← to projects

Personal site and tool

A solo practice in AI adoption for small and mid-sized businesses in Uzbekistan. Two artifacts: a personal site and a tool, Quiz — a narrowly focused diagnostic.

The site

The site is a face, not a sales landing page. The home page is “nice to meet you.” Projects is what I do. Audit is a separate door for those who want a business breakdown.

Home hero: photo, name, a few basic facts
Home/

A personal page: photo, a short bit about me, date of birth, city, what I do, a timeline of my life, interests.

Projects hero: the foyer and the first door — Personal infrastructure
Projects/projects

A list of projects: personal infrastructure, this page, RPGameBot. Each one has its own subpage with a write-up.

Audit hero: “You’re digging in the wrong place” + a button into the Quiz
Audit/audit

The landing page for the diagnostic quiz: what pain it solves, who it’s for, what the client walks away with, and how going through it works.

Quiz

When a small-business owner in Uzbekistan asks an AI “what should I do,” they get a generic answer about the US and Europe. The benchmarks are someone else’s, the advice is shallow, the local market isn’t taken into account.

A narrow quiz. Separate for three industries. Several AI stages, each with its own context and role. The benchmarks are real, gathered from publicly available Uzbek sources, and they get better with every run (we’ll also expand them in the future with paid sources).

architecture · 4 stages
Q
questions
fixed + conditional numeric
↓
S1
initial diagnosis
hypothesis, scoring
↓
S2
diagnosis
author + self-critic
↓
→
report
tailored to the business type
in parallel: lead → CRM, the bot pings the author

The facts

Who
SMB owners in Uzbekistan
Tracks
ecom · services · b2b
Stack
Next.js · Postgres · Anthropic
Benchmarks
Uzbek sources + dynamic quartiles
Profiles
tailored to the business type within a track
Leads
mini-CRM + Telegram bot
Status
active development

The user’s path

1
The user lands on /audit, reads the promise “5 minutes → a personal plan,” and hits the CTA.
2
They go to quiz.gapurik.com and pick their industry out of three.
3
They answer a series of questions. Some of the numeric ones are only asked for certain answers.
4
Stage 1 works through the answers: an initial hypothesis + scoring. The numbers are compared against Uzbek benchmarks and the track’s dynamic quartiles.
5
Stage 2 writes the final report by frames: profile, blind spots, the causal chain, anti-recommendations, growth areas, CTA. On top of it — a self-critic (checking tone and jargon).
6
The user gets a report tailored to their business type — the track, the profile within the track, the local context in every block.
7
The lead drops into the mini-CRM, the bot pings — then sales takes over by hand.

The best of the project

The benchmarks grow along with the database

On top of the existing ones from publicly available Uzbek sources, dynamic quartiles from real runs are added. With each new user they get more accurate.

Adapts in real time

The quiz shifts as you go: the direction is chosen from the first answers, the text and options adjust to the business type, and numeric questions appear only when there’s a point in asking them.

Moving toward any business type in Uzbekistan

Today there are three tracks (e-com / services / b2b) with subtypes and profiles inside each. The goal is for any business in Uzbekistan to recognize itself in the report. Each version is more accurate.