Skip to content
avernek.com
The live avernek.com homepage showing the hero, positioning line and the three-stage system diagram

Plate 01 · Company · Product · Sales

Building an AI company from zero

Avernek Technologies is a registered Nepali company building AI inquiry systems for businesses that lose customers between an advert and a reply. I started it, sell it, direct its design, and answer for it when something breaks.

Role
Founder, Sales, Team lead, Design direction
Year
2026
Status
Live
Stack
Founder-led sales · Client discovery · Design direction · Delivery

The problem

Businesses here spend real money getting people to message them. A message arrives at 9pm on a Saturday. Nobody sees it until Monday. By then the customer has booked somewhere else.

The ads worked. The business still lost the sale. Nothing owns the gap between an enquiry arriving and someone being ready to deal with it, and that gap is what Avernek sells a fix for.

My role

Founder and Managing Director. In a three-person company that means sales, requirements, design and delivery.

I find the clients and run the meetings. I work out what a business is actually asking for, which is rarely what they said first, and turn it into something we can quote and build. I price it, close it, and design what we ship. Then I lead the team through delivering it.

Pragyan is our CTO and builds the automation. Sushant runs the creative and ad side.


Four decisions I made early

  1. 01

    Sell the outcome, never the technology

    The site doesn't open with the word AI. It opens with turning enquiries into tracked sales opportunities, because that's the sentence an owner repeats to their partner. The technology is how we do it, not what they are buying.

  2. 02

    Promise handling, not revenue

    Our footer says we improve enquiry handling, tracking and reporting, and that we don't guarantee sales. It costs us a few deals. It also removes every argument six months in, which is worth more.

  3. 03

    Testimonials in the client's own words

    Every quote on our site is the client's own message, tidied for grammar and nothing else. One of them opens by saying they were not convinced at first. That one performs better than the polished ones.

  4. 04

    Three packages, one recommended

    Core, Growth and Scale, with Growth marked as recommended. Owners don't want a configurator. They want to know which one businesses like theirs pick.

The website

The site is dark, typographic and deliberately not playful. The audience is a business owner deciding whether a small Nepali company can be trusted with their enquiries, and warmth isn't what reassures them. The design direction, the positioning and most of the words are mine.

It was built with AI in the loop. That isn't a confession, it's the point: the distance between deciding something and having it live was days rather than months, which for a company with no runway is the whole advantage.

The Avernek homepage on a phone-width viewport, showing the same hero and system diagram stacked.
The same page at 430px. Most of the audience is a business owner reading a message on their phone, so the mobile layout is the one that actually has to work.

Under it

My part
Positioning, copy, design direction, the build
Not my part
The AI automation engine behind it, which is our CTO
Approach
AI-assisted, so a decision reached the live site in days

What I got wrong

I under-priced the first few deals to get signatures, and then had to argue my way back out of the number I had set. I built features clients described rather than the ones they would use. I spent too long polishing the website before there was anyone to send it to.

Most of this is being learned on live clients.