Agent Lab: How We Built a Hybrid Team of Humans and AI Agents

At the World Economic Forum in Davos, Salesforce CEO Marc Benioff predicted that future workforces would be hybrid, combining humans and agents, which inspired a team to experiment with building their own hybrid team by integrating AI agents into their workflows. The team, led by Sam, embarked on a project to create AI agents using platforms like n8n, aiming to automate tasks and enhance productivity, with a dedicated Slack channel for communication and learning resources to support the initiative. They structured their efforts with clear goals, including making AI agents part of their quarterly objectives, which tied into performance bonuses, and utilized tools like Miro boards for idea pitches and feedback. Through various stages, they built diverse agents such as a personalized newsletter generator, a content workflow system, and a language model benchmarking tool, learning critical lessons on the importance of simplicity, effective tool connections, realistic scoping, and understanding processes before automation. While not all agents proved useful, some became valuable organizational tools, like Lucie’s LLM tracker, which evolved into a company-wide resource, and the team plans to continue developing AI solutions in their work.