The latest food and beverage industry news and trend analysis

Delivery Hero has unveiled Herogen, a proprietary autonomous software delivery agent designed to streamline product and engineering workflows across its global delivery platform. Built in-house using Large Language Models (LLMs), the tool has already demonstrated a technical output equivalent to 130 senior engineers, significantly accelerating the company’s development pace.

The launch marks a transition from manual AI-assisted coding to a fully autonomous model. Herogen is currently utilised by 18% of the company’s developers and is responsible for 9% of all code change requests across the organisation.

Autonomous Security and the Council of Agents

To ensure the integrity and quality of its codebase, Delivery Hero utilises a "Council of Agents" framework. While traditional AI coding assistants require constant manual oversight, Herogen operates by interpreting natural language instructions to write, test, and iterate on code independently.

The Council of Agents, comprising multiple LLMs from various providers, reviews Herogen’s output from diverse technical perspectives before a human developer performs a final check. This multi-model approach is designed to eliminate the risk of blind spots inherent in any single model's training data. Currently, Herogen maintains an 85% success rate, defined by the ratio of merged to rejected pull requests.

Operational Impact and Time Reclamation

By autonomously merging more than 100 code change requests per day, Herogen has reclaimed an estimated 250,000 hours of manual coding annually. This efficiency allows the existing engineering team to focus on higher-level system architecture and complex problem-solving rather than repetitive syntax tasks.

Benjamin Mann, Chief Technology Officer at Delivery Hero, stated that Herogen allows the company to achieve greater output with the same elite team. Mann emphasised that the tool handles mundane tasks and repetitive toil, allowing engineers to focus on building resilient systems on a global scale.

Evolution of the Software Engineering Profession

The deployment of Herogen reflects a fundamental shift in the interface between human engineers and machines. By moving from complex programming languages to natural language intent, the company is repositioning its developers from traditional programmers to software architects.

Key shifts in the engineering workflow include:

• Focus on Architecture: Engineers spend less time on syntax and more time on high-level system design.

• Rapid Innovation: Natural language commands allow for faster transition from concept to code.

• Human Oversight: Developers act as the final decision-makers and reviewers for autonomously generated proposals.

Future Integration and Global Reach

Delivery Hero plans to expand Herogen’s integration throughout the year, with a target of handling 20% of all code change requests by the end of 2026. As the company continues to develop its "Everyday App" for customers in 65 countries, the autonomous agent serves as a critical tool for maintaining a competitive lead in the quick-commerce and local delivery sectors.

Headquartered in Berlin, Delivery Hero continues to pioneer quick-commerce solutions, aiming to deliver groceries and household goods in under 30 minutes. The implementation of Herogen is a core component of the company's strategy to scale its technical infrastructure to meet these demanding delivery windows.