Skip to main content

How it works

From conversation to calories

Ziva gives your AI assistant the ability to look up real food data, log meals, and track your daily nutrition — all from a normal conversation. When you say "I had a chicken salad for lunch," your AI can look up the food, figure out the macros, and log it. No app switching, no barcode scanning, no manual search.

What happens when you connect

Ziva uses something called MCP (Model Context Protocol) to give AI assistants new abilities. Think of it like installing a plugin. When you connect Ziva, your AI gets access to a set of nutrition tools it didn't have before.

Those tools include searching a food database (USDA FoodData Central, the same data the FDA uses), logging meals to your account, looking up saved meals and recipes, and pulling up your daily progress.

How food gets matched

Your AI searches the USDA database using what you described. The database has hundreds of thousands of foods with full nutrient profiles — everything from raw chicken breast to branded packaged products.

The AI picks the closest match and selects an appropriate portion size. If you said "a bowl of rice," it'll match something like "Rice, white, long-grain, cooked" and pick a portion that makes sense for a bowl (roughly 1 cup).

Sometimes the exact food isn't in the USDA database — maybe it's a specific homemade dish, a local restaurant item, or a combination that doesn't have a clean match. In those cases, the AI falls back to a "quick entry" where it estimates the macros based on what it knows about the food.

You'll see these two types labeled in your meal log. Verified items come from USDA FoodData Central with full nutrient profiles — vitamins, minerals, amino acids, and more. The data comes from the same source the FDA uses for nutrition labeling. Quick items are AI estimates of the macros when there's no good USDA match. They don't have the full 50+ nutrient breakdown, but the numbers are reasonable.

Either way, you can see exactly what was matched and what portion was assumed. Nothing is hidden.

Getting started

Before you start logging, make sure Ziva is enabled in your AI client — not having it enabled is the most common reason logging doesn't work. See the integrations page for setup instructions for ChatGPT, Claude, and other MCP clients.

What gets tracked

After a meal is logged, everything shows up on your Ziva dashboard. Here's what you get:

  • Daily calories and macros (protein, fat, carbs) with progress toward your goals
  • Each meal broken down by food item, with the exact USDA match and portion used
  • Full nutrient details (50+ nutrients) when you tap on a food item
  • Weekly and monthly trends on the Insights page
  • Saved meals and recipes you can log again quickly

Your AI also gets access to this data. You can ask it things like "how much protein have I had today" or "what did I eat yesterday" and it'll pull from your actual logs.

Building a custom integration?

See the detailed technical specs for developers.

Ready to connect your assistant?

Setup takes about a minute.