A new paradigm for analytical systems

Introduction

Building industrial analytics involves more than writing individual calculations: teams also need to manage dependencies, validate changes, and move work into production. Traditional workflows often handle these concerns separately and manually. Uptimely takes a system-wide approach, describing the information and calculations needed so their relationships can be resolved as one analytical system. The comparison below highlights the differences.

Comparison

Paradigm	Traditional	Uptimely
Design principle	How, when, and where (imperative)	What (declarative)
Calculation code structure	Case by case	Standard
Code reuse	Moderate	High
Who can write functions	Data scientists	The person who needs them
Required skill level	Skill level defines who can write functions	All skill levels
Validating function changes	Many manual steps	Automatic
Function deployment	Manual	Automatic
Variable dependency tracking	Manual documentation	Resolved graph
From idea to production	Weeks	Minutes
Focus	Single calculation	The whole analytical system
Plan level	Physical	Conceptual
Deployment optimization	None	Automatic
Supported backends	SQL or Python	SQL and Python, extensible
Notebook role	Use for everything	Research and exploration
Roles involved	Data scientists and data engineers	Data scientists, data engineers, managers, site engineers, and sales
Benefits from AI	Partial, due to scattered resources	Strong, due to modularization