frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Goes Out to Everyone Trying Their Best and Top Things I Learned

2•AllaTurca•1h ago
Hi everyone,

I'm Anwaar, founder of AllMind AI. My startup is a research terminal for institutional investors, so my experience and this post are going to be tailored to people selling into harder-to-reach audiences (finance, legal, and healthcare). I started my startup after leaving my job at TD (I spent a few years in Finance & ML before starting this company). TLDR: I never realized how difficult starting your own business truly is so I wanted to make this post.

This post is not meant for people who spun up a Lovable vibe-coded project wanting to sell it. A lot of people do that, and it works great, but that's not what I wanted to do. My advice/experience is meant for people aiming to create a real, valuable business who want to hear honest feedback from someone who left a pretty good-paying career to do something crazy and start his own company. So here I go explaining the top 5 things I learned:

1) Get comfortable not knowing what's happening. People who say you should have clarity when starting a business are out of their minds. It's not possible. You learn something new every single day. You struggle, you make mistakes, you pick yourself up and start again the next day. You keep going until something clicks, and then you hit another plateau or dead end. Get comfortable being unsure, uncertain, and not really understanding what's going on. If you keep going, things start to make sense.

2) Sell yourself before you talk about the product or traction, especially to customers. If your customers are from a regulated industry, they need to trust that you, the founder, are the best possible person to lead your company. In the early days, if you can't sell yourself properly to your customers, nothing else will work. Be comfortable telling them the vision you have and where things stand today.

3) Solving for a specific problem doesn't always work the way you think. There are a million different problems people have, and the majority of the annoying ones can now be solved with a Claude Code session in two hours. If your business can be replicated in under one week with Claude Code, think hard about your positioning. When working on a problem, focus on the value you are creating, not just the end output. People have gotten used to getting super high-quality stuff at low cost (again, because of tools like Claude). If you are selling to hard-to-convince audiences, make it known that your tool is genuinely valuable to them. Customers are very aware of what's new in the market so please don't try to fool them!

4) Focus on the UI. If your UI sucks, trust me, it becomes an instant turn-off during demos. At bare minimum, look better than the incumbents even if you're not as good. People can get used to bugs or minor annoyances, but if no one is willing to use the product because the UI sucks, it becomes pointless. Nobody wants a bad UI. If nothing else works, at least make it look presentable in a demo to get interest started.

5) Get comfortable talking to people. Trust me, this was something new for me. As a quant and technical finance guy, I was never an extrovert. You have to build this skill because, funny enough, you are selling to people.

These are my top 5 lessons. My view and experience may differ from yours because of the audience I'm selling to, but I think the majority of founders can relate to at least two or more of these points. Hope this helps someone out there!

Concerns about low-quality PRs beeing merged into main

https://discourse.llvm.org/t/concerns-about-low-quality-prs-beeing-merged-into-main/89748
1•todsacerdoti•2m ago•0 comments

Show HN: Turn text prompts into playable battle pets with stats and skills

https://www.webattle.ai
1•wdkwdkwdk•3m ago•1 comments

A Metal-Detection Objection at the Reception

https://aribrin.substack.com/p/metal-detection-objection-at-the
1•ctoth•3m ago•0 comments

A History of a Novel Idea

https://digitalcommons.cwu.edu/cgi/viewcontent.cgi?article=1006&context=hmvla_jampa
1•andsoitis•5m ago•0 comments

Undifferentiated Phenomenological Continuum

https://alexbierach.com/undifferentiated-phenomenological-continuum/
1•andsoitis•8m ago•0 comments

Decagonal and Quasi-Crystalline Tilings in Medieval Islamic Architecture [pdf]

https://www.physics.rutgers.edu/~pchandra/physics601/Lu_Steinhardt.pdf
1•andsoitis•15m ago•0 comments

A New AI Math Startup Just Cracked 4 Previously Unsolved Problems

https://www.wired.com/story/a-new-ai-math-ai-startup-just-cracked-4-previously-unsolved-problems/
1•battle-racket•16m ago•0 comments

Quitting

https://thepointmag.com/examined-life/quitting/
1•NaOH•17m ago•0 comments

Why is price comparison still so bad in 2026?

1•YourPriceHub•17m ago•0 comments

Trapping Electrons (or: how data is stored)

https://www.makingsoftware.com/chapters/how-is-data-stored
1•MatthewBF•19m ago•0 comments

Causal World Modeling for Robot Control

https://arxiv.org/abs/2601.21998
1•mountainview•19m ago•0 comments

Arthur Brooks: Can We All Relax About Bad Bunny's Spanish Halftime Show?

https://www.thefp.com/p/arthur-brooks-can-we-all-relax-about-bad-bunnys-spanish-halftime-show
4•donsupreme•26m ago•2 comments

Major political upheaval in Japan? Sanae Takaichi may dissolve the Diet

1•jocelyner•27m ago•1 comments

The Logic of State Surveillance

https://www.nber.org/papers/w34492
2•bediger4000•28m ago•0 comments

Moltbook was peak AI theater

https://www.technologyreview.com/2026/02/06/1132448/moltbook-was-peak-ai-theater/
3•spenvo•32m ago•0 comments

Show HN: Claude Code style personal website

https://www.ajwaxman.com/terminal
1•awaxman11•32m ago•0 comments

MRFF says US Military members were 'pressured' by commanders to see 'Melania'

https://www.businessinsider.com/us-military-members-pressured-to-see-melania-documentary-mrff-2026-2
14•hn_acker•32m ago•1 comments

The Original Power Rangers Is Officially Dead After 51 Years

https://screenrant.com/super-sentai-franchise-original-power-rangers-ending-gozyuger/
3•thunderbong•35m ago•0 comments

Sensing methane emissions from African livestock by drone, tower, and satellite

https://bg.copernicus.org/articles/22/4163/2025/
2•PaulHoule•52m ago•0 comments

Are LLMs Becoming Components Rather Than Systems?

https://www.corvic.ai/blog/part-1-of-3-indicting-the-villain-how-the-pipeline-tyranny-created-you...
1•GurbinderGill•54m ago•1 comments

From Interfaces to Intelligence: Where Agentic AI Shines

https://dvitsios.org/2026/02/07/from-interfaces-to-intelligence-where-agentic-ai-really-shines/
1•djif•54m ago•1 comments

An attempt at a First-Proof AI challenge

https://abhvio.us/posts/pony/
1•abhv•54m ago•1 comments

Seedance 2.0 preview: The best video model of 2026, outperforming Sora 2

1•Alisaqqt•55m ago•0 comments

Show HN: We built a tool to track the millions of dollars I ignore during ads

https://twitter.com/k0ba_eth/status/2020650212906418532
1•k0ba•57m ago•0 comments

A Remarkable Achievement (Old Fashioned Software)

https://mcauldronism.substack.com/p/a-remarkable-achievement-old-fashioned
2•nottheg•58m ago•0 comments

Show HN: Faaadmv – check and renew CA DMV registration

https://github.com/AshKash/faaadmv
1•innagadadavida•1h ago•0 comments

IA dans le service client: les 20 cas d'usage à prioriser en 2026

https://nadiaaccompagne.substack.com/p/ia-dans-le-service-client-les-20
1•NadiaAccompagne•1h ago•0 comments

Bounded Rationality

https://en.wikipedia.org/wiki/Bounded_rationality
2•toomuchtodo•1h ago•0 comments

More intelligent agents behave less coherently (2023)

https://sohl-dickstein.github.io/2023/03/09/coherence.html
1•627467•1h ago•0 comments

I found a domain for $1.99 on Vercel and bought 100 of them – this is just one

https://chars.monster/
2•luthiraabeykoon•1h ago•1 comments