DEV Community: Harsha Kumar

Why I built a page with no immediate growth upside, and shipped it anyway

Harsha Kumar — Thu, 16 Jul 2026 10:05:42 +0000

XEdge has a discovery problem that's already solved. Describe a goal, get a stack, done. What it didn't have was a reason for anyone to come back after that first answer.
A friend pointed this out plainly a few weeks ago, and independently, someone in a community I'm active in asked almost the same question from a different angle, is there any incentive to keep using this. Two people, no coordination, same gap named.
The fix didn't need to be complicated. Real builders, their actual AI stack, one short quote, a photo, a permanent page that grows over time. No login system, no notification engine, no new infrastructure.
It's live now. First person featured is the friend who suggested it, which felt like the only honest way to open something built on someone else's idea.
xedge.tech/builders, and if you want to be one of the next few featured, this post is an open invitation.

The newsletter stack that gets you to your first $50 — no audience required

Harsha Kumar — Wed, 15 Jul 2026 08:47:06 +0000

Most "start a newsletter" advice skips the order of operations. Here's the exact one, five tools, in sequence:
Perplexity for niche validation — search actual complaints in your space, not trend reports. Claude for the first five issues, written in one batch sitting, not live day-by-day. Beehiiv free tier for the platform — zero cost to start. One insight post daily for growth, subscribe link in the first comment so the post itself reads as value, not a pitch. First $50 from one manually-sold sponsored slot, before any subscriber threshold, because the first sale validates the model faster than waiting for scale.
$0 total cost. First subscriber same day you start.
Full playbook with every prompt used at xedge.tech.

What actually matters when someone asks you to "build a chatbot" for their business

Harsha Kumar — Sat, 11 Jul 2026 07:52:00 +0000

Got a real lead this week — small business owner wanting an AI chatbot for his site. Here's the honest breakdown of what that actually requires, since "chatbot" means five different things depending on who's asking.
Scope before building. Most small businesses don't need a custom-trained bot. They need their top 10-20 repeated customer questions automated well. Find that out first.
Platform over custom build. Tidio, Crisp, Intercom's lower tiers — pre-built AI chat widgets connect to a knowledge base fast. Custom builds are rarely worth it below a certain scale.
The bot is only as good as its input. Feed it real past customer conversations, not generic FAQ copy. Specificity is what makes it feel useful instead of annoying.
Always build the human handoff. A bot that traps a frustrated customer in a loop does more damage than having no bot. This is non-negotiable.
Test before launch. Twenty real customer-style questions, see what breaks, fix it before it's client-facing.
Full customer support AI stack at xedge.tech.

The no-code SaaS stack that actually gets you to a paying user

Harsha Kumar — Thu, 09 Jul 2026 05:32:28 +0000

No-code SaaS consistently surprises founders because the barrier to a real, working product is genuinely gone now. Here's the stack:
Validation: Claude — before touching a builder, describe the idea and ask Claude to steelman the 5 biggest reasons it fails. Cheapest possible filter.
Building: Bubble.io — real database, real auth, real workflows, no code. Free tier is enough to validate and launch a basic version.
AI features: OpenAI API via Bubble plugin — the unlock most no-code founders skip. Connects in 20 minutes, instantly makes the product feel more capable than it looks.
Payments: Stripe + Bubble — charge from day one, even a token amount. Paying users tell you the truth; free users tell you nothing.
Distribution: show up where your users already are — solve a real problem in a real thread first, link second.
Total cost: $0. Time to live product: one weekend. Time to first paying user: 2-4 weeks with real effort.
Full playbook at xedge.tech.

this is how I started Xedge from 0

Harsha Kumar — Sun, 05 Jul 2026 03:53:32 +0000

I'm Harsha and i'm from india , once i'm in college i was obsessed with this ai tool market when i tried to use them so many questions ,

which tool is better ?

-does hype really matches that output ?

is there any underrated tool that i'm missing

i just become from obsessed to overwhelmed and then i got an idea like what if there is a guide to tell me , but when i clearly did research there are already a bunch of tools directories out there like the futurepedia , there is an ai for that , etc . i noticed that all they do is like dumb a ton of tools and describe about them . that doesn't make any sense to me . i asked myself " is there a better way of organisation rather than dumping tools ?". then i thought about an ai ecosystem , where they'll get ai tools , stacks , professional prompts , automations , templetes , agents , api's , and i even imagined a interface within it where they can build their custom software by using open sources . they are not only going to buy the templetes from others they can build and sell them too . simply a places where people can access everything related to ai at one place where you can buy , sell and create custom for your own .

i think about it like an ai paradise . but at that point i really don't know how to code , i only know the c language . then just did some research got the elite sources to learn html , css and js i got the documentations in 1 tab and the videos in another and opened vs code for practice . in less than a month i've learned css , html , java script , python and MYSQL .

Then I heard about ai open source high end models writing code for free . Im like " damm i have to use this free 1 time usage " i explained almost everything related to xedge i imagined about and it just gave me a single vibe coded home page , i'm very dissapointed at that point because i thought it might build a whole website for me . then my inner voice kicked in and told me " you have the skills , just do it " , i used that same colour pallete generated by ai and created like 4 pages and learned a lot of stuff in that building phase , and it became very easy but time taking process , then i get to know about claude , gpt , and a bunch of other tools .

I used them to write code faster and after it done doing things i'll check all the code based on my knowledge and do corrections if i want , that is how i built the my ideas first prototype .

then i increased my connections mastered marketing certainly the organic marketing and the SEO game , i did a lot of research and learned a lot of stuff , the on i've been grinding on gaining knowledge among tech side and distribution side while building the product , i put all my efforts and now i have over 600+ users in my platform and i spent almost 0$ ( except for the domain ) .

even though i havent made a single penny from my skills or from my product , I'm really proud to myself for doing all this in less than 3 months and starting from 0 and spending 0 . And i believe i could really get my xedge to its final phase and gnerate the real revenue ,

How to validate any startup idea in 48 hours with $0

Harsha Kumar — Thu, 02 Jul 2026 08:18:10 +0000

Most founders skip validation because it feels like delaying the real work.
Validation IS the real work. Building the wrong thing for 6 months is the delay.
Here's the exact process:
Find real complaints: Perplexity
Search for people already expressing the problem you're solving — Reddit, review sites, forums. Copy the exact language they use. That's your marketing copy and your proof of demand in one step.
Size the market honestly: Claude
Describe your idea and ask Claude to calculate TAM/SAM/SOM using three different methods — top-down from industry reports, bottom-up from customer count, and value theory. If the three methods don't roughly converge, you don't understand your market well enough yet.
Map your real competitors: Claude + Google
Not just direct competitors — what do people do instead of your solution right now? That "instead of" answer is your real competition, and it's almost always more honest than a competitor matrix.
Test demand without building: Carrd
Free one-page landing page with a waitlist signup. If 20 strangers won't give you their email for something free, the idea has a messaging problem at minimum and a demand problem at worst.
Talk to 5 real people
Not friends. Not family. Strangers who actually have the problem. One conversation where someone says "I've been looking for exactly this" is worth more than any analytics dashboard.
Total cost: $0
Time investment: 48 hours
Time saved if the idea is wrong: months of your life
Full validation playbook with every prompt and template at xedge.tech
What idea are you sitting on right now that you haven't validated yet?

Why I stopped trying to grow and started trying to help — and what happened to XEdge's traffic

Harsha Kumar — Wed, 01 Jul 2026 05:02:04 +0000

For the first two weeks of building XEdge I didn't mention the product once.
I showed up in Discord servers, answered questions on Dev.to, replied to LinkedIn posts from bigger accounts — all without pitching anything. Just genuinely trying to be useful.
By the time I mentioned XEdge, people already knew I knew what I was talking about. The click-through rate on those first mentions was dramatically higher than anything I've posted since, because trust was already there before the ask.
700 users later, $0 marketing spend, I keep coming back to the same lesson: distribution isn't a numbers game, it's a trust game.
You don't need more reach. You need more trust density in the right rooms.
What room are you in right now where you've been lurking instead of contributing?

A month in, 0 sales, and what I think I actually got wrong

Harsha Kumar — Mon, 29 Jun 2026 03:45:02 +0000

Real numbers: XEdge has 700+ users, $0 revenue, and a $29 playbook that hasn't sold a single copy in a month, even with a free coupon code live.
Here's my honest read on why.
I assumed "real, lived experience" would sell itself. It doesn't. I wrote a genuinely specific document — actual TAM/SAM/SOM math from XEdge, the exact validation prompts I used, the pricing mistake I almost made — but my posts about it stayed vague. "Check out my playbook" doesn't tell anyone why it's different from the thousand other startup guides already out there.
Specificity is the thing I skipped. Generic hooks don't sell specific value.
So today I'm being as concrete as possible: this is the actual market sizing process for XEdge, shown step by step. These are the actual prompts, copy-pasteable. This is the actual pricing fear I almost gave into and how I avoided it.
If you're early-stage and stuck before you've even started, free with code FIRST50, or $29 otherwise.
The lesson: vulnerability gets attention. Specificity converts it. I had the first one. I was missing the second.

The 3am question that stops more startups than bad ideas ever do

Harsha Kumar — Sun, 28 Jun 2026 08:06:56 +0000

Everyone assumes the hardest part of building something is the idea, or the execution.
It's neither.
It's the moment — usually late at night, usually alone — where you ask yourself: does anyone actually want this, or have I just convinced myself it matters?
That question killed more of my early momentum than any technical problem ever did. The fix wasn't motivation. It was a framework — proper TAM/SAM/SOM math, validation prompts that surface the truth instead of letting you guess, pricing psychology that removes the fear-based mistakes.
I wrote all of it down while I was still in that uncertain place — not after success gave me hindsight clarity.
If you're sitting in that 3am question right now: that's exactly who this playbook is for. $29, free for early grabs with FIRST50.
The lesson: the fear isn't a personal flaw, it's universal. The only difference between people who start and people who don't is whether they have a way to answer the question fast instead of sitting in it for months.

The idea you haven't told anyone about

Harsha Kumar — Sat, 27 Jun 2026 04:33:44 +0000

Five months ago I had an idea for XEdge and didn't tell anyone for weeks.
Not because it was bad. Because saying it out loud made it real, and real meant I'd have to actually try.
I think a lot of people are sitting on something like that right now. Not lacking ideas — lacking the nudge to say it out loud to someone who won't laugh.
So here's an open question: what's the idea you haven't told anyone about? Drop it in the comments, here or anywhere else you've seen this question today.
No pitch coming. Just genuinely curious what's out there.

Why I stopped waiting on Fiverr and went direct to my audience instead

Harsha Kumar — Sat, 27 Jun 2026 04:19:43 +0000

Spent the last few days setting up Fiverr gigs — landing pages, AI stack research, content writing. Zero impressions, which is normal for a brand new seller, but also slow.
Here's the realization: I already have an audience that knows and trusts me from 5 months of building XEdge in public. Why wait on an algorithm to surface me to strangers when I can offer the same service directly to people already paying attention?
So today I'm opening 3 paid spots for 1:1 AI stack research — same service I'd list on Fiverr, just sold directly through the audience I've already built.
The lesson: platforms with built-in audiences (Fiverr, Upwork) are good for cold discovery, but if you've spent months building any kind of audience already, sell to them directly first. It's faster, it's warmer, and you skip months of algorithm patience.
xedge.tech — and if you want a stack built for your specific goal, the offer's open this week.

The skill I built for my product turned out to be sellable on its own

Harsha Kumar — Thu, 25 Jun 2026 04:53:01 +0000

For the last 4 months I've been building Xedge — helping people find the right AI tool stack for their specific goal.
This week, needing to cover some short-term costs, I realized something: I've been doing this exact service for free in DMs for months. Someone asks what they're building, I figure out their AI stack, I reply with the breakdown.
That's a sellable skill completely separate from the product itself.
So I'm testing it directly — listing "AI stack research and setup" as a freelance service while XEdge keeps compounding in the background.
The lesson here: sometimes the tool you build to solve a problem at scale has a manual, sellable version hiding inside it. Don't assume the only output of months of building is the product itself — the expertise you gained along the way often has its own market.
Anyone else found unexpected income inside a skill they built for something else entirely?