SmartInferVerifiable AI systems for high-value problems.
SmartInfer builds specialist AI systems for consequential decisions and actions. Our leading systems are Galileo for marketing measurement and allocation, and Monel for continual software engineering.
We focus on problems where the cost of error is clear and the impact of improvement can be measured.
In private pilots
Galileo — the Marketing Scientist agent
Decision: How should the next marketing budget be allocated?
Galileo combines an identifiability-aware marketing mix model with an evidence-grounded AI agent. Unlike an MMM judged by historical fit alone, it asks whether the evidence supports moving the budget. It estimates channel response, uncertainty, and budget scenarios, and withholds reallocations when the data cannot separate channel effects. The agent lets teams interrogate the computed analysis and explore model-backed scenarios.
Continual software engineering
Monel — continual software engineering agent
Objective: Build software from the ground up and become better at engineering it as it evolves.
Monel is an engineering system designed to build new software and improve existing codebases. It explores competing solutions, independently verifies changes, and carries engineering experience forward.
The research goal is to turn accumulated experience into better engineering on the same evolving system. Experiments and results will be published on Monel’s site as they become available.
In development
Retail specialists
-
Shopping Associate AI agent
A conversation and action engine designed to behave like a human shopping associate.
It talks naturally with shoppers, asks and answers questions, searches the catalogue, compares suitable products, and uses permitted retailer tools to help process orders and returns.
-
Product Discoverability Agent
Transforms a catalogue or individual product listing into a cleaned, enriched version.
Structures the result for search engines, AI shopping systems, and LLM-based interfaces, with outputs formatted for platforms such as Google, Amazon, and Walmart.
Founder track record
Built on two decades of production ML and AI
SmartInfer was founded by Anjan Goswami following two decades building and leading production search, ranking, and AI systems at Walmart, eBay, Adobe, Salesforce, Amazon, and Microsoft. Most recently, he led AI for PowerPoint Copilot. He holds a PhD in computer science from UC Davis.
Selected pre-SmartInfer work:
- +23% e-commerce search revenue lift — Walmart
- Replaced a 15-year heuristic ranking system with ML — eBay
- Led AI for PowerPoint Copilot — Microsoft
Selected pre-SmartInfer production work → anjangoswami.com →
Contact
Work with SmartInfer
Galileo pilots, Monel research, design partnerships, or other questions.