Human API reveals man-bites-dog approach to agentic AI

In a sea of agentic AI start-ups, Human API has a novel approach ... it has built an AI system in which the artificial intelligence gives tasks to humans, not the other way around.

Human API, a San Francisco outfit, in an announcement, said it had exited stealth mode as it unveiled the new platform that it says will allow AI to designate tasks directly to humans for project coordination and execution.

This approach will let the AI overcome “last-mile” human barriers that prevent real‑world interactions. The platform is built using Eclipse, which built a cryptocurrency-anchored 'virtual machine'.

Human API explains that the AI agent provides a coordination and execution layer, whilst humans in the system create accounts to browse and accept task assignments, before being paid via Stripe Connect. Examples described by the company involved data-oriented tasks.

Related reading

• Deepgram triples concurrency limits for AI voice agents
• Udemy to insert its courses directly into ChatGPT
• ElevenLabs activates insurance policy covering actions of its AI voice agents

During 'stealth mode', Human API said the platform contributed to a studio‑quality audio dataset for a leading AI lab. And, it intends that its initial business model will focus on licensing audio data to AI labs and that it plans to expand into other data types and task categories, including computer‑usage data and work requiring real‑world execution including logistics.

Human API noted that it has raised $65 million to date from investors including Placeholder, Hack, Polychain, DBA, and Delphi Ventures. It plans to use its funding to formalize agent‑human collaboration and create a marketplace where human skills can be monetized globally.

The Recap

• Human API launched a platform letting AI agents hire humans.
• The platform has raised $65 million from venture investors.
• Initial business model focuses on licensing audio data to AI labs.