Agniva Mahata

No Map

Tue, 10 Feb 2026 00:00:00 +0000

In 2018 I went to Vishakhapatnam for 11th and 12th. First time I left home for real. I was seventeen, I didn’t know what I was doing, and that was fine because nobody around me knew what I was doing either.

When you leave and come back, something breaks. Not dramatically — you don’t have some movie moment. You just start hearing things differently. The stuff everyone at home agreed was true — study, degree, government job — you realize those aren’t laws. They’re just another perspective. From another person. Someone who loves you, who absolutely wants good for you, but who just doesn’t know what else is out there. They’re not wrong. They’re just working with what they have.

And that’s the part that’s on you. Not to resent them for it. To teach them. To show them what you’ve seen. Because they can’t want better for you if they don’t know better exists.

My grandfather was a primary school teacher. My dad is a high school teacher. My mom held the entire family together and nobody ever gave that a title. The path was obvious. Three items on the menu. I didn’t fight it because I had some grand vision. I just couldn’t make myself believe it.

Came back. Did IIT coaching. Failed.

I’m not going to dress that up. I failed. Didn’t get into a good college. And here’s the thing about failing something everyone around you treats as the only thing that matters — it doesn’t free you. It just makes you desperate.

COVID hit. I had a plan to go to Germany. My parents said no, stay in Kolkata. So I stayed. 2020 to 2023. Three years.

First year — COVID. Everything shut. Second year — still mostly COVID. The world was frozen and so was I.

Third year I was depressed. Properly depressed. I was going to college, I was trying to run an agency, I was trying to make money online, I was doing a job on the side because I already knew — I knew in my bones — that college wasn’t going to give me shit. I was doing five things at once and none of them were working and I was twenty-something in Kolkata watching my life not start.

Here’s what nobody tells you about that phase: it’s not that you don’t know what to do. It’s that you’re trying to do everything because the fear of picking wrong is worse than the exhaustion of doing all of it badly.

I have a friend. Ambitious guy, wants to build things, wants to make money, wants to move. His parents want to see him in an office uniform. So he’s trying to do both.

Both doesn’t work. I’ll say that again. Both doesn’t work.

Not because the things are individually bad — they’re not. But they pull in opposite directions. You can’t build something new while performing loyalty to the old thing. You’ll tear yourself in half and then wonder why you’re tired all the time.

There’s a lot you have to unlearn before you can learn anything useful. And we — city people, middle-class people, educated people — we live in this scarcity mindset where starting from zero feels like dying. Like if you let go of the thing you already have, even if the thing is garbage, you’ll fall and never get back up.

I’ll tell you something. In my entire life I have started from zero more times than I can count. Every single time. I move to a new city — zero. I try a new thing — zero. I leave something that isn’t working — zero. And every time it feels like the end and every time it isn’t.

Everything is uncertain. That’s not a motivational poster. That’s just the reality. The sooner you stop pretending you can control it, the sooner you can actually move.

Third year of college. Bangalore was the only thing that made sense. I had one college there I could transfer to, and honestly the college was an excuse. I needed to leave. Kolkata was killing me slowly and politely and I was letting it.

So I quit. Quit college. Moved to Bangalore.

Grandfather: teacher. Father: teacher. Son: dropout who moved to a city where he knew nobody. That sentence doesn’t go over well at family dinners. But here’s the thing — it was my sentence to write. Not theirs.

People love to talk about how someone else’s decision ruined their life. My parents said no to Germany. My college didn’t teach me anything useful. The economy was bad. COVID happened. All true. All irrelevant. Because at some point you have to look at your own life and say: okay, but what am I going to do about it?

That’s the whole thing. That’s the entire point. There is no map. There is no path. There is no sequence of correct decisions that someone can hand you. It’s your life and you fundamentally have to decide. Not your parents. Not your teachers. Not the economy. You.

And that’s terrifying. I know. I’ve been terrified every time. But the alternative is what? Sitting in Kolkata being depressed and running an agency that doesn’t work while attending a college that doesn’t matter? That’s not safe. That’s just slow.

Bangalore didn’t fix anything. Let me be clear about that. I didn’t show up and suddenly everything clicked. I showed up and I was broke and confused and starting from zero again. The only difference was I had chosen it.

That matters more than you think. When the suffering is something you chose, you can work with it. When it’s something that happened to you, it just eats you.

I kept building. Said yes to everything. Worked for free — not because I was being strategic about building trust or whatever LinkedIn nonsense people say. I worked for free because I genuinely could not bring myself to send an invoice. I built entire systems for people and then couldn’t say “pay me.” I still can’t pitch. I can build you everything. Top to bottom, the whole thing. Ask me to sell it and I become a different person. Smaller.

But here’s what happens when you keep showing up. People don’t talk about how talented you are. Nobody’s saying that. They say: he delivers. He doesn’t bullshit. He shows up. And slowly — in a way you’ll never see happening — trust forms.

Trust is the only real currency. I didn’t learn that from a business book. I learned it from the guy who ran a shop in my neighborhood growing up. Never cheated anyone. Didn’t have a brand. Didn’t optimize anything. The whole area trusted him with their lives. He had more real capital than most startup founders I’ve met. Just a different kind.

The internet was my only window. The only thing that showed me the menu had more than three items. I found out product management existed from a YouTube video. That people got paid to think about systems. That there were entire careers nobody in my zip code had mentioned.

When you go from three options to infinite options, you go a little insane. You try everything. Every new thing feels like the thing. You start a hundred projects. Finish three. I did this for years. It looked like chaos from outside. From inside it just felt like my life.

What kept me going wasn’t talent. It was stubbornness. The same stubbornness that made my grandfather walk to school every day for forty years. The same thing that made my mom hold everything together when nothing was designed to help her. I inherited the refusal to stop. I didn’t choose it. I just have it.

I don’t have a map. I’ve never had a map. The map everyone else was using was drawn by people who could only see what was in front of them. Not malicious. Just local.

Nobody told me I couldn’t go this fast. So I didn’t know I wasn’t supposed to.

Some days I’m on a call with people in different countries talking about products that handle real money and I think — I dropped out of college. I can’t pitch. Five years ago I had nothing. How is this real?

I don’t know. I don’t think it matters. The game doesn’t care about your backstory. It cares about whether you show up.

I show up.

The decisions are the thing. Not the outcomes, not the circumstances, not the luck. The decisions.

Every time I decided — leave Vishakhapatnam, fail IIT, stay in Kolkata, quit, move to Bangalore, build for free, say yes when I should’ve said maybe — every single one of those decisions was mine. Some were smart. Some were stupid. Most were just desperate.

But they were mine. And that’s the only thing that made the difference.

You’re reading this and maybe you’re in your version of Kolkata. Stuck somewhere, doing five things badly, waiting for permission to start your actual life. Nobody is going to give you that permission. The map doesn’t exist. The path isn’t real. The sequence of safe decisions that guarantees everything works out — fiction.

Your life is yours. Decide something. It’ll probably go wrong. Start from zero again. It’s not the end. It never was.

SaaS 3.0: The End of Faster Horses

Sat, 12 Jul 2025 00:00:00 +0000

For twenty years, the SaaS playbook has been embarrassingly simple: move data to the cloud, wrap it in a dashboard, and charge by the seat. Salesforce, HubSpot, and their 10,000 copycats built billion-dollar empires by merely showing us what was happening inside our pipelines. They were mirrors, not muscles.

And then AI arrived—not as a feature, but as a rupture. The gap between insight and action, once bridged only by human hands, can now be closed by machines that do. We are not iterating on SaaS; we are graduating from it. Welcome to SaaS 3.0: software that doesn’t just inform, but executes.

The Old Stack: Information → Decision → Work

Traditional SaaS sits at the first two layers. It collects data (CRM fields, analytics events, support tickets) and visualizes it so a human can decide what to do next. The third step—actual work—is outsourced to people. More leads? Hire more reps. More tickets? Add more agents. The business model is linear: one seat equals one unit of labor.

This is why Semrush tells you which keywords are “opportunities” but cannot write the article, publish it, or build the backlink. It’s why your CRM flags “10 leads to follow up” but cannot send the email, research the prospect, or book the meeting. The software ends where the work begins.

The New Stack: Insight → Autonomous Action → Outcome

SaaS 3.0 collapses the stack. AI doesn’t just surface insights; it acts on them. The same CRM now:

Researches each lead (LinkedIn, earnings calls, tech stack).
Drafts hyper-personalized follow-ups.
Sends them at the optimal time.
Books meetings directly into your calendar.
Tracks replies and loops back to step 1.

The human role shifts from doing to orchestrating. You don’t write the email; you approve the strategy. You don’t chase leads; you decide which segments the AI should prioritize. The pricing model flips: you pay for outcomes (meetings booked, articles ranked, tickets resolved), not seats. One user can now orchestrate an army of agents.

The Execution Gap: Where the Real Opportunity Lies

Most founders today are using AI coding tools to build faster horses—traditional SaaS in days instead of months. They optimize for the wrong variable: speed of construction, not magnitude of value. It’s like strapping a rocket engine to a bicycle. The real opportunity is software that wasn’t possible yesterday:

SEO agents that don’t just audit sites but publish 500 optimized articles and build backlinks at scale.
Legal agents that don’t just store contracts but negotiate redlines and file trademarks.
Sales agents that don’t just score leads but close them end-to-end while you sleep.

The winners won’t be those who moved fast, but those who recognized that software is no longer a tool—it’s a digital employee.

The Transition: AI SaaS as a Bridge

SaaS 3.0 isn’t a binary leap; it’s a gradient. Start with AI SaaS: software that still asks for permission but does the work. Your SEO system today researches keywords and publishes posts, but you approve the calendar. Over time, as data accrues and trust grows, the system graduates to SaaS 3.0: autonomous loops where human oversight is strategic, not tactical.

This is why outcome-based pricing works from day one. If your SEO agent generates 100 ranking articles in a week, charging $5/article is trivial compared to hiring five writers. The customer pays for the delta, not the dashboard.

The Mindset Shift: From Builder to Employer

Stop building software that helps people work. Start building software that does the work. The question is no longer “What features should we add?” but “Which job roles can we absorb?” The future belongs to founders who treat AI agents not as features, but as staff.

In ten years, the distinction between software that shows and software that does will feel as quaint as the difference between online and offline businesses does today. The only question left will be: how much agency are we comfortable granting our digital employees?

Build for that future. Or keep building faster horses.

Sensitive to Small Moves

Sun, 05 Jan 2025 00:00:00 +0000

Fire. Running water. Wind through trees.

We can stare at them for hours and not know why. Something holds us there. I think I know what we’re actually looking at.

In chaos theory, systems following deterministic rules can become wildly sensitive to tiny differences in starting conditions. Two leaves drop into the same stream, inches apart. One washes up on the left bank. The other travels two miles downstream and gets caught in a drain. Same water. Same rules. Completely different destinations.

This property has a name in physics. I want to give it a name in life.

Liminal sensitivity. The moments when your trajectory is most responsive to small inputs. When the old structure has dissolved and the new one hasn’t formed yet. When a single conversation, a single “yes,” a single person can bend the entire arc of what comes next.

Most people run from these moments. The ones who change are the ones who stay.

In 1961, Edward Lorenz was running weather simulations at MIT. He re-entered a starting value as 0.506 instead of 0.506127. Tiny difference. Three decimal places. The simulation produced a completely different weather pattern. Not a small deviation. A different world.

He called it sensitive dependence on initial conditions. The public called it the butterfly effect. Most people think that’s a metaphor. It’s not. It’s a mathematical property of deterministic systems. The rules are fixed. The outcomes are not. Past a certain horizon, prediction becomes impossible. Not because the system is random. Because it is exquisitely, structurally sensitive.

This isn’t a physics essay. This is about when it happens to you.

There’s a word anthropologists use for the space between. Liminality. From the Latin limen: threshold. The doorway. You’re not in the old room anymore. You’re not in the new one yet. You’re standing in the frame.

No label. No title. No clean answer to “what do you do?”

This is when your life becomes a chaotic system.

When you’re stable, your trajectory is locked. Job, identity, routine, paycheck, address. You’re a boulder rolling downhill. Massive inertia. A conversation can’t move you. A chance meeting can’t redirect you. Small inputs bounce off. The path is set.

But in the liminal phase, you’re a leaf in a stream. Everything is sensitive. One conversation changes a career. One meeting changes a city. One “yes” to something that makes no sense rewrites the next five years.

The math is the same. The system is the same. The difference is where you are in it.

Most people hate liminal space. The ambiguity. The not-knowing. The way it makes your chest tight at 2 AM when you can’t explain your life in a sentence.

So they skip it.

Stable job to stable job. Known identity to known identity. One relationship into the next without breathing. They grab the first thing that looks like solid ground and hold on.

This is why most people never really change. Not because they lack talent. Not because they lack opportunity. Because they collapse the uncertainty before chaos can do its work. They exit the doorway before the sensitivity peaks.

This pattern repeats across careers, relationships, cities, companies. The people who transform are never the ones who made the safest move. They’re the ones who tolerated the liminal space long enough for a small input to find them.

A cannabis venture that went nowhere. Months of work, connections, planning. It didn’t fail. It just didn’t converge. At the time it felt like wasted motion. Looking back: a data point. A calibration.

A company with Sharique, my best friend since we were kids. By December 2024, we were two weeks from dead. Burn rate eating through what was left. No clear path forward. Pure liminal space. Every instinct screaming to grab something stable.

Then a single meeting. Ben van Sprundel. One conversation that didn’t just save the company. It changed the city we operated in. It changed who we were building for. It changed everything downstream in ways we couldn’t have predicted and didn’t try to.

One “yes” to something that made no sense at the time. One small move during maximum sensitivity.

Steve Jobs said you can’t connect the dots looking forward. Chaos theory says the same thing, but with math. The dots connect into shapes you never would have drawn on purpose. You only see the pattern from the other side.

Here’s the thing about chaotic systems. Even the wildly unpredictable ones orbit around certain shapes. In physics, these are called strange attractors. Plot a chaotic system over time and it looks like noise at any given moment. Zoom out, and the shape is unmistakable. A butterfly wing. A spiral. A pattern that was always there but invisible from inside any single loop.

Five years of apparent noise. Failed ventures. Dead-end projects. Cities that didn’t work out. Ideas that landed nowhere.

But orbiting the same things the entire time. Building. Operating. Solving problems for real people. Getting obsessed with how things actually work, not how they’re supposed to work on paper.

The attractor was always there. You just can’t see it from inside the doorway. You have to trust that the orbit has a shape even when every data point feels random. Because it does. The math guarantees it.

This isn’t philosophy. It’s operating strategy.

Liminal phases are rare. Most of life is the stable part, the boulder rolling, the locked trajectory. But when you’re in one, the rules change. Small moves compound harder than they ever will during stable phases. The sensitivity is real and it’s temporary.

Which means three things:

Freezing is the worst response. You’re wasting the only period where small inputs actually matter. Sitting still during maximum sensitivity is like having the wind at your back and refusing to move.

Grabbing the first exit collapses optionality. The desperate lunge for stability feels like progress. It’s not. It’s choosing a trajectory before the system has shown you where it wants to go. You lock in too early and spend years course-correcting from a premature decision.

Staying open is the strategy. Saying yes to things that don’t make sense yet. Talking to people outside your current world. Making small moves in directions you can’t fully justify. Not because you’re being reckless. Because you understand that during liminal sensitivity, the small move that changes everything won’t announce itself. It’ll look like a normal Tuesday.

The fire. The running water. The wind.

Maybe we stare because they’re mirrors. Constant transition. Constant sensitivity. Movement that never settles but holds a shape over time. Chaos that isn’t chaos at all, but a system too complex to read from the inside.

The liminal space isn’t the gap between your real life. It is your real life. The doorway isn’t the obstacle. It’s the instrument.

Chaos isn’t the enemy. It’s the operating system.

Stay there a little longer than feels comfortable.

Dev Log: Stateful Agents and the RAG Problem

Thu, 02 Jan 2025 00:00:00 +0000

I’ve been spending a lot of time thinking about the limitations of current Retrieval-Augmented Generation (RAG) systems, particularly when we try to build AI agents that feel truly useful. RAG’s core concept is elegant: a retriever finds relevant documents from a knowledge base using Vector Search, and a generator an LLM embedding model, synthesizes that information into a coherent response. This is a solid way to ground LLMs in facts and minimize those hallucinations.

However, there’s a fundamental issue that keeps nagging at me: RAG feels inherently stateless. Each interaction with the system is like a brand new conversation. We can try to work around this by feeding previous conversation turns into the context window, but that’s just a textual bandage. It’s not real state. The agent has the memory of a goldfish, forgetting everything once you open a new chat or the context window is full [when working with local LLMs], and perhaps even more critically, forgetting why it retrieved specific documents in the first place.

The core problem, as I see it, is that RAG treats all knowledge as either completely static – the unchanging documents in the vector database – or entirely ephemeral – the current conversation history. This black-and-white view misses a crucial category of information, especially in enterprise settings: bounded ephemeral knowledge.

Consider active support tickets. A ticket might be open for hours or days, representing critical information for a support agent during that time, but eventually becoming irrelevant once resolved. Or maybe real-time system alerts. A sudden spike in server latency is incredibly important right now, but its significance diminishes as the system returns to normal. Consider, too, ongoing sales negotiations. The specifics of a current deal are vital for the sales team today, but those details become much less important once the deal is closed, whether won or lost. Even Slack or Teams discussions, particularly those tied to a specific project or product launch, fall into this category.

These diverse examples all have a common thread: the information is highly relevant for a defined period, and then it should gracefully disappear or be archived. Typical RAG implementations struggle with this. You could, in theory, constantly update and delete entries in your vector database, but that quickly becomes a data management nightmare. You also lose the crucial context of why a particular piece of information was relevant at a specific moment.

The solution I’m exploring is a concept I’m temporarily calling NEBULA: Neuro-Ephemeral Bounded Universe for Language-based Augmentation. The fundamental idea is to explicitly model and manage this bounded ephemeral knowledge, alongside the traditional “persistent” knowledge that enterprises already maintain.

The architecture I’m visualizing starts with an Ephemeral Ingest Layer. This acts as the entry point for all the real-time, streaming data – everything from Slack messages and system logs to CRM updates and transaction events. It leverages lightweight Transformers, or perhaps even simpler embedding models, to vectorize these snippets of information as they arrive. Importantly, this layer also incorporates filters. These can be rule-based or learned, and their job is to determine which ephemeral signals are actually important enough to be passed on to the rest of the system. We don’t need to index every single chat message or log line.

Next comes the Bounded Universe Index. This is a separate vector database, purpose-built for ephemeral knowledge. It has a few key characteristics. First, it supports Time-to-Live (TTL) or decay policies. Every entry has a built-in expiration date. This might be a strict TTL, like “delete after 24 hours,” or a more nuanced decay function, where the relevance score gradually diminishes over time. Second, it’s rich in metadata. Each entry includes information about its source, its time-sensitivity, an estimated “weight” indicating its importance, and potential links to entities in the enterprise’s main knowledge graph.

[Ideally imagine it to be like Redis but for Vector Database]

The Enterprise Knowledge Graph and Persistent Vector DB remain, representing the traditional knowledge store. This holds the stable, long-term information about products, policies, historical data, and so on – essentially what you’d find in a standard RAG system.

The real magic happens in the Neuro-Ephemeral Orchestrator. This is the “brain” that connects everything. When a query arrives, the orchestrator doesn’t just retrieve information from the persistent database. It also identifies relevant ephemeral items from the Bounded Universe Index. It then creates composite embeddings. These merge the embeddings from the persistent and ephemeral sources into a single representation, capturing both long-term knowledge and the immediate context. This merging could involve weighted averaging, concatenation, or even a more advanced learned fusion technique. The orchestrator also dynamically weights the relative importance of ephemeral versus persistent information. During a major system outage, for instance, ephemeral alerts would be given significantly more weight.

An Adaptive Summarization and Memory Decay component acts as the system’s housekeeper. It regularly scans the Bounded Universe Index, performing two crucial functions. It summarizes. If an ephemeral item has demonstrated lasting relevance – for example, a support ticket that revealed a bug – it’s summarized and potentially archived into the persistent knowledge base. And it decays or prunes. Items that have expired or fallen below a relevance threshold are removed, keeping the Bounded Universe Index lean and performant.

Finally, the Generative/Reasoning Layer, where the LLM resides, receives these composite embeddings from the orchestrator. This allows the LLM to reason about both long-term knowledge and the current ephemeral context. This layer might consist of a single LLM or a collection of more specialized agents.

This approach, I believe, offers several significant advantages over standard RAG. Ephemeral knowledge is treated as a first-class citizen, not an afterthought or a clumsy workaround. The merging of ephemeral and persistent knowledge happens at the embedding level, rather than simply by concatenating text, leading to a more nuanced and powerful representation of context. The system is inherently bounded and incorporates decay, preventing data bloat and ensuring the LLM isn’t misled by outdated information. And the orchestrator can dynamically adjust the context based on the situation, making the agent more responsive to real-time events.

My next steps involve tackling some significant, but exciting, challenges. I need to prototype the Bounded Universe Index, likely starting with Redis and its TTL capabilities, while experimenting with various decay functions and metadata schemas. Developing the orchestrator logic is the most complex task, requiring me to determine the optimal way to merge embeddings and dynamically adjust their weights, possibly starting with simple heuristics and progressing to learned approaches. I plan to test this with a real-world use case, probably a simulated IT operations scenario with streaming system logs and a knowledge base of past incidents. I also need to determine the best way to represent the Composite embeddings.

I anticipate encountering plenty of obstacles, but that’s inherent in the process. The ultimate goal is to create agents that are not merely knowledgeable, but also genuinely aware of the present, and capable of gracefully forgetting information when it’s no longer relevant. I believe this is a crucial step towards building truly useful AI systems for enterprise applications.

How to Get Rich in Your 20s: Understanding Value & Leverage

Tue, 17 Sep 2024 00:00:00 +0000

I never asked my parents for money after I turned 18. I started making my own, spent it all, went broke, started something, went broke again, started something else. This happened more times than I want to admit.

I’m from Jhargram. A tier-3 village in West Bengal that most people in India haven’t heard of. I failed to get into IIT. I spent years in Kolkata feeling like I was drowning in slow motion. I moved to Bangalore with no plan and no safety net. I built BenAI with Ben van Sprundel. I worked with Greg Isenberg on Boring Marketing. I started from zero so many times that zero started feeling like a familiar address.

But somewhere in that mess I figured out a thing. Not a trick, not a hack. A thing. The kind of thing that once you see it you can’t unsee it, and once you can’t unsee it you start making money in ways that feel almost unfair.

It comes down to two words: value and leverage. That’s it. Everything else is decoration.

Value is simple. Someone has a problem. You solve it. They pay you. The end.

But here’s what nobody tells you: the size of the payment has almost nothing to do with how hard you work. I’ve worked 16-hour days and made nothing. I’ve spent two hours setting up a system for someone and made more than I did in three months of freelancing.

The difference wasn’t effort. It was what I was solving and for whom.

When I was freelancing in Kolkata, during those depression years where every day felt the same shade of grey, I was solving small problems for people who didn’t have money. I was trading time for rupees. There’s a ceiling on that and you hit it fast and it hurts when you hit it because you’re already exhausted.

Then I started solving expensive problems for people who had money. Not harder problems. Not more complex problems. Just expensive ones. Everything changed.

A creator with 100K subscribers who doesn’t know how to monetize? That’s an expensive problem. A company bleeding money because their systems are manual? Expensive. These people will pay you real money because the cost of not solving it is higher than whatever you charge.

That’s value. Not what you can do. What you can solve, and for whom.

Leverage is the multiplier. It’s what separates the person making ₹50K a month working 80 hours from the person making ₹5L working 30.

There are four kinds and they stack.

Skill. This is where everyone starts. You learn something. Code, design, sales, whatever. And you trade that skill for money. Fine. But skill alone is a job. Skill alone means you stop working, you stop earning. Skill is necessary but it’s not enough.

Time. Not working more hours. Working on the right things. I wasted years doing things that felt productive but weren’t. Entire months in Kolkata spent “being busy” while going absolutely nowhere. The moment I started asking “is this the highest-value thing I could be doing right now?” and genuinely asking, not performing the question for myself. I started moving faster than people who worked twice as hard.

Money. Once you have some, you can make it work for you. But honestly, in your 20s, spend it on experiences and experiments. Every rupee I spent on trying something new compounded in ways I couldn’t have predicted. The ₹10K I spent going to that event where I met the person who introduced me to the person who changed my trajectory? That’s a return no stock market can match.

Distribution. The internet is the great equalizer. One piece of content can reach a million people while you sleep. One system you build can serve a thousand clients simultaneously. Distribution means your output is no longer limited by your input. This is where it gets unfair.

Stack two of these and you’re doing well. Stack three and you’re dangerous. Stack all four and you’re building something real.

Now here’s the part I actually want to talk about. The part that’s specific to right now, to this moment, to you reading this.

AI broke the game open. But not in the way most people think.

I keep seeing people call AI “a fifth type of leverage.” No. That misses the entire point. AI is not a new category you add to the list. AI is the adoption layer. It’s the thing that makes all four existing leverages suddenly accessible to people who never had access before.

That distinction matters. A lot.

If AI were just a fifth lever, it would be one more thing to acquire. One more skill to stack. But that’s not what’s happening. What’s happening is that the four levers that used to be locked behind years of experience, or capital, or connections, or audience. Those locks are breaking. AI is the thing breaking them.

Let me be specific.

Skill leverage used to take years. You’d spend a decade getting good enough at something to charge real money for it. I know this because I lived it. Now someone with six months of focused learning and the right AI tools can deliver work that used to require that decade. Not because the AI does the work for them. Because the AI collapses the gap between what they know and what they can deliver. It’s like having a senior partner looking over your shoulder who never gets tired and never judges you for asking basic questions.

Time leverage used to require hiring people. You’d need assistants, employees, systems, all of which cost money you didn’t have. When I was building things in Kolkata, I couldn’t afford to hire anyone. I was the entire team. Now AI handles the manual parts. The research, the first drafts, the data processing, the repetitive stuff that used to eat your entire day. What’s left is the high-value thinking. The decisions. The things that actually move the needle. One person with AI tools now has the time leverage that used to require a team of five.

Money leverage used to require capital. You needed money to test ideas, build products, run experiments. Now you can prototype in a weekend. You can test a business idea for nearly zero. You can build a working product before spending a single rupee on development. The cost of experimentation collapsed overnight. The money barrier, the thing that kept people from even starting, barely exists anymore.

Distribution leverage used to require an audience or a budget. You’d need followers, ad spend, connections. Now AI helps you create content at scale, understand what resonates, build systems that reach people automatically. One person can now build distribution that used to require a marketing team. I’ve seen this firsthand building content systems for Boring Marketing with Greg. The game changed completely.

So AI didn’t add a new lever. It lowered the barrier to entry on every existing one. And that’s way more powerful than a new category. Because it means someone starting from zero today, someone like I was in Jhargram or during those lost years in Kolkata, has access to all four leverages in a way that wasn’t possible two years ago.

The levers were always there. AI just made them reachable.

Every time the market shakes, every time there’s a massive technological shift, a window opens. The window doesn’t stay open forever.

It was open when the internet started. It was open when mobile happened. It was open when crypto hit. Each time, a bunch of nobodies walked through the window and came out the other side with money and freedom. Not because they were smarter. Because they were there, and they moved.

The AI window is open right now. Wide open. And the thing about this particular window is that it doesn’t just help you build faster. It fundamentally changes who can play the game.

The 45-year-old executive running a company? He can’t keep up with this. He’s in meetings all day. He doesn’t know what these tools can do. He doesn’t know that the thing his team of five is doing manually can be automated in an afternoon by someone who understands the tools.

You do. Or you can.

And that gap between what you understand and what he doesn’t? That’s not a small advantage. That’s asymmetric leverage. The kind that lets a kid from a village nobody has heard of compete with people who have ten times the resources.

I’m not saying this theoretically. I tested this.

I have two friends. Both their names start with Y. Both were at zero. No real income, no clear direction, the kind of stuck where you wake up and the day feels heavy before it starts.

I took one fundamental belief and tested it on them: if you understand value and leverage, and you have access to AI tools that collapse the barriers on every lever, you can make money. That’s the whole thesis.

One of them is now doing $10,000 a month. From zero. In months, not years.

The other one was harder. Honestly, I thought he was beyond repair. The kind of stuck that feels permanent. He’s making progress now. Not $10K yet. But moving. Actually moving, for the first time.

I didn’t give them money. I didn’t give them clients. I gave them a framework. Value and leverage. Then I showed them how AI tools collapse the barriers on each lever. How skill becomes accessible faster. How time opens up. How money stops being the prerequisite to starting. How distribution becomes possible for one person.

That was the entire intervention.

The markets are shaky. The economy feels uncertain. AI is eating jobs. Half the internet is panicking.

Good.

When everything is uncertain, the person who moves with intention has an absurd advantage over the person who freezes. Most people freeze. That’s not a judgment. It’s just what happens. Fear is the default.

But if you can look at a shaky market and see opportunity instead of threat. If you can see that the chaos is just the old structure rearranging itself and there’s money lying on the floor during the rearrangement. Then you’re already ahead.

I come from a village. I’ve started from zero more times than I can count. I don’t have a degree that matters. I failed the exam that was supposed to be my ticket out. I spent years in a city that was slowly crushing me before I figured out how to leave. I still can’t send invoices without feeling weird about it.

But I understand value and leverage. And I understand that right now, this window, this moment, is the easiest it has ever been to build something from nothing.

Not easy. Easiest it’s ever been. There’s a difference.

It’s still going to be hard. You’re still going to fail. You’re still going to feel like you’re drowning. But the tools exist now. The access exists now. The barriers that used to keep people like us out are collapsing in real time.

The only thing that’s missing is you deciding to move.

Being rich isn’t a number. I’ve met people with more money than I’ll ever have who are trapped. Trapped by their companies, their obligations, their image. That’s not rich. That’s just expensive.

Rich is being able to wake up and decide what to do with your day. Rich is having options. Rich is not being afraid of starting over because you’ve done it so many times that zero doesn’t scare you anymore.

You get there through value and leverage. You get there faster right now because AI just collapsed every barrier that used to keep people out. Not by being a new lever. By making the existing levers reachable for anyone willing to learn.

And you get there at all because at some point you stop reading essays about getting rich and you actually go do something.

The window is open. Walk through it or don’t. But don’t pretend it isn’t there.