How much will my AI product cost?

Most complete products land between €15K–€45K. The sprint (from €1,500) gives you an exact number for your case within two weeks — with no obligation to continue.

How do I know AI will actually work for my business?

You don't yet, and neither do I — that is what the data assessment is for. If the data isn't ready or a model isn't the right answer, the report says so and recommends a better approach. Six years of production engineering means there is always a version that ships, with or without AI.

What if the project doesn't deliver the expected results?

Success metrics are defined in week one and you see progress every 3–4 days, so there is no surprise at the end. Every milestone leaves you with software that runs; you can stop at any of them.

What happens after the initial project?

You own all code and documentation — nothing is locked to me. Most clients continue on the retainer for new features; it is optional and month to month.

Which models and cloud do you use?

Claude, GPT or open-weight models, chosen per problem, on AWS, GCP or Vercel — in your account, so the bills and the keys are yours.

Why not just hire an AI engineer?

A senior hire runs €90K–€120K a year and takes three to six months to find and onboard. The retainer starts at €3,000/mo, starts next week, and covers the full stack — not only the model.

Is there a minimum term?

No. It is month to month. Most clients stay because the work is good, not because of a contract.

Founder-led · Pristina · Working software from day 5

We ship the AI product.Not the deck.

Complete AI products — model, backend, UI, deploy — built by one senior engineer in 2–4 weeks. Prices below, not on request.

Book a 30-min call See the work

Scoping call outputLive

scope: support-bot for 200 tickets/day · est. €18K · 3 weeks · start Mon

€1,500: Strategy sprint · 2 weeks
€15K–€45K: Product sprint · 2–4 weeks
€3,000/mo: Retainer · month to month

The problem

You've tried to build this before.

Five ways an AI project stalls. Every one of them is an engineering problem, not a model problem.

The pilot never left the notebook.

A demo that works on your laptop is not a product. Nobody owned the deploy, the monitoring or the bill.

The agency quoted three months.

Discovery, then a deck, then a statement of work. Four weeks gone before a line of code exists.

You got juniors on the invoice.

The senior on the pitch call was never on the project. You paid for the ramp-up twice.

Nobody would name a price.

Every estimate arrived as a range with a caveat. The scope moved every time you asked.

The model works, the product doesn't.

Prompt quality was never the hard part. Auth, data plumbing and uptime were.

Our approach

Working software from day 5, then every three days after.

No discovery phase that produces a deck. The first deliverable is a URL, and it never stops being one.

Day 5 — it runs.

A URL with the core loop working end to end by the first Friday. Rough, but real.

Every 3–4 days — you see it.

A live demo on a call, not a status report. What you say on Tuesday is in the build by Friday.

Week 4 — it's live.

Deployed, monitored, documented, source in your repo. Thirty days of support after.

Measured on real traffic

What shipped, and what it did.

Five products in production. The number on the left is the one the client measured after launch.

See how they work

64%tickets resolved without a humanInstant first response. €110K a year saved on support.

−85%review time per contract99.2% accuracy on key clauses. 40 hours saved per employee, monthly.

−90%manual reporting timeReporting runs end to end. Executives query it in plain language.

8 min → 45 saverage wait time70% of routine calls handled. €180K a year saved.

3×qualified leads processed85% scoring accuracy. Outreach runs on its own.

How they work

Five products, drawn from their own interfaces.

Each panel is the real shape of the thing that shipped — the thread, the extraction table, the agent pipeline, the queue, the scoring run.

Ticket #4821 · Billing

My invoice shows March twice. Was I charged double?

Yes — the annual plan renewed on the 1st and the old monthly plan billed on the 3rd. I’ve refunded the monthly charge (€49); it lands in 3–5 days.Source · Billing FAQ §3 · Stripe refund re_2Kx…

Resolved automatically · 0.8 s

Support bot for a SaaS help desk

Challenge: 200+ tickets a day. First response took 4 hours.
Solution: Retrieval chatbot over the docs, wired into the existing ticketing system.

Retrieval chatbot · Ticketing integration

Master services agreement · p.7

Extracted clauses4 / 4

ClauseValueConf.

Termination90 days notice99%

Liability cap€250,00098%

Governing lawZürich99%

Auto-renewal12 months97%

Contract review platform

Challenge: 2–3 hours of manual review per document.
Solution: Clause extraction model, a review dashboard and a workflow around it.

Extraction · Dashboard

Pipeline · month-end closeRun #218

✓ extracted 1,204 rows✓ 2 anomalies flagged✓ report sent 06:12

Financial reporting agents

Challenge: Finance team losing 15+ hours a week consolidating data by hand.
Solution: A pipeline of agents: extract, validate, analyse, write the report.

Multi-agent · Reporting

Live · inboundRec

“I need a copy of last month’s invoice, the one from March.”

00:00Call in+41 44 …
00:03Transcribedde-CH
00:07Intentbilling · invoice copy
00:09Routedbilling agent
00:45Resolvedinvoice re-sent

Voice agent for a call centre

Challenge: 500+ inbound calls a day, long queues, high staffing cost.
Solution: Voice assistant with live transcription and intent routing; hands off to a person for complex cases.

Voice · Routing

Inbound · todaySynced → CRM

LeadCompanyCo.ScoreAction

M. KellerHelvetia Logistik92Demo booked

A. RossiNordwind GmbH81Email sent

J. BerishaPrishtina Tech64Nurture

S. NovakAdria Retail31Skip

Lead scoring for a sales team

Challenge: Reps qualifying leads, booking demos and following up by hand across platforms.
Solution: Agents for qualification, outreach and CRM sync.

Agents · CRM sync

Shipped & live

6 products, all with a URL.

Open one and click around. Nothing here is a mockup or a case study written after the fact.

Resumemasterresumemaster.net

Resumemaster

AI résumé engine · Next.js + React Native · 10K users

Codepilotcodepilot.dev

Codepilot

SaaS starter kit · Auth, billing, dashboards · Next.js

Linktobiolinktob.io

Linktobio

Link-in-bio product · Web + mobile app

Cremericremeri.com

Cremeri

Skincare storefront · Next.js

Le Plombierle-plombier-gmbh.ch

Le Plombier

Service business site · Lead forms · Next.js

Yunakinyunakin.com

Yunakin

Perk-sharing platform · Web app · Next.js

What it costs

Three ways in, all of them priced.

No “contact us for a quote”. Pick the tier, book the call, get a fixed number in writing.

012 weeks

Strategy Sprint

from €1,500

Two weeks to find out what is worth building, whether your data can carry it, and what it costs. Ends in a fixed quote.

Data & feasibility audit
3–5 scoped use cases
Fixed-price quote
90-day roadmap

What’s in it

022–4 weeks

Product Sprint

€15K–€45K

A complete AI product: model, backend, UI, deploy. Working software from day 5, a demo every 3–4 days, live by week 4.

Model + backend + UI
Deployed on your cloud
Demo every 3–4 days
30 days of support

What’s in it

0360–80 h / month

Retainer

from €3,000/mo

Senior full-stack and AI hours every month, without hiring. Roadmap features, model tuning, infrastructure cost — one engineer who already knows the codebase.

60–80 senior hours
Weekly call + daily Slack
Same-day emergencies
Month to month

What’s in it

Start a project

The terms

What you can hold me to.

Four commitments that are in every contract, whichever tier you pick.

The price on the card is the price on the invoice.

Fixed fee, agreed before the first commit. Scope changes are quoted before they are built, never after.

You own everything from the first commit.

Source, documentation and infrastructure live in your accounts. There is nothing to hand back.

One engineer, start to finish.

The person on the scoping call writes the code. There is nobody to hand off to.

No lock-in on the retainer.

Month to month, cancel any month. Clients stay because the work is good, not because of a contract.

Why one engineer

Most people can do AI or production. The graveyard of AI projects is the gap between them. I've spent six years on the production side and build the AI on top of it. You get one engineer, the whole stack, and a demo every few days.

The engineer

Fortan Pireva

Founder & Senior Full-Stack Engineer · Pristina, Kosovo

Six years building production systems for startups and scale-ups, then AI on top of that foundation. On every Codeks project you talk to the person writing the code.

LinkedIn GitHub

Built for

ResumemasterCodepilotLinktobioCremeriLe PlombierYunakinCC Sporting EventsCrafted Tools

Start here

Your product could be live in four weeks.

Thirty minutes, no deck. Bring the problem and you leave with a scope, a price and a start date.

Book the call