Worthunt

We are building the Data + Infra layer for the next generation workforce: be it teams of up to 10 people or individuals monetizing their skills, building a real audience, meeting clients, handling invoices, and selling services who are going digital-first at work.

@jerkeyray Push the .env file as the thank you present. Live your main intern life

internship gonna come to an end soon so i gotta start looking for something new now :(

Ahh my codex and opencode agents are all fired up now

@AbhijeetWH Larper larping huge

@AbhijeetWH kids after reading a claude generated 5 page PRD for the 5th time

I’m not kidding when I say I got around ~$3k in grants or sponsors, talked with 100+ people if not including comments... did 30+ demos, and went through countless product changes. I even got unicorn startups to partner before launch. I’ve been lying to my family about sleeping, studying, or even eating just to work on this. Now I’m scrapping a big chunk of that idea to rebuild and improve it into something new, after going all in on the knowledge work industry for so long and reading real theses and articles. While others are finalizing their YC applications, I decided to change so much just a day before because I want to solve a bigger problem than what I originally set out to do. This will either be a banger or a big L, but it could also be the most impressive thing in our space. Like… who even am I without my delusional self? I definitely never wanted to launch or say I’m building a startup just for the sake of it. I genuinely love working and solving problems for people i fkn love you all too :))

What "just one more feature" before launch looks like. I ain't explaining anything atp lol just the last messed up UX before final fixes and also a major standpoint pivot of what I wanna build under which Worthunt will be (something like what Anthropic is to claude or Palantir is to ontology but for knowledge work. will post soon!!) In the meantime if you know anyone working in ML for a few years now especially in world models do lmk :))

@AbhijeetWH @dextertwts @handotdev Get back to work!! you wannabe🥀💔

Most definitely, output quality is a function of input quality. If you treat prompts and documentation as variables, then the system begins to behave near deterministically. Better prompts produce better outputs. Better documentation increases the probability distribution of correct responses. What most people underestimate is that English and Markdown are quietly becoming the new programming interface. Not because they are more powerful than traditional languages, but because they operate at the optimal abstraction layer for AI systems. They compress intent efficiently while remaining interpretable. This is where compaction becomes important. The goal is to encode maximum intent in minimum tokens. Well structured Markdown is not just readable, it is a compression strategy. It reduces ambiguity and improves alignment between instruction and execution. In that sense, prompting is less like casual instruction and more like writing a specification. You are not asking the model. You are defining constraints, shaping context, and bounding the solution space. I see this very clearly while building Worthunt. I do not rely on prebuilt SDKs or integrations. Instead, I operate directly with structured knowledge bases and MCP pipelines such as Context7, Exa, and similar retrieval layers. The delta in output quality is non trivial. When the context is well structured and properly compacted, the system can outperform a typical entry level developer in both speed and coherence. Now, there is a common criticism: "the UI feels rough" , the security is not enterprise grade, or the system does not scale like a hundred million dollar startup. That criticism is valid but misapplied. A single AI agent was never meant to represent the global optimum of software engineering. It is not the best possible developer. It is closer to a highly capable individual contributor operating under constraints. If we model this correctly, the comparison changes. It is not AI versus elite engineering teams. The correct baseline is AI plus a human in the loop with foundational knowledge versus a single developer or a very small team. Under that framing, the results are already compelling. You can think of it in terms of parameter sensitivity. Let: Input quality = prompt clarity + documentation depth Context richness = retrieval systems + structured knowledge Human guidance = domain understanding Then output quality is approximately proportional to the product of these variables. This is multiplicative, not additive. Small improvements in each variable compound. Change the parameters, and the outcome scales non linearly. Looking forward two years, this becomes even more pronounced. As context pipelines improve, as compaction techniques mature, and as we see better token efficiency through what can be described as turboquantization of resources, the effective context to token ratio will increase significantly. More signal per token. Less redundancy. Tighter alignment. That means smaller, cheaper, even open models will operate with far higher effective intelligence, simply because they are fed better context. At that point, the performance curve shifts. These systems will not just match current developers; they will surpass a large percentage of them, particularly in the 70th to 90th percentile range. The implication is clear. The skill of the future is not just coding but rather it is structured thinking, expressed through language. English becomes the interface. Markdown becomes the medium. And the real differentiator becomes taste. Taste in architecture. Taste in design. Taste in product decisions. Taste in marketing and distribution. In other words, the new stack is: English + Markdown + Taste

@AbhijeetWH We can make it dynamic later

I see this move as “riding the wave,” particularly when in depth analysis is required. However, I’ve long believed that UX will increasingly shift toward generative, natural language driven interfaces rather than traditional dashboards. That said, a more effective approach today may be to double down on context simulation bringing together dashboards, contextual analytics, and chat interfaces into a unified experience, with chat serving as a seamless, quick access layer across screens. While powering analytics with behaviour knowledge.

@AbhijeetWH Everything but launch 🫠

Did a boilerplate type shi for the agents... having telegram for your app feels like the next boilerplate move for agentic saas. But soon we will see a lot of GUI work transitioning into TUI for agents. This will be a bigger shift than MCPs

Type shi🗣️

@claudeai Chill out big bro

You can now enable Claude to use your computer to complete tasks. It opens your apps, navigates your browser, fills in spreadsheets—anything you'd do sitting at your desk. Research preview in Claude Cowork and Claude Code, macOS only.

Oh yeah finally team workspace is there too not just solo mode. Waiting for OAuth and other approvals for few features and we start making some real money

We’ve just launched the new mobile UI of Worthunt cleaner, faster, and built for a smoother experience. Every detail has been redesigned to make navigation effortless and discovery seamless. Take a look and tell us how does it feel?

The 4 apps you need to run & scale in India by Indian 🗣️🔥 If you’re building from your bedroom, co-working space, or college canteen… AI → Sarvam AI Email → Zoho Mail Payments → Dodo Payments Distribution → Worthunt connect, acquire, distribute, and scale all in one place India isn’t just consuming global software anymore. We’re building the operating layer for the next decade.

In the last 4 weeks ~40 demo beta users who validated Worthunt. Secured 2 unicorn partnerships. Build the auric+circle system benchmark outperformed a major chunk of the distribution and routing market 2nd only to uber's geohashing Scheduled a meeting with one of the biggest pre-seed vcs in the world. A million+ impressions. If you believe in my work do checkout worthunt.com and join it and share it. I do understand I was fumbling in between and I'll improve on it. I'm a born fast paced speaker and sharing it all in 1 min was something that took me a while hopefully I can improve myself on this domain fast🫡 gmi

@AbhijeetWH Gmi🤌🏻🗣️

If there exists even a non zero probability of empowering individuals to build a company of one, I would commit fully to making it inevitable. The future does not belong to bloated organizations competing on marginal improvements; it belongs to singular operators leveraging leverage software, capital, and now intelligence to create disproportionate value. Most people are trapped not by lack of ambition, but by structural dependence on teams, hierarchies, and permission. Break that dependency, and you unlock an entirely new category of professionals. I’m all in on building that infrastructure. Not because it’s incremental but because it’s contrarian and true. A world where one person can build, distribute, monetize, and scale independently isn’t just efficient; it’s inevitable. The incumbents won’t see it coming, because they’re too busy optimizing headcount. We’ll be busy eliminating the need for it. That’s the kind of future Worthunt should underwrite.

The more AI labs talk about replacing people, the more they miss the real picture. This will not happen the way it is commonly perceived. A large share of their revenue comes from the very people whose jobs are said to be at risk. Replacing them would mean undermining their own market. That is not a winning strategy. What will actually change is the quality of work and the level of individual agency. Humans will not disappear. They will become more powerful. Output will increase, leverage will increase, and small teams will outperform large organizations. Freshers will face pressure, no doubt. The bar will keep on rising. You either become a strong generalist who can operate across functions, which works well in startups and product roles, or you become a deep specialist in one domain, which large enterprises will continue to value. Average profiles will struggle. Exceptional ones will compound. At Worthunt, we are building for this future. Our focus is not on replacing humans, but on amplifying them. We want to enable individuals to build a company of one, with real agency, visible work, and ownership of outcomes. Creativity is not replaceable. Judgment is not replaceable. Taste is not replaceable. AI is not here to eliminate humans. It is here to remove friction. Humans who use it well will matter more, not less. Join the waitlist at Worthunt.com

Freelancing has outgrown side hustles. It can now be run like a company of one.

Most freelancers run businesses on scattered tools. Worthunt changes that. At Worthunt, we help freelancers and agencies focus on what matters most and help them become a company of ONE