MaxHermes is a cloud sandbox AI agent that autonomously extracts reusable skills from completed tasks and improves itself across sessions. For knowledge workers and enterprise teams who want an AI assistant that compounds with use.
MaxHermes is a cloud sandbox AI agent from MiniMax that autonomously extracts reusable skills from completed tasks and applies them in future sessions.
Problem → Solution: The statelesness of AI agents is a design problem that the industry has mostly ignored. You do the work of teaching the agent your context, it performs, the session ends, and nothing carries forward.
MaxHermes introduces a learning loop at the execution level. Every complex task triggers automatic skill extraction. Those skill documents persist, get refined through feedback, and stack over time. The agent that handles your tasks in month three is meaningfully more capable than the one you started with.
What makes it different: Built on MiniMax M2.7, it achieves this without requiring any manual capability setup or server configuration. The evolution is autonomous.
Key features:
Learning loop that auto-generates skills post-task
Cross-session persistent memory
7x24 cloud availability with no infrastructure on your end
Integrations with Feishu, DingTalk, and Enterprise WeChat out of the box
Benefits:
Your AI assistant improves from real usage, not from prompt tuning
Zero setup cost for enterprise teams already in Chinese productivity ecosystems
Parallel task execution means it handles multiple workstreams without queuing
The compounding skill model is the bet worth watching here. Whether it holds up at scale and across diverse task types is the open question at launch.
P.S. I hunt the latest and greatest launches in tech, SaaS and AI, follow to be notified →@rohanrecommends
cloud sandbox approach makes sense for this. how does the skill library work across team members? wondering if there's a way to share learned skills between agents or if each instance builds its own knowledge base.
About MaxHermes by Minimax on Product Hunt
“AI agent that builds skills from every task you give it”
MaxHermes by Minimax launched on Product Hunt on April 28th, 2026 and earned 91 upvotes and 3 comments, placing #10 on the daily leaderboard. MaxHermes is a cloud sandbox AI agent that autonomously extracts reusable skills from completed tasks and improves itself across sessions. For knowledge workers and enterprise teams who want an AI assistant that compounds with use.
MaxHermes by Minimax was featured in Productivity (650.5k followers), Task Management (84k followers) and Artificial Intelligence (467k followers) on Product Hunt. Together, these topics include over 227k products, making this a competitive space to launch in.
Who hunted MaxHermes by Minimax?
MaxHermes by Minimax was hunted by Rohan Chaubey. A “hunter” on Product Hunt is the community member who submits a product to the platform — uploading the images, the link, and tagging the makers behind it. Hunters typically write the first comment explaining why a product is worth attention, and their followers are notified the moment they post. Around 79% of featured launches on Product Hunt are self-hunted by their makers, but a well-known hunter still acts as a signal of quality to the rest of the community. See the full all-time top hunters leaderboard to discover who is shaping the Product Hunt ecosystem.
Want to see how MaxHermes by Minimax stacked up against nearby launches in real time? Check out the live launch dashboard for upvote speed charts, proximity comparisons, and more analytics.
Rohan Chaubey
MaxHermes is a cloud sandbox AI agent from MiniMax that autonomously extracts reusable skills from completed tasks and applies them in future sessions.
Problem → Solution: The statelesness of AI agents is a design problem that the industry has mostly ignored. You do the work of teaching the agent your context, it performs, the session ends, and nothing carries forward.
MaxHermes introduces a learning loop at the execution level. Every complex task triggers automatic skill extraction. Those skill documents persist, get refined through feedback, and stack over time. The agent that handles your tasks in month three is meaningfully more capable than the one you started with.
What makes it different: Built on MiniMax M2.7, it achieves this without requiring any manual capability setup or server configuration. The evolution is autonomous.
Key features:
Learning loop that auto-generates skills post-task
Cross-session persistent memory
7x24 cloud availability with no infrastructure on your end
Integrations with Feishu, DingTalk, and Enterprise WeChat out of the box
Benefits:
Your AI assistant improves from real usage, not from prompt tuning
Zero setup cost for enterprise teams already in Chinese productivity ecosystems
Parallel task execution means it handles multiple workstreams without queuing
The compounding skill model is the bet worth watching here. Whether it holds up at scale and across diverse task types is the open question at launch.
P.S. I hunt the latest and greatest launches in tech, SaaS and AI, follow to be notified → @rohanrecommends