Writer has released Action Agent, an autonomous AI designed for complex enterprise tasks. This 'super agent' connects to over 600 tools and services, coordinating tasks across an organisation's tech infrastructure. It aims to execute intricate, multi-step knowledge work from start to finish, incorporating research and computer use.
Action Agent is powered by an updated version of Writer's adaptive reasoning LLM, Palmyra X5. It achieved a score of 61% on the GAIA Level 3 benchmark, surpassing OpenAI's performance. It also outperformed computer use agents on the CUB leaderboard, scoring 10.4%. The agent is currently available in beta, complementing Writer's existing library of agents designed for repeatable team workflows.
Action Agent is designed to act as a trusted partner for individuals, assisting with strategic projects requiring creativity and precision. It enables users to accomplish more in less time by independently building solutions, such as detailed reports, data analysis, and fully functional webpages. It can browse the web, read and write files, analyse data and write and run code.
Related Articles
AI 'Hallucinations' Remain Problematic
Read more about AI 'Hallucinations' Remain Problematic →AI Agents Reshape Enterprise
Read more about AI Agents Reshape Enterprise →CBA Automates Call Centres
Read more about CBA Automates Call Centres →ChatGPT Agent: AI Autonomy
Read more about ChatGPT Agent: AI Autonomy →