AI Consulting that lands in daily work: analysis, pilot, rollout.
Mijo has been working with AI actively since 2022, from the GPT-3 API through the first Midjourney releases to today's multi-model practice. This website itself was built with coding agents, and the logged usage is public on the AI in Numbers page. For your team that means: creatives straight from the model, clean data for Smart Bidding, custom GPTs for recurring workflows and an EU AI Act classification that runs along from day one.
First call 30 min · honest about what's worth it and what isn't.
Who does it
Mijo Jurisic, Google Ads consultant, founder of MJ Marketing.
Working in Google Ads since 2021: self-taught on own online projects, deep dive into auction mechanics, conversion tracking and search-engine logic. 2023-24 as a consultant in Google's Premier Partner programme with 500+ direct client accounts and top DACH agencies, €20M media budget under direct responsibility.
MJ Marketing since 2024: Premier-Partner methods delivered directly to SMEs and local service providers. No junior handover. No marketing fluff. No lock-in.
— Mijo Jurisic
How it runs
Analysis, pilot, rollout: in that order.
- Step 1
Analysis: what your team actually does today
We walk through your marketing and ops routine: which tasks repeat, where the data sits, which licences you already pay for. The output is a list of use cases sorted by effort and expected benefit. In my experience the spectacular ideas rarely sit on top, recurring text and data work does.
- Step 2
Pilot: one use case, real data
The top use case gets built and tested on real data from your business, not on demo examples. The success criterion is fixed upfront, for example handling time per case, error rate or turnaround. If the pilot misses it, it gets rebuilt or dropped instead of rolled out anyway.
- Step 3
Rollout: into your team, not into a dependency
Prompts, workflows and access live in your own accounts, not in mine. On top of that you get a handover doc and a session with the people who work with it every day. The goal is reached when you can carry on without me.
- After that
Sharpening instead of a permanent retainer
Models, prices and limits change on a monthly rhythm, so a review belongs to it: does the workflow still run, is there a better or cheaper model by now, does the EU AI Act classification still fit. If you want that continuously, book the Monthly Advisor, everyone else gets in touch when something comes up.
Illustrative AI scenarios
What typically comes out.
Illustrative, anonymized AI consulting scenarios from practice: use case, tool stack, outcome. The real measured numbers on my own AI usage are public on the AI in Numbers page.
| Industry | Use case | Before AI setup | After sprint |
|---|---|---|---|
E-Commerce · fashion DTC Sprint Q4/2025 | Ad-creative pipeline |
| Creative production cost €2,400/mo external€600/mo internal |
B2B services · SaaS sales Sprint Q3/2025 | Custom GPT · sales notes |
| Time for call notes 45 min/call8 min/call |
Agency · whitelabel Sprint Q1/2026 | Audit + internal tools |
| Workflow automation 0 automated12 workflows live |
Coaching · content producer Sprint Q3/2025 | Content repurposing |
| Posts from long-form 3 posts/webinar18 posts/webinar |
Healthcare · telemedicine platform Audit Q4/2025 | Compliance audit |
| EU AI Act readiness Unclear, 0%Mapping done |
Tax advisory · SME Sprint Q1/2026 | Document classification |
| Client doc processing manual, 2h/clientsemi-auto 20 min |
Three principles
ROI-first, no AI theatre.
ROI before tool choice
I start with the concrete use case, not with "which AI platform should we buy". If the benefit isn't clear, no tool comes in. In my experience AI projects rarely fail on the technology, they fail because a tool gets bought first and a problem for it is searched afterwards.
Build vs. buy
Some use cases need their own agent or an API integration, others are solved better by an off-the-shelf subscription. I test new models through a gateway like OpenRouter before recommending them, automations run through Make, and whatever should stay local runs locally on ComfyUI. You hear honestly what pays off at your volume and where vendor lock-in lurks.
EU AI Act built in
In force since August 2024, strict from 2026. I build use cases classified from the start (low-risk / limited-risk / high-risk), no rework when audits come.
Pricing
Three formats: audit, sprint, advisor.
Hourly rate, sprint package or monthly subscription. First call always free.
- Marketing stack audit
- Use-case prioritisation (ROI)
- Build-vs-buy tool recommendation
- 30-min walkthrough
- 4× 90-min working sessions
- Tool setup + configuration
- Handover doc for your team
- 30-day Slack support afterwards
- Monthly strategy call
- Slack support for tool questions
- Vendor reviews + recommendations
- EU AI Act updates
What's not included
Tool licences (OpenAI API, Claude API, Midjourney etc.) are not in the price. You pay vendors directly. I help with selection + setup. Custom software development beyond sprint scope is a separate web-dev mandate.
Free, 30 min: honest assessment whether AI solves your problem.
3 months minimum, then cancellable monthly.
Compliance check built in by default on every mandate.
What you don't get
Anti-bullshit guarantee.
What's currently standard in the AI consulting market, and I explicitly don't:
AI makes teams more efficient, not redundant. Anyone selling you "AI transformation" that saves 30% headcount either lied or has unrealistic assumptions. I talk about concrete workflows, not headcount reduction.
Some agencies sell €10k solutions that at their core are a ChatGPT API call. I honestly say when custom code pays off and when €20/mo SaaS suffices.
I get no commissions from OpenAI, Anthropic, Microsoft, or other AI vendors. Recommendations are independent. You pay me for advice, not affiliate links.
Strategy is 5 bullet points + decisions. Whoever needs 6 months for AI strategy without concrete delivery is selling theatre. Audit is 5 days, implementation 4 weeks.
FAQ
Common AI consulting questions.
First call
30 minutes. We check the use case.
You explain business model + current marketing stack + AI ideas. I honestly say what's worth it, what's hype, which tier fits. If AI isn't your main problem right now, I'll tell you that too.
Free. No sales pressure. No vendor bias.
Hands-On Tool Stack
What I actually run daily, incl. Alpha & Beta.
Active AI practice since 2022: from GPT-3 through the first Midjourney releases to today's multi-model reality. How much of it actually runs through the accounts is on the AI in Numbers page, with snapshot date, source and a breakdown per model. Stealth-tested pre-release models like Horizon or Sonoma show up there too, attributed to their confirmed maker. Recommendations per use case, not per favourite vendor.
Models & providers
Alpha & Beta · Pre-Release
Code Agents & IDEs
Image & video
Design, automation & research
Gateway & local hardware
No commissions, no vendor partnerships. If a €20/mo SaaS suffices, I say so, even if it means fewer consulting hours for me. If a pre-release model solves your use case better than the official release, I say that too.