How iBizAI builds a business

Executive summary

iBizAI is the business-AI initiative of CEMI. It does two things that are deliberately kept apart. As a guide it is independent and vendor-neutral: sector guides, quick wins, a glossary, opinion articles and four AI-readiness assessments, published free in English, Spanish and French, that help a company work out what AI actually does for it. As an offer it is a curated suite of tools — the Ibizai AI Suite — that CEMI builds, integrates and operates on the client’s own knowledge and stack.

The separation is the product decision. A recommendation is only worth something if the recommender does not have to win the build; an implementation is only accountable if someone owns it. iBizAI recommends; CEMI builds and operates.

The problem

Outside the largest markets, AI arrives as a demonstration. The demo works, the pilot stalls, and the company is left with a subscription nobody uses. Three things break it: the tool does not know the company’s own reality, it does not speak the company’s language well enough to put in front of a customer, and it is priced for a budget the company does not have.

The published site addresses all three: answers grounded in the client’s own documents with citations, trilingual by construction rather than by translation afterwards, and two regional price tiers instead of one.

The offer

The Suite is organised in three tiers — Basic (reactive helpers), Optimum (active assistants) and Premium (strategic intelligence) — and a Group / Holding edition for diversified groups, where each subsidiary keeps its own vault and brand voice while the holding gets one contract, one compliance console and cross-subsidiary benchmarking.

Alongside it sits the build service: custom helpdesk and customer-service assistants trained on the client’s own knowledge base, with anti-hallucination guardrails, multimedia answers, smart routing and CRM lead capture. The published case is the Franco-Dominican Chamber of Commerce assistant, which answers investment questions, cites Dominican law and decrees, and routes qualified leads to the chamber’s team.

Go to market

The buyer is reached through institutions rather than one at a time: chambers of commerce, business associations, cooperatives and diversified holdings, each of which brings a membership or a portfolio with it. The Franco-Dominican Chamber engagement is the template — an institution buys one assistant, and its members meet the platform through it.

Geographically the sequence is the same as the published price list: the Caribbean and Latin America first, where the developing-economy tier makes the offer affordable and the trilingual content already exists; Europe and North America second, at the developed-economy tier.

The free educational layer is the acquisition surface. It is the reason a decision-maker arrives before there is a sales conversation, and the round keeps it current for exactly that reason.

How we compete

Not on model quality — the Suite orchestrates frontier models rather than training its own, and that is stated on the site. The competition is on the wrapper: what the assistant knows, whose language it speaks, what it refuses to make up, who answers when it is wrong, and what it costs in a market where a first-world price is simply a no.

Against global SaaS the advantage is proximity, language and price tier. Against local integrators it is a published product with a published price instead of a bespoke project every time. Against doing nothing — the real competitor — it is a free educational path that ends in a priced, specific offer.

What the company owns

The published trilingual content library and the assessment instruments; the curated Suite definition and the tool catalogue; the delivery method for grounded, citation-bearing assistants, including the anti-hallucination prompt discipline used in the Chamber build; the brand and domains; and access to the shared CEMI platform — the persona engine, the production pipeline and the delivery arm — which this company does not have to build.

There is no patent and none is claimed. The defensibility is accumulation: content, method, institutional relationships and the client knowledge bases already integrated.

Market

Four in five EU enterprises with ten or more employees still do not use AI. Eurostat measured 20.0% using AI technologies in 2025, up from 13.5% in 2024 and 7.7% in 2021, with adoption running from 42.0% in Denmark to 5.2% in Romania — an eightfold spread between the leading and lagging member states. That is an adoption gap, not a market size: it counts how many firms have started, not what they spend, and it is the gap this company sells into. Source: Eurostat, 20% of EU enterprises use AI technologies (11 December 2025), ec.europa.eu/eurostat/web/products-eurostat-news/w/ddn-20251211-2.

The World Economic Forum’s Future of Jobs Report 2025, surveying over 1,000 employers across 22 industry clusters and 55 economies, projects 170 million jobs created and 92 million displaced between 2025 and 2030 — a net gain of 78 million, or 7% of total employment. That is a global employer survey’s projection of job creation and displacement, not a market size: it counts the jobs those employers expect to appear and disappear, not what anyone will spend. Source: World Economic Forum, The Future of Jobs Report 2025 (January 2025), weforum.org/publications/the-future-of-jobs-report-2025/digest/.

This plan states no market size. We hold no figure for the addressable market that comes with a source we can cite on the page, and the standard this kit is built to is explicit that a section without a source does not exist. Directional evidence for demand is in the published opinion articles, each carrying its own citations; the commercial evidence that matters here is narrower and concrete: an institution paid for an assistant, and the price list for the Suite is public.

Risks, and what we do about them

Model dependency. The Suite runs on third-party frontier models; pricing and availability are outside our control. Mitigation: the orchestration layer is model-agnostic by design, and the site says so — the best model is picked per task.

Concentration. A business built through institutions is exposed to a small number of large relationships. Mitigation: the self-serve subscription tiers this round funds exist precisely to widen the base beneath them.

Trust. One fabricated answer in front of a client’s customer costs more than the contract. Mitigation: grounded answers with citations, human-in-the-loop approval for irreversible steps, and an auditable trail — the six principles published on the Suite page.

Key-person concentration. One named human runs the company. Mitigation is honest rather than reassuring: it is a real risk, it is why the round funds delivery capacity, and it is on the table at diligence.

Regulatory change. AI rules are moving in every jurisdiction we sell into. Mitigation: compliance monitoring is itself one of the Suite’s tools, and the Group console was designed around a general counsel’s view across jurisdictions.

What this plan does not claim

It does not state revenue, a run rate, or a customer count. It does not project a five-year P&L. It does not name a market size. It does not describe the offer’s terms as measured results. Everything the plan does not publish — including revenue and the financial model — is available to investors on request, and the diligence list on the FAQ page says exactly what opens and how fast.