01 — Sound familiar?
Six questions every Uzbek SME owner answers "yes" to at least twice.
If two of these sound like your business, the next ten minutes are worth your time.
Biznesingiz hammasi Telegramdami?
Orders, customers, employees — all in private chats with no audit trail.
Hisobotni oyiga bir marta, qo'lda yig'asizmi?
Monthly P&L assembled in Excel by hand, two weeks after month-end.
Laboratoriya daftari hali ham qog'ozdami?
Production decisions written down once and never queried again.
Egasiz savdo to'xtaydimi?
Every quote, every approval, every dispute — through one person.
Yangi xodim uch oyda o'rganadimi?
Onboarding lives in someone's head; one in three leaves before it ends.
Odoo migratsiyasiga $50,000 tashlamoqchimisiz?
Six-month rollouts that finish where Excel started. There is a smaller way.
02 — What changed
Something changed in the last eighteen months.
For ten years the answer to those questions was the same — a bigger ERP, a foreign vendor, six months of integration, and reports nobody trusts. AI collapses the answer. Language models read your messy inputs. Deterministic engines do the math that has to be right. Your existing software stays where it is. We build the bridge in weeks. Your team owns it forever.
03 — Services
Four ways we work with companies.
Pick where you are, not where the vendor wants to put you.
AI Strategy & Discovery
Two weeks paid. We sit with your data and your team, and you walk away with a written automation roadmap, scored by ROI versus complexity. If your problem doesn't need AI, we say so on day five.
AI Automation & Implementation
Production AI systems on top of your existing software. Telegram bots that replace paperwork. Reports that build themselves. Standards-compliant engines for safety-critical work. Your developer co-builds with us from day one.
AI Training Programs
Audience-segmented, hands-on, full-day on-site. Owners, knowledge workers, developers — each track has its own curriculum and its own tools. 30-day adoption telemetry included; we measure use, not satisfaction.
Long-term AI Partnership
For companies treating AI as an operational discipline, not a one-time project. Framework agreement, retainer, right of first refusal on AI work, quarterly strategy reviews with the founder. You pay for the foundation once.
04 — What we actually build
Five patterns we ship.
The patterns repeat across industries. The data and the language change; the architecture doesn't.
| № | Pattern | Description |
|---|---|---|
| 01 | Conversational AI workflows | Telegram, WhatsApp, web. Replace handwritten or phone-call processes with structured intake that still feels like a chat. |
| 02 | Standards-compliant computation engines | Where safety or regulation requires deterministic answers, AI handles the language layer and a deterministic engine handles the math. This is how we build for concrete recipes, financial statements, and clinical workflows. |
| 03 | Document & knowledge intelligence | Search, summarization, drafting, translation across your contracts, manuals, and operational records. NotebookLM-style knowledge bases that actually get used. |
| 04 | Operational dashboards & executive reports | Real-time intelligence drawn from your existing ERP, accounting, or dispatch systems. P&L, balance sheet, cash flow, custom KPIs — exported to PDF and Excel, reconciled against the last four quarters. |
| 05 | Custom AI agents | Multi-step workflow automation using Anthropic Claude, MCP servers, and the modern agentic stack. Not chatbots — workers. |
05 — Selected work
What we have shipped.
Hadid Beton — Manufacturing AI · 2026
A Telegram-native concrete recipe intelligence system, an internal ERP security review, and a four-statement financial reporting suite. First phase of a multi-year partnership extending into logistics, sales, and predictive maintenance.
Intercity Taxi & POS Tech — AI Training · 2026
Designed and delivered a full AI adoption program for the engineering and marketing organization of a regional intercity taxi platform that also builds POS and distribution software.
p57.uz — Founder project
A published prompt-engineering and critical-thinking platform used by individuals and organizations across the region to build internal AI literacy.
E-commerce image-editing AI — Founder product
A production AI application that automates product-image editing for e-commerce sellers — background removal, color correction, multi-angle generation.
06 — How we work
Six rules that came from watching too many AI projects fail.
| № | Principle | Detail |
|---|---|---|
| 01 | Stage 0 before you build | Two weeks of paid discovery. You get a real plan with real numbers. If we find your problem doesn't need AI, we tell you. That has saved more than one client over USD 30,000. |
| 02 | Phase by phase, with payment milestones | No phase begins until the previous one is signed off and paid. You are never out more than a single milestone ahead of acceptance. |
| 03 | Hybrid Symbolic-Neural design | Language models are great at language. They are not great at safety-critical math. We use AI for understanding and explanation, and deterministic engines for any computation that ends up in a safety, financial, or compliance decision. This is why our systems do not fail publicly. |
| 04 | Co-delivery with your team | Your developer or operator participates from day one. When we hand over, your team owns it. No vendor lock-in. No 2 a.m. emergency support contracts. |
| 05 | Documentation in two languages | Internal technical docs in English. End-user materials in Uzbek (Latin) or Russian, depending on your audience. Always. |
| 06 | Acceptance gates with real metrics | Every phase has measurable acceptance criteria written into the contract. "It works" is not a deliverable. "≥80% match rate over 50 production batches" is. |
07 — Pricing
Where engagements start.
Anchors only. Final scope and price come out of Stage 0. Full pricing breakdown shared on request.
| Engagement | Starts at | Form |
|---|---|---|
| Discovery sprint | USD 2,500 | Fixed price · 1–2 weeks · written report you keep |
| AI training session | USD 2,500 | Fixed price · full day · on-site · Uzbek / Russian / English |
| AI implementation | USD 9,000 | Per system · 4–10 weeks · co-delivered with your developer |
| Multi-phase build | USD 25,000+ | Framework agreement · phase-by-phase · 12–24 months |
08 — Founder
Xurshid Maroziqov.
Founder & lead consultant
Xurshid Maroziqov founded HEAD after a decade of building and operating businesses across AI, education, and e-commerce. AI builder since 2021. Day-one user of Anthropic Claude Code.
- p57.uz · prompt engineering and critical thinking platform
- E-commerce image-editing AI · production application
- Education Express · English-as-second-language academy
- Hadid Beton, Intercity Taxi, two single-session enterprise trainings · 2025–2026
Based in Kokand, Uzbekistan. Working in Uzbek, Russian, and English.
09 — Common questions
Frequently asked.
We're not a tech company. Will AI work for us?
That is exactly who we work with. Our clients are concrete manufacturers, taxi operators, language schools, and retail businesses. The question is not whether your business is "tech" — it is whether you have data and processes. Most businesses do.
We already use Claude or ChatGPT. Why a consultant?
Using a chatbot for occasional drafting is one thing. Building a Telegram bot that runs your lab, a security-hardened reporting system that runs your finance, or an agent that handles your dispatch — that is a different kind of work. You do not need us to type prompts. You need us when AI has to be production-grade, integrated with your systems, and used by people who are not power users.
Can we just have you train our team and build it ourselves?
Often, yes. About 30% of our clients run their first AI project entirely in-house after a structured training program. Others bring us in for the build phase. Both paths are real.
How is HEAD different from international AI consultancies?
Three differences. Locality — we work in your language, in your time zone, in your business culture. Speed — we ship in weeks because we operate with a small team and modern tooling. Price — our regional pricing is a fraction of what global firms charge, with comparable engineering depth.
How is HEAD different from cheaper local freelancers?
We are a consultancy with a methodology, a contract structure, and accountability. We sign for outcomes. We deliver documentation. We hand over to your team in a way that is auditable. Freelance work is a great fit for some problems. Multi-stakeholder, safety-sensitive, or commercially material work is not one of them.
Will my data leave the country?
For most builds, no. Our default architecture keeps your data on your servers. Where AI model APIs are involved, we minimize what is sent, never send sensitive personal information, and discuss data residency with you in Stage 0.
How do we know in advance if AI is right for our problem?
That is the entire point of Stage 0. Two weeks. USD 2,500. You walk away with a written report — what is automatable, what is not, what to do first, in what order, and what it will cost. If the answer is "AI is the wrong tool for this," that is also a deliverable.
What does the first 30 days look like?
Week 1–2: Stage 0 — interviews, data review, code walk-through, written report. Week 3: scope and price the first build phase, sign the addendum. Week 4: phase 1 kicks off; your developer co-builds with us from day one. By day 30 you have a discovery report and a moving build, not a slide deck.