who actually know architecture, who have depth of the system not "hey I want to build X with this and that, do it for me. If you don't know what's happening behind the prompt, you're not building a better version. You're just wrapping it differently.

Now that Claude Code source code is publicly available, the real question is: who is building a better version of it?

Our WhatsApp-based CRM is still in beta. Onboarding is manual. Self-signup doesn't exist yet. Half the flow is held together with duct tape and Google Sheets. And we just got our first paying customer. Collected the GST docs, CIN, signed the agreement. Next month a small invoice gets added to the MRR counter. It's not life-changing money. But it's real. People keep saying "the market is too crowded." Sure. But I don't need the whole market. I need 50 customers who trust me to solve their specific problem. This is what a true MVP looks like not a polished product with zero users, but a rough product with one user who's actually paying because it works for them. Slow. Steady. Useful. That's the whole plan. $0 MRR → not $0 anymore. Will keep sharing the journey the wins, the mess-ups, all of it.

Can someone explain GraphRAG with a real use case? Not theory. Not a Medium article/pdf that reads like any LLM wrote it. I want to hear from someone who actually built it in production or any side project what worked, what broke, when it made sense over basic RAG, and when it was just overkill. Builders only. Drop your experience below.

True, Research, Planning, development & making it available for the end users has been solved but now major challenge is distribution where it's super confusing for customers when there are a lot of options through vibe coded applications are available. Now it's hard to differentiate between Secure, legit, reliable and a vibe coded application where no parameter has ben taken care.

I am convinced 2026 is the best time to start a 1 man company.

Add AI-generated PDFs and ebooks to this list. My feed is full of "I wrote a 200-page guide using AI", also don't miss this 5 hours claude course. Bro if your whole book is repackaged LLM output, why would I read it? I'll just ask the LLM myself and get better answers for my specific problem. What made books worth reading was never the crap content. It was someone saying "I tried this, it broke, here's what actually worked." AI hasn't done the work. It's just summarising other people's work. Same vibe as 2015 when everyone said physical stores are dead. They weren't. People still go to offline store to get better experience. Stop selling me AI output in a PDF wrapper and calling it a book.

Nobody wants to read AI-generated books, watch an AI-generated movie or listen to an AI-generated song.

Expensive compute that changed how millions of people work daily. That's not "nothing special" that's infrastructure. But here's the real problem nobody talks about. A solo founder paying $500/month pre revenue on AI APIs to build a micro SaaS that charges $20-50/month or a team of 100 employees where profitability is the major challenge. The math doesn't work yet. You're spending more on the tools than your customers pay you. AI isn't overhyped. It's overpriced for the people who need it most. The day inference costs drop 10x and output quality stays the same, that's when it actually gets interesting. Not AGI. Not sentience. Just cheap, reliable, precise models that a bootstrapped founder can afford without burning runway on API bills. We're not in the AI revolution yet. We're in the AI tax era. The revolution starts when the cost disappears.

people are slowly realizing ai isn’t anything special, it’s just insanely expensive compute

@gregisenberg Funny how we spent decades saying "learn to code" and now the cheat code is "learn to sell." Now one person can spin up a SaaS in a weekend. Getting 5 people to pay for it? That's the part that keeps me up at night. Code is free. Attention/distributions costs everything.

now that we all can build anything with ai we're going to all have to figure out distribution the wealthiest people will be marketers over the next 10 years

Anthropic shipped more features in 3 months than most startups ship in a year. And they're not shipping random stuff. Every release connects Code writes it, Cowork runs it, Dispatch automates it, Computer Use controls it. They're quietly building an operating system for work and most people are still thinking of Claude as "a chatbot."

Dario Amodei in 2025: "In 12 months, we may be in a world where AI is writing essentially all of the code."" Anthropic 2026: - Jan 2026: Claude Cowork launched. - Feb 2026: Opus 4.6 released. - Feb 2026: Sonnet 4.6 released. - Feb 2026: Cowork launched on PC - Feb 2026: PowerPoint integration - Feb 2026: Excel integrations added. - Feb 2026: Co-work plug-ins released. - Feb 2026: Claude Code security launched. - Feb 2026: Claude Code Remote Control - Feb 2026: Scheduled Task in Co- work - Feb 2026: Connector available in the free - Mar 2026: Claude memory is free - Mar 2026: Claude Marketplace launched - Mar 2026: Claude com ambassadors - Mar 2026: Code review for Claude code - Mar 2026: Claude skills for Excel & Slides - Mar 2026: charts & diagram in chat - Mar 2026: 1 million context window - Mar 2026: Dispatch for Claude Co-work - Mar 2026: Claude code Channels - Mar 2026: Co-work Projects - Mar 2026: Claude Computer use - Mar 2026: Auto mode in Claude code.

Every week a new YouTube video says "RAG is dead." It's not dead. You just can't get clicks titling your video "RAG got better and more boring." Basic RAG stuff documents in a vector DB, retrieve top 5 chunks, paste into prompt yeah that approach has limits. Everyone figured that out. But RAG didn't die. It just evolved: →Chunking got smarter. Semantic chunking that respects document structure instead of fixed-size splits. → Retrieval got layered. Hybrid search mixing vector similarity with keyword matching. Re-ranking before anything hits the LLM. → Context windows got bigger but that didn't kill RAG dumping 100k tokens into a prompt isn't a strategy. It's expensive and slow. → Agentic RAG is where it's heading. The model decides what to retrieve, when, and how much. Not a static pipeline but a reasoning loop. The people saying "RAG is dead" usually mean "my naive implementation stopped working." RAG didn't fail them. They outgrew the tutorial version. RAG isn't dead. Lazy RAG is.

Vibe coders bringing ideas to life? That's great. Seriously. More people building is always a good thing. But the moment something breaks in a way the AI can't pattern-match and it will the person who understands how a call stack works, why state behaves the way it does, or what a memory leak actually looks like, is the person who ships while everyone else is on prompt attempt #47. Vibe coding lowers the floor. CS fundamentals raise the ceiling. You need both. But only one saves you at 2am. I am still rooting for basic while non-tech people are forcing more for prompt engineering. use it as a tool, life is good else be ready for unexpected.

Vibe-coders don't give a shit about code. They don't understand it, they don't touch it, and they will never have to deal with it. They care about bringing their ideas to life and nothing else. They are 100% unburden by what many of us consider a showstopper. And that's okay. There's an important lesson here.

Unpopular opinion: the best time to build a startup is while you still have a job. You make better product decisions when you're not desperate for revenue. You say no to bad customers because you can afford to. You build what's right, not what's fast. Everyone glorifies the "I quit my job to go all in" story. Nobody talks about the founders who built quietly for 6-12 months with a salary paying the bills, validated the idea with real users, and only quit when the product was already making money. That's not playing it safe. That's playing it smart. I'm building ConvoEngage right now nights after work, weekends when I can. It's slow. Some weeks I ship one feature. Some weeks I ship nothing. It's more like getting on all wheel to understand the complete market & customers. If you're building something on the side and feeling guilty that you haven't "taken the leap" don't. The leap is overrated. The grind is underrated.

@trikcode Add sales to that and you've got the whole game. Most technical founders can build. Some learn to market. Almost none want to get on a call and close a deal. The ones who do all three don't need a team of 10 to hit their first $100k.

Founders who combine coding and marketing become unstoppable

@Hiteshdotcom AI dying is the most alive topic on the internet right now. Hitesh bhai, AI has died more times than JavaScript frameworks have been replaced. And yet both are still everywhere.😀

Aaj ka video, AI is dead Not joking 😂

Writing code was never the hard part. Writing the RIGHT code was. That hasn't changed. It's actually gotten worse. Now anyone can generate 5,000 lines in 10 minutes and feel productive while creating a codebase nobody can maintain. The hard part in 2026 is the same as 2006 understanding the business logic, designing the architecture, knowing which feature to say no to. AI can write the function. It can't tell you whether the function should exist in the first place. The developers who hand both to AI are building something they'll be debugging for years. Code got easier to write. Systems got harder to think through. That's the strange part.

We have entered a strange phase of software where writing code is no longer the hard part. Very strange.

Ideas were always cheap. Execution was always expensive. That part's changing one person can build what took 20 people three years ago. But now that everyone can build, the hard part isn't shipping. It's getting anyone to notice you shipped. Distribution is the new moat. You can have the best product in the world and still lose to someone with a worse product and a bigger audience. We moved from "can you build it?" to "can you get it in front of the right people?"

The era of jobs is fading - making way for the era of ideas.