TechPrufer
Get started
Uber
Uber logoUber · Feature trail

AI-Powered Cart Generation and Management

2 posts · Jun 2026 – Sep 2026

What this trail covers

Uber is developing advanced AI capabilities to transform user interactions with its commerce platforms, starting with Uber Eats. This initiative focuses on moving beyond traditional search-based interactions to more intuitive, intent-driven experiences. Key developments include the Cart Assistant, which uses multi-prompt state graphs and LLMs to interpret natural language and image inputs, generate draft grocery carts, and manage complex constraints like pricing and quantity. This evolution aims to improve model effectiveness by ensuring training and inference feature consistency through a feature logging framework, addressing bandwidth, volume, feature name length, and selective logging challenges. The system logs features at inference time, uses a feature allow list, feature name aliasing via enums, and impression filtering with Flink for efficient data processing and storage.

All Uber trails
  1. Taming the ML Firehose: Scaling Feature ConsistencyThis post details the implementation of a feature logging framework to ensure consistency between ML model training and inference data at Uber Eats. It addresses scale challenges including bandwidth, volume, feature name length, and selective logging by implementing a feature allow list, feature name aliasing via enums, and impression filtering with Flink. The Flink job optimization involved profiling, custom state management with RocksDB, and aggressive eviction strategies to handle large state footprints and high throughput.
  2. Cart Assistant: Agentic Grocery Shopping on Uber EatsThis post details the system architecture of Uber Eats' Cart Assistant, an AI-powered feature that generates draft grocery carts from natural language or image prompts. It introduces a multi-prompt state graph architecture where each step (Cart Plan Generation, Candidate Retrieval, Semantic Relevance Judging, Price and Deal Constraint Enforcement, Quantity Selection, Guardrails, Cart Assembly, Content Refinement) has a focused responsibility. Key technical contributions include: 1. A structured output schema for the Cart Plan, separating retrieval language from reasoning context and capturing item-level and cart-level constraints. 2. A semantic relevance judging step with a constrained rubric to distinguish direct matches from substitutes. 3. A dedicated Quantity Selection stage combining LLM reasoning with deterministic calculations to map colloquial quantities to purchasable units. 4. A robust guardrail system (deterministic and LLM-based) to enforce boundaries and ensure safe execution. 5. Strategies for parallelism and latency reduction using dynamic asynchronous execution and mixed model sizes. 6. An evaluation-driven development framework for reliable measurement and improvement of the stochastic LLM system.