AI Projects
An Organized Overview
I am currently developing four main Artificial Intelligence projects:
1. Personal Assistants (Private Use Only)
This suite includes four AI assistants, each tailored to my personal needs:
Travel Assistant
A smart travel guide that suggests specific flights based on my preferred travel window and destination. The goal is to receive concrete proposals, not vague advice. Gemini has performed well so far; GPT-4.0 was serviceable but too generic. This tool replaces the need for clumsy travel platforms by working from a set of detailed travel preferences that I provide.
Health Assistant
An assistant that understands my physiological conditions and health records. I want to consult it when making health-related decisions—complementing or replacing human advice through AI that knows my specific profile.
Finance Assistant
A financial guide that knows everything about my situation: bank accounts, expenses, long-term goals, and personal preferences. My conversations with GPT-4.1 on managing my UN pension have been useful. This assistant should offer well-informed, contextualized advice.
Writing Assistant
Probably the most important: a writing partner that understands my voice—spoken, slightly ironic, yet grammatically sound and stylistically refined. I want correct, natural English that still sounds like me—not overly formal, not broken.
Business Model (Internal)
- Input: My time and instructions
- Output: High-relevance tools to enhance productivity and personal well-being
2. DionBot for Wuhan-Umbria.com
DionBot (from Dionysos Bot) is being rebuilt from scratch to better serve food and wine-related interactions. The existing version in the Meow suite is not satisfactory.
New DionBot modules will include:
- Food-Wine Pairing
- Menu Creation
- Recipe Generation
These modules will be powered by interactive Forms that collect detailed input (e.g., dishes, spiciness, preferences) and generate structured GPT prompts. A key design principle is the separation of content from programming—logic should live in editable configuration files, so changes don’t require code rewrites.
The final product will likely be a WordPress plugin, though the choice of no/low-code platform and language model (LLM) is still undecided. Different LLMs may be used in parallel depending on performance and cost.
Business Model
- Input: Medium work, low AI subscription cost
- Output: €30/year subscription
- Target: 500 users globally
3. PDAI.org (Parkinson’s Disease AI)
PDAI.org is a scientific and public information platform focused on Parkinson’s Disease, beginning with content on Tai Chi and expanding into broader topics.
Its goals are to:
- Deliver open, reliable, AI-curated information
- Create a clean, eye-catching website (non-WordPress)
- Explore sustainable funding models such as subscriptions and research grants
Business Model
- Input: High time investment; variable compute resources
- Output: €80/year subscription
- Target: 1,000 subscribers
4. SCIAI.eu
SCIAI.eu powers PDAI.org with deeper infrastructure for scientific literature analysis using AI.
Two core offerings are envisioned:
- On-Demand Dossiers
Example: “Want everything on metformin?” We’ll create a structured dossier from thousands of scientific articles. - Scientific Collaborations
SCIAI.eu may partner on funded research projects. For now, it’s a personal research environment, with monetization as a secondary goal.
Scalability Outlook
Currently private, but both SCIAI.eu and PDAI.org may face scalability questions in the future, depending on uptake and collaboration opportunities.