Hikmah Labs
contact →
We build web services and Telegram bots, automate document flow and routine operations, and embed AI into the systems a company already runs. The price is fixed before we start, you see the result at every stage, and a 90-day warranty applies after handover.
Web services, Telegram bots and internal tools. We take the task from idea to launch: design, development, testing, deployment to production and monitoring afterwards.
first working version in weeks
web services — from $4,000 · bots — from $900
more →Document flow, reconciliation, reporting, approvals. We set up a system that runs the routine operations on its own.
before we start, we count the hours your team spends on that routine each month
from $1,000 · 2–4 weeks
more →We embed AI into the systems you already run: 1C, internal services, knowledge bases. Agents process documents, answer clients and staff, and prepare drafts and reports.
from our work: a counterparty risk report in under a minute
from $1,400 · 3–5 weeks
more →The agent answers inbound calls and handles routine questions, runs outbound campaigns and captures leads. Complex cases go to a human operator.
built for a financial services company and a logistics operator
from $1,400 · 3–4 weeks
more →We keep the product running after launch: monitoring, fixes and new features as your needs change. Retainer format, so the cost of support is known in advance.
we find out about failures before your users do
from $170/month per system
more →24/7 counterparty monitoring: the service tracks court cases, bankruptcies and debts and warns you when something changes.
A vault for API keys: the real key is encrypted and never leaves the service, while apps and AI agents get revocable passes instead.
Counterparty checks inside Telegram: send a tax ID and get a risk report back in under a minute.
Automatic comparison of signed documents against the master copy: the service catches swapped dates, amounts and payment details.
An employee sends an idea as text or voice, and the bot drafts a news item to the company's editorial rules and walks it through approval.
A B2B cross-posting platform: write a post once and publish it everywhere, with AI adapting the text to each platform's format.
An AI agent takes client calls and answers routine questions by voice. Operators are left with the complex cases.
The agent takes inbound calls and makes outbound ones: it captures requests, clarifies details and hands them off for processing.
A support bot built on the company's knowledge base: it answers clients from your own materials and keeps the base updated from new documents.
An AI pipeline for an online school: a recorded lesson becomes a full package of teaching and marketing materials.
We discuss the task and tell you what is feasible and what is not — free of charge. We sign an NDA before the details if you need one. For complex projects there is a paid discovery phase: $170, 3–5 days, a written risk report. Its cost is credited against the main project.
Scope, timeline and price are put in writing before we start. Changes along the way are agreed separately, so the invoice does not grow quietly.
We work in stages with 50% upfront. Each stage ends with a demo of working functionality, so problems surface early, when they are cheap to fix.
Two rounds of revisions are included in the price. After the final payment, the code and exclusive rights to the product transfer to you under the contract.
For 90 days after handover we fix technical faults caused by us at no charge. After that, support and further development run on a retainer.
An AI agent once took down a client's live site. We restored it in three minutes from a copy made before the work started. Since then, a backup is the mandatory first step of every task.
On anything involving payments, access rights or third-party data, we review the risks first and write the code second. We carry responsibility for the result.
Before handover we hand the project to separate AI agents tasked with finding vulnerabilities. That is how we once closed a payment bypass before the client ever saw it.
The project cards show real statuses: live, in progress, testing. The client always sees the actual state of things.
Everyone around me was talking about ChatGPT, and I sat with it in the evenings like everyone else: ask something, laugh, close the tab. Then one day it solved in a minute a problem I had spent an hour on. I remember putting the phone down: this is not a toy. That thought never let go.
Came back from wintering in Saudi Arabia with the feeling that something had to change — and sat down to code. By the end of January I tried Google Antigravity, the first agentic tool in my hands, and spent whole evenings building anything and everything. Two dozen practice projects and not a single client.
Went to an entrepreneurship forum in Sharjah, just to listen. Watched people my age talk about their startups from the stage. Went home with mixed feelings, somewhere between envy and anger at myself. And my tool was already sitting in my pocket.
Enough practice projects — I went looking for real clients. Started messaging businesses on Instagram directly: ten to twenty messages every evening. Silence in return. But I held on to one thought: if even one person answers, there is a chance.
The first client did not come from Instagram. In a Telegram chat for developers, someone was looking for a person who builds products fast. I answered right away. A week later I showed the result — he liked it. One order turned into a second, the second into a third. A month of being ignored paid off with one message.
Started a Telegram channel — decided to keep the log in public. In the first post I wrote it as it was: second year of university, a dorm room, a few live projects, $500 in two weeks. That was the first time I said out loud what I did not fully believe myself: this could become a studio.
There was more work than my own two hands could take, and the first three students appeared. I teach them to build products with AI the way I do it myself. While explaining, I laid out my own process properly for the first time. You can only explain what you understand all the way down.
A day I will not forget. An AI agent, adjusting something small on the server, took down a client's live site. The backup made before the work started saved it: three minutes and the site was alive again. Since that day, no task in the studio begins without a backup.
I circled the Claude Code subscription for a long time: $200 a month is real money on a student budget. Then I did the honest maths: a client product means weeks of revisions, and my time is worth more than that. Bought it. Never regretted it once.
The project with payments was ready for handover. But instead of shipping it and closing the invoice, I opened a clean session and gave the AI agents one task: break it, get past the payment. And they found a hole the first audit had missed. The client will never know about it. That is exactly the point.
Reread the log from the first entry. In winter I was thrilled by any reply at all — now there is more than $7,000 of delivered projects, and the studio has contracts, stages and warranty obligations. The numbers are nice, but here is what matters: this stopped being a side gig.
The circle closed. My first student has been working with me on the team for a month — there are two of us now. Seven projects in progress, and the studio has this site. I look back at autumn 2024 and cannot believe it all started with evenings in ChatGPT. More to come.
The founder works on every project, so questions get answered directly and fast. The commitments are still formal: a contract, defined stages, a 90-day warranty. The log above shows how the studio actually works.
We take backups before any work starts. On anything touching payments or third-party data, the code goes through a risk review, and separate AI agents audit the project before handover. We sign an NDA by default.
After the final payment, exclusive rights to the code and documentation transfer to you under the contract. We do not reuse your product or show it to other clients.
We work in stages and demo each one, so problems surface early, when they are cheap to fix. Two rounds of revisions are included in the price, and fixes are free for 90 days after handover.
The first consultation is free: describe the task and we will tell you what is feasible and what it costs. Ballpark figures: AI agents and voice agents from $1,400, automation from $1,000, Telegram bots from $900, support from $170 a month. The price is fixed in the spec before we start and does not move during the work. For complex projects there is a discovery phase: $170 and 3–5 days for a written report with risks and an exact price, credited against the project.
Write to us — the founder answers, not a manager.
The consultation is free, two sentences about the task are enough.
While you are deciding, read the studio log: @dorm_to_studio