All Posts

ACI.dev reduces AI agent context bloat by replacing hundreds of tool schemas with two meta-functions that dynamically load only needed tools on-demand.

We've partnered up with Cognee to demonstrate how to build an HR Application Tracking System.

At AI Tinkerers London this week, we showed how to turn any agentic IDE into Lovable — simply by adding a single ACI.dev unified MCP server. We want to bring on the VibeOps and Vibe Experience Engineering movement.

ACI.dev is now fully open source. Learn how our platform enables AI agents to integrate with 600+ tools via a robust multi-tenant architecture, MCP support, and natural language permission boundaries.

Preliminary evaluation of the Mercury API, a large language model based on the diffusion model

An in-depth analysis of the technical limitations of Anthropic's Model Context Protocol (MCP)

The parallel between agent-computer interface (ACI) hybridization and biological sensory evolution.

Examining the parallels between generative AI agents and consumers