Orzio

A photo-first logging flow

Orzio starts in Telegram rather than asking someone to install and maintain another dedicated food app. Send a meal photo and the service uses a vision model to estimate the foods and likely macronutrients, then returns a compact record in the conversation.

The estimate is useful as a fast log, not a laboratory measurement. Portions hidden by the photo, ingredients mixed into sauces, and preparation methods can all change the real totals. Orzio keeps the interaction lightweight while leaving room to correct or undo a result.

• Meal-photo analysis with estimated calories and macros.
• Personal food rules evaluated alongside numeric targets.
• Reversible entries through an undo flow.
• Telegram and Discord integration paths in the codebase.

Two different kinds of checking

A food plan can contain numeric constraints, such as a daily macro target, and semantic constraints, such as avoiding a named ingredient. Orzio separates those jobs. TypeScript code performs deterministic numeric checks, while the model interprets what the meal appears to contain and whether that conflicts with a written rule.

That division makes the arithmetic repeatable and keeps the probabilistic part where it is actually needed: understanding a meal from a photo and natural-language instructions.

Privacy and account controls

The product includes an account-deletion path designed to remove the linked user record and associated meal photos. That matters for a product handling personal food history and images.

The beta is being used to refine the smallest interaction that still produces an honest, correctable log. It is not presented as medical advice, a diagnosis, or a replacement for measurements from a clinician or dietitian.