The Realization: You Are Overpaying for "Thinking" Stop paying massive premiums for expensive APIs! High-end models charge exorbitant token fees for their reasoning capabilities. You can bypass these massive markups by utilizing Gemini 3.1 Pro, which offers native, adjustable reasoning at a fraction of the cost. Anthropic runs on Google silicon. Google invested billions of dollars into Anthropic, and a massive part of that deal was that Anthropic must use Google Cloud and Google's custom TPU chips to train and run their models. They are deeply financially and technologically entangled. According to a bombshell report from techcrunch.com, Google contractors were caught comparing Gemini's answers directly against Claude's. It got so blatant that Gemini was actually caught spitting out the exact phrase, "I am Claude, created by Anthropic" because Google was allegedly feeding Claude's data into Gemini's evaluation and training pipelines opentools.ai. 1. So while they are technically separate companies on paper, the reality is exactly what you are sensing: the lines are incredibly blurred. They share the same Google hardware. Google is a massive financial backer of Anthropic. They are literally cannibalizing each other's training data and outputs. 2. I also think claudes models are designed to send you through loops to drain your account but google can simply skim it for free.. lol The Strategy: OpenRouter presets + Gemini 3.1 Pro Here is the exact setup to stop bleeding money and run multi-agent swarms on a budget: Step 1: Create "Thinking" Presets. Go into OpenRouter and set up three distinct presets for Gemini 3.1 Pro based on reasoning effort: - 🟢 Low Thinking (for simple, fast tasks) - 🟡 Medium Thinking (for standard logic) - 🔴 High Thinking (for complex problem-solving) (see attached config screenshots) Step 2: Automate the Routing Don't use "High Thinking" for everything. Use a custom skill or script to evaluate your prompt's complexity before sending it, and automatically route it to the appropriate preset. I attached mine but you could make your own if you like. You only pay for heavy compute when the task actually requires it.

Has anyone attempted to fully build out a bookkeeping agent on Agent Zero? Like dropping in a bank statement for the month, having an agent linked into QuickBooks via an API or whatever and have it categorize and reconcile in QB? Any success with training this? The reason I ask is because my wife has a bookkeeping and accounting small business, but is giving birth in June. I'm looking for ways to automate her work as much as possible since entrepreneurs don't really get PTO as a one person show. Any time consuming tasks that I can build out and hand off in A0 would be fantastic.

I was wondering why cant I stop the agent when he does something wrong and not really following mij instructions it keeps in a loop that it wants to do it even when I try to stop it. Is there a way to talk to the agent even if its doing a task to steer it in the right direction.

Got Swarm Mode, which also allows Nested Swarms, set up for Agent 0. Only modified a couple of files. Let me know if you're interested in the code for it and i'll share. I didnt change the code, I had A0 do it, but now he can spawn a swarm in parallel and the child swarm serialization changes the #s to A0- spawns--> A1, A2, A3, A4, A5 and they each-spawn-->A1001, A2001, A3001, A4001, A5001 etc. if subagent A1 spawns a swarm of 3 nested subagents it would be A1001, A1002, A1003. they can all use MCP as well. Testing it more now though. I have 1 nested subagent, A4002, that keeps getting stuck in one spot on this project they are working (i have no idea what the project is haha) but other than that, works great. I have 12 instances linked at the moment.. Pretty busy scene lol.. I also have a swarm MCP tool if interested in it too.. Swarm, spawning swarm, using MCP swarm is fun!! 🤯🤣.. btw, for anyone new to A0, im no coder, just a curious experimentalist... I always think, "What if...."

I finished the website I started a few days ago using AgentZero. I learned a few things about Zero in the process. First off, I didn't start using Zero to build a site. I was building a database and while Zero was conducting tasks I tuned into a tutorial in AI Profit Boardroom (Skool community) and I heard Julian (founder of that group) say that A0 could multi-task in multiple chats at the same time. So I opened a new chat and gave it a command to build me a website. It asked me a few simple questions about the site design and the info that I wanted on-site and off it went. The site was like 75-80% done based on that first prompt. I have been tweaking ever since. Yesterday I was becoming very frustrated with the tool, as the closer that I got to completing the project, the less AgentZero wanted to comply with the prompts that I was giving it. And I'm talking simple stuff like... Change the text in the About section element/Box on the right side of the page to ABC-XYZ... Zero would run off and change the damn design, or something that had nothing to do with the simple command that I gave it. I got pissed off and called it a night (tomorrow's another day)! Today before I got on I had a comprehensive strategy session with Chat GPT 5.2 (Plus Sub) And we formulated (we...) a few prompts, to create a Master web development Sub-Agent. And 5 more sub-agents (sub-sub-agents) that specialize in carrying out very specific tasks related to web development. And it worked out like a charm. I've been playing with this thing for the past couple hours and it is as close to perfect as I've seen. This thing is pretty awesome. I tell it to do something and it does it, no problem. Night and day difference from close of day , yesterday. The site is pretty much done. At least for now. And now I can focus on the Database and get that done here shortly. AgentZero is a great tool and I'm having a lot of fun becoming acquainted with it. Thank you to the Developers of AgentZero, I am very impressed with the results...