Tendem: A Hybrid AI+Human Platform

TL;DR

Tendem is a hybrid AI-human platform that combines AI efficiency with human expertise, achieving higher quality and faster results than AI-only or human-only workflows, with competitive costs.

Contribution

The paper introduces Tendem, a novel hybrid system integrating AI and human efforts, and demonstrates its superior performance on real-world tasks and benchmarks.

Findings

Tendem outperforms AI-only and human-only workflows in quality and speed.

Tendem's AI agent performs near state-of-the-art on web browsing and tool-use tasks.

Operational costs of Tendem are comparable to human-only workflows.

Abstract

Tendem is a hybrid system where AI handles structured, repeatable work and Human Experts step in when the models fail or to verify results. Each result undergoes a comprehensive quality review before delivery to the Client. To assess Tendem's performance, we conducted a series of in-house evaluations on 94 real-world tasks, comparing it with AI-only agents and human-only workflows carried out by Upwork freelancers. The results show that Tendem consistently delivers higher-quality outputs with faster turnaround times. At the same time, its operational costs remain comparable to human-only execution. On third-party agentic benchmarks, Tendem's AI Agent (operating autonomously, without human involvement) performs near state-of-the-art on web browsing and tool-use tasks while demonstrating strong results in frontier domain knowledge and reasoning.