Paras Kuhad (@paraskuhad) header
Paras Kuhad (@paraskuhad) profile picture

ad infinitum

Pune
Joined March 2009
Photo by Paras Kuhad (@paraskuhad)Photo by Paras Kuhad (@paraskuhad)Photo by Paras Kuhad (@paraskuhad)Photo by Paras Kuhad (@paraskuhad)Photo by Paras Kuhad (@paraskuhad)Photo by Paras Kuhad (@paraskuhad)Photo by Paras Kuhad (@paraskuhad)Photo by Paras Kuhad (@paraskuhad)Photo by Paras Kuhad (@paraskuhad)Photo by Paras Kuhad (@paraskuhad)
Ab samjha mujh mein aspiration ki kamee kyon hai
Morgan Stanley India MD btw..
19
33
2
361
36,361
I've never found a shortcut while building things that didn't come back to bite me. It usually begins with something minor. Maybe you ship a feature quickly by skipping the spec review, hardcode a value instead of setting up a config, or skip a test because you’re short on time. The reward is instant and feels good. With AI, these shortcuts happen even more often. We end up with more work that’s only ‘almost ready.’ But sooner or later, we have to pay for it, and it often costs even more. The problems don’t always show up immediately. Sometimes everything works fine for months. But over time, things get more complicated. Quick fixes turn into technical debt. You have to explain choices you barely remember, and it takes twice as long to get them up to speed compared to doing it right from the start. Even worse, each shortcut makes the team trust the codebase less. When you know the code is barely holding together, you end up working more slowly. People lose confidence, and progress slows down. Going slow is actually the fastest way forward now.
31
The near future is all about whether your product can work for code, scripts, and agents that automate the real work. MCPs and REST APIs that can solve data queries in your chat agent or within the product UI. MCP and CLI are becoming must-have components for any product. Automation is simply not about integrations. LLM can already talk to other products and orchestrate. The value is in the workflow, not in the presentation of data. If AI-based agents can hit your endpoints, orchestrate actions, and extract data without you being in the loop, productivity rises. That’s how your executives, power users, would like to work.
32
Some of your best work will start as a gut feeling you can't fully explain. In the beginning, that's often all you have. People talk a lot about data-driven decisions and structured validation. Those things matter, and they're non-negotiable once you're shipping at scale. But the first sparks, the ideas you can't let go of, rarely arrive with a spreadsheet attached. Every builder I know has followed ideas without knowing exactly where they came from. The hard part is trusting your instincts until something real takes shape. It can feel uncomfortable. You might not be able to explain it to your team, others involved, or even yourself. But if you ignore that feeling, you often miss out on what could be truly original. Systems and processes can wait. First, you need the courage to act on something that doesn't fit into a presentation or spreadsheet and might not even have a business model yet. Those strong beliefs are your real strengths, no matter how things turn out. At its core, building things is about having the discipline to validate, but also the courage to start when all you have is an intuition you don’t fully understand.
27
Trying long posts on X... Here we go. I still feel tense every time I ask for a deal. Building is where I'm comfortable. Give me code, architecture, a new workflow to design, I’m in my tune. Selling? That means switching gears to frame, persuade, and repeat the story. Every founder who ships knows this tension. But until your idea can support itself, you have to tell its story. You might hope the product will speak for itself, but it won’t. You might think customers will just understand, but they usually don’t. Every feature and every detail needs a story that someone else can believe in. The best tech founders I know aren’t natural salespeople, but they learned to treat sales as just another system to understand. Craft the story, ship it, get feedback, debug, iterate. The job isn’t done when the code compiles; it’s done when someone else cares enough to pay. Yes, it’s uncomfortable. Yes, most of us would rather be building. But that moment you connect the dots for a customer, and they say, “I need this”, that’s a different kind of build. You don’t have to love selling. But you have to do it, or you’re building for an audience of one. What's the trick. Reframe the challenge. Just think you are not connecting to sell, you are connecting with the prospect because they need something you are expert on, and they need help! #founder #sales #builders #startups
40
Hey everyone! Super excited to kick off mitramandal.ai at @One2NC. It's a meetup where we’ll dive into real stories about implementing AI. The goal is to help software engineers see how AI can fit into the systems and projects they already know. Mitramandal creates space for this kind of learning. Engineers sharing implementation experiences - the successes, failures, and everything in between. Stay tuned, we’ll be announcing the date soon!
6
8
6
46
9,046
What’s a well-lived life? To me, it’s a life with minimal friction. One where your nature and your actions are in alignment. One where you crave nothing other than the present moment. One where you’re deeply okay with everything.
45
83
7
827
22,504
Maybe something changed, check back with your dream.
40
A new browser war has begun.
1
1
41
Startups have become really mainstream when the producer of Kal Ho Na Ho brings Akshay Kumar to judge startup pitches In the last 4 years starting with Shark Tank, it has become far easier for me to explain my grandmother what I do for work
6
4
119
5,801
Next stage is recollecting your GST number.
In India, one really starts adulting from the moment you start recollecting your PAN without actually seeing the pan card.
32
.@livspace You got to stop spamming, please! .. There is a limit for outbound.
1
105
Just crazy, got the hit again. Looks like they have bought numbers in bulk. Surprising when TRAI is pushing against spam with telecom operators. Scary. If spammers can pull this job at such a big scale, no wonder so many people are being robbed of life savings.
1
41
I hope this spamming get indexed on search engine, if not then I plan to write a detailed post on medium or somewhere. This needs to stop.
23
Three years later. Never stop dreaming.
316
643
191
18,436
2,363,923
Vibe coding is a loose concept. Writing code has always been a deterministic process. It is the single source of truth for all decisions about what happens, under what conditions, and with what variations. Variations also use data, again deterministic in nature
1
1
46
Unless we understand this part of the process, and if we keep assuming it will give us somewhat (emphasis on somewhat) useful results, I think it’s dangerous. It carries even more risk if the end users are not aware that what they are using was generated from this kind of process
22
It will be a sleepy day when you will make the most progress.
48
Vinod Khosla: “Don’t rely on what the world tells you is possible. Create the world you want.” Vinod points out that in 2010, the Department of Energy projected that there would be 2,305 electric cars on the US market by 2035. In 2014, Tesla alone shipped about 80,000 electric cars. Vinod’s takeaway: “This tells us an important story: they talked to all the experts, and experts only extrapolate the past.” He then shares a few other examples of “experts” who were both overconfident and wrong about the future: • A cartoon from 1889 painting electricity as an existential thread to humans (a narrative that resembles the scaremongering taking place around AI today) • “Lord Calvin, who was president of the Royal Society and maybe the most informed person in science in the early 1900s, said in about 1906 that heavier than air flying machines were impossible, just before the Wright Brothers flew their plane.” • “Fooling around with alternating current is just a waste of time. Nobody will every use it.” — Thomas Edison • “Remote shopping, while entirely feasible, will certainly flop. It has no chance of success.” — Time Magazine He concludes his talk with the following line: “The future belongs to those who dream the dreams & are unreasonable enough to try and make them come true, & who aren’t afraid to fail, & hence take, bold & radical risks.” And he tells the audience: “Don’t rely on experts. Don’t rely on what the world tells you is possible. Create the world you want. I say there’s two types of people: people who extrapolate the past and people who invent the world they want.” Video source: @khoslaventures (2023)
19
100
12
532
41,074