I spent ten years teaching finance professionals how to prove what things cost. I now do the same for AI budgets, and I publish the method so anyone can check the work.
Most people advising on AI spend arrive from marketing or from engineering. I arrived from both, and from cost accounting before either.
A CIPFA diploma in public finance, then nine and a half years as an academic tutor in cost accounting for CIPFA in London. Job order and process costing, activity-based costing, variance analysis, budgeting, taught to people who would go on to sign off public money.
Founded Online Marketing Academy in Skopje and have run it for more than fifteen years: email marketing programs, content strategy, SEO, and courses that trained B2B founders and marketers. Fifteen years of watching what actually converts, measured, before AI made it faster.
A global automotive manufacturer needed thousands of recorded customer conversations turned into market research its team could act on. The obvious approach, one model prompted to read everything and write the report in a single pass, produced confident, inconsistent output and described features the vehicles did not have. For a marketing team, that is not a formatting error. It is a wrong strategy shipped with confidence.
Structured facts pulled out of each transcript before any writing happens.
Each claim tied to the manufacturer's own documentation, not to model memory.
The report draws from verified material and nothing else.
Review sits inside the pipeline, not after it.
Every instrument I build states how it works and where it stops. Clients can audit the method before they hire anyone.
The AI Citation Audit runs a brand through the engines buyers actually use, checks every citation, and validates each result against the page it came from. Every finding carries an evidence grade.
The Executive AI ROI Scorecard has one calibration rule: a practice earns a high score only if it is true, documented, and would survive the person who runs it leaving. A roadmap in someone's head scores a 1.
The AI Citation Report carries a live source on every statistic and states the limits of the audits in print, including the things we will not promise. Third-party benchmarks are labeled as benchmarks.
"If a vendor's numbers cannot be interrogated, they are decoration. Ours can be."
A documented review of AI spend and use cases, actual presence in AI-generated answers, the condition of company data, and the workflows where implementation pays back fastest. Delivers a prioritized roadmap, defined KPIs, an implementation plan, and an executive presentation.
Where a company stands against competitors inside ChatGPT, Claude, and Perplexity, with every citation validated against its source page and benchmarks attached to every metric.
Built with Ollon. Ingestion, chunking, retrieval tuning, and evaluation, grounded in a company's actual documentation. Running on Google Cloud, Vertex AI, and n8n, with no coding required on the client side.
Getting a company cited, and cited correctly, by the engines its buyers ask. Measured against the published citation-audit methodology rather than against screenshots.
Orchestrated systems that turn scattered market signal into research a team can act on, with the same grounding rules that keep the automotive reports honest.
Each one computes in your own browser. Nothing you enter is stored or sent anywhere.
Twenty statements, five minutes, one number for your next board conversation, plus the section that tells you what to fix first. Score what is in place today, not what is planned. The full report ranks your four sections, quotes your own lowest answers back to you, and sequences a 30-day plan.
Start the scorecard revenueexperts.aiOne row per AI initiative, tool, and subscription your company funds. One owner, one cost, one number each. It returns your measured-to-unmeasured ratio and a decision list. Rows you cannot complete are the finding, not a gap in the exercise.
Build the inventory onlinemarketingacademy.aiCloud AI is rented capability on someone else's infrastructure. Sovereign AI is capability you own. Which one you need depends on your regulatory exposure, data sensitivity, and operational capacity, not on a vendor's demo. Twelve questions, about three minutes.
Run the checkThese give directional signals. The deployment and investment decisions belong in a proper assessment, based on your data, regulatory exposure, and governance requirements.
Certified Google Partner, building on Google Cloud and Vertex AI. Google AI Professional Certificate and AI for App Building, both earned March 2026. CIPFA diploma in public finance. Bachelor's degree in marketing, Ss. Cyril and Methodius University, Skopje.
One short issue a week on AI ROI, AI visibility, and data activation, written for the executives who sign the AI budget. One verified source per issue, scope notes included.
An evidence-first report on AI visibility. Every statistic carries a live source, and every limit of the audits is stated plainly, including the things we will not promise.
The Executive AI ROI Scorecard: twenty statements, five minutes, across spend and return, AI visibility, data readiness, and workflow change. Free.