@Scott Graham have you tried reporting the video for sharing unlawful content? These types of videos are dangerous. It takes my session ~2 minutes to scrape 20 videos of this type and instantly integrate the teachings into a workflow that is ready to use on command. I think the truth here is that humanity is simply going to have to learn how to defend themselves and protect themselves from wrongdoer’s. Equivalent in learning how to effectively wield a sword and shield during the dawn of medieval times, etc. 🏰⚔️🛡️We need to start teaching this as mandatory classes in school, starting at grades as young as 6th grade. This should have been a mandatory requirement for education systems, yesterday. If at least 4 of 10 people are highly proficient in Cybersecurity (specifically AI tech) and 3 of 10 have the general understanding/exposure, that increases not only our ability to stay secure but also promotes increased cyber defence innovation, with 7/10 people educated in the minimum level of academia. I also believe that this should be one of any government’s top priorities, dedicating funding to this education campaign. Personally, I would not hire anyone that isn’t already, or wouldn’t be willing to become educated and proficient in AI Cybersecurity. The risks involved and the exposure of naivety is not merely a chance at this point in time, but a guarantee. The almost laughable fact about all of this is that in ~20, maybe 30 years MAX, quantum compute + AI systems will make any innovation or technology obsolete, and that goes both ways. It will be asymmetric to yin and yang ☯️ at a cosmic level, two opposite forces continuously pushing into the other, likely locked at a stand still with minor moves in either direction. It’s impossible to know for sure, but that’s what I feel will be most likely. Simply by following the Law of Correspondence, we can nod to this as being a reasonable assumption.

Similar name (agent0 vs Agent Zero) and not much information on the system itself (is it a platform or is it a training/learning methodology). The concept of Agent 0 is not novel, nor was it invented by Agent Zero AI, however the concept was capitalized on and implemented via the Agent Zero software & platform. So, while I’m sure there are many different “versions” or “types” of workflows and/or systems that also use the agent 0 concept, the Agent Zero AI platform is still unique as a program. I hope this clears things up! If even just a little bit. P.S. I don’t think anyone would be surprised to see copycats pop up, but you know what they say to that regard; it means we are doing something right.

To Scott’s point, thats an inevitable truth. In the meantime, my recommendation would be to integrate a cost & token efficiency throttle directly within the agent.system.main.role.md file of your Agent Profile, perhaps utilizing the A0 MCP or A2A in order to work from various models through triggered calls of which trigger a model switch when a task does not need the high complexity reasoning and compute that Claude uses throughout your building processes. I am sort of brainstorming here so this is a very brittle solution however perhaps it can provide some inspirational thought that will lead you to a solution that works best for you. I think most importantly is utilizing A0 memory as effectively and efficiently as possible, so that even if you decide you’re willing to spend the $ for an initial (first time) action, build, task, etc; the complex token-consuming reasoning does not have to be repeated in future scenarios, but rather can be pulled from memory skipping maybe ~30% of token use. Over time this % should theoretically increase as A0 memorizes more of the “how”, “where/when” (application of how), and I would think over time you would be looking at a 50% decrease in token in/out usage costs. Test the idea with a low cost model for PoC and refinement in a different container, play around with efficacy of including in directions a “Least Token Use Approach”, “without sacrificing quality”, “utilize memory for the most resource consuming areas of a task”, “for the purpose of adherence to Least Token Usage protocol”. Again all just ideas off the top of my head, I would also most definitely set up a Dashboard UI that tracks everything, providing valuable insights into where the resource exhaustion is occurring. Etc.

@Jordan Urbs not a problem at all! Happy to assist where able. Look forward to hearing how any testing results turn out!

@Jan Tomášek Perhaps down the road we can work together to build, develop, train, and utilize a model that is built BY Agent Zero, FOR Agent Zero, and use the same Venice API token staking mechanism for access. Ive thought about this a bit recently, we would have to have a huge discussion about the training materials and optimization for Agent Zero-type functionality/capability. (Just one of my many, many, … many floating thoughts)💭 😅 I’ve been finally back into actual A0 usage now that all the months of prep has been done and I missed it so much. I’m really beyond impressed and happy with every single feature, change, edit, and other decisions that the team has made during my time away. Great, GREAT work Jan (and team). I’m going to do my best to integrate back into community calls now that I have an hour or two of my workday that I can spare.

Multi-Agent Orchestration of Specialized Variants allowing for even the most simple input to result in a full-on factory of planning, designing, testing, and production. Like having an entire enterprise quality corporation capable of delivering anything you need without using more than one sentence to inhibit the process, let alone what an entire paragraph can produce. A key agent role to include in the workflow is an AI Linguistics Translator and an R&D Role Architect among a few other key workflow cycle agents that ensure a task or project does not fizzle out or run into paradoxical situations. I’ll share my work once I have reached a point at which I am willing to do so! Almost there. 2 - 3 months I recon and then I’ll begin to do walk-throughs.