OpenHuman
An open source AI harness built with the human in mind
2K followers
An open source AI harness built with the human in mind
2K followers
90% of people who try AI agents give up. Three reasons: memory that resets every session, your data sitting in someone else's cloud and a terminal just to get started. Real blockers. OpenHuman fixes all of it. Local-first, privacy-first. It remembers everything about you and actually gets smarter the more you use it. Every feature lives in one simple interface. Fully open source. One-click setup. P.S. The product is in beta, so expect bugs, but we're building and shipping fast.
OpenHuman
@aylin334194 It's as easy as pressing a few buttons and you get setup in 5 mins 😄
The gap between powerful agent and usable by normal people is still massive and most projects only solve the first half.
OpenHuman
@bruce_warren Hey Bruce, nice to see your comment.
You just described the entire reason we built this.
We've spent more engineering hours on installer, defaults, error messages, and recovery than we have on the LLM layer. So I agree when you state the gap between usability of agents and normal people.
So, I would like to say this one is different. It is something that anybody can use because of its easy and simple interface.
OpenHuman
@bruce_warren LFG thanks for the comment, give it a try and lmk what you think about it.
OpenHuman
@fatih919979 wow what a question. basically you can let OpenHuman handle all the chaos for you and just get summaries regularly basically. So it's an actual operating layer yes.
What is the subscription price to use this? Is it open source or need to pay anything?
OpenHuman
@agastya_patel it is completely open source and you choose to either run things locally in which it is as good as free. or use a cloud if you don't have decent enough hardware
Hi al! I am curious how the team thinks about the storage trade-off underneath the 【TokenJuice】. The raw feed from Gmail, Slack, and Notion is preserved in SQLite, and then a separate Markdown memory tree is generated for Obsidian. So the same corpus lives twice on a local disk. I ran a similar local-first RAG setup last year, and at around 20,000 chunked documents the query latency climbed from under 200ms to multiple seconds. The double index (database plus filesystem) was the bottleneck.
OpenHuman markets "automatic growth" as a feature. Has the team tested the local index at tens of thousands of memory nodes? And is there a path for users to prune the raw OAuth feed after compression without losing the distilled memory entirely?
I would love to hear how the makers are thinking about retention policies or compaction. Local storage is not a free resource after all.
The persistent memory angle is what makes this actually agentic rather than just a pretty chatbot wrapper. Session resets are one of the biggest reasons agents feel useless in practice. The dad story resonates too- accessibility is an underrated problem in this space. Curious how the memory layer handles conflicts or outdated context...if I told it something about my workflow six months ago that's no longer true, does it overwrite, append, or does the user have to manually prune? Also wondering how the local Gemma3 orchestration holds up when you're chaining several tool calls together? Does it struggle with longer reasoning chains compared to a hosted model?
GraphBit
How it's handle hallucination thing?
OpenHuman
@imrulkaayes Good question.
Three layers here:
First, OpenHuman grounds its answers in your actual data (emails, slack, notion, etc.) rather than generating from training memory alone.
Second, every memory chunk has a deterministic ID and we can show you the exact source for any claim the agent makes.
Third, when the agent isn't confident, it tells you and asks rather than guessing. We're not pretending hallucinations are solved, but grounding in your real corpus plus auditable retrieval cuts the worst of it.
Happy to go deeper if useful.
OpenHuman
@imrulkaayes The default prompts used by the agent already prevent halucination. And if you want to use a model that is completly yours and high end, you can easily switch to that as well. Most frontier models if prompted well do not halcuinate (much).