Hi everyone, I’d like some advice on the project I’m describing below. For the past weeks, I’ve been building a WhatsApp chatbot demo in n8n to handle appointments and answer FAQs for a medical clinic. I’m using Google Sheets and Google Calendar. I don’t have much experience building this kind of thing in n8n, and I’ve noticed that the AI ​​model (Gemini) makes several errors in its responses. It mixes up dates and times, and sometimes claims there are no free slots when there actually are. I know this is part of the troubleshooting and fine-tuning process, but I already have a fairly extensive System Message with many rules designed to prevent these errors, and it keeps making them. So, my question is: for a WhatsApp chatbot, is it better to use fixed options (like a menu) to avoid these errors, rather than having the AI ​​respond with free-form text? My goal is really to give customers the feeling that they’re talking to a person, not a robot. However, if the AI ​​is prone to so many errors, I’m wondering if it would be better to use buttons, fixed menus, or a hybrid approach?m What do you recommend? Thanks in advance.

I'm Aatar Atta, and I help businesses automate operations, reduce costs, and scale with AI. I've joined this community to learn from Brendan and its members, to share my knowledge and contribute to making this community more valuable, and to network with like-minded people.

One thing I keep noticing in AI automation: People sometimes use an AI agent where a simple workflow would be better. For example: Form submitted → validate data → update CRM → send confirmation That probably doesn’t need an autonomous agent. But: Customer asks a complex question → search knowledge base → understand intent → decide what information is needed → use tools → respond → escalate if necessary Now an agent starts making more sense. So I’m curious: **Where do you personally draw the line between a workflow and an AI agent?** Would love to hear how other builders are deciding this in real projects. ---

Same model. Same task. Same success rate. Half the bill. A new study from UC Berkeley and Arena put a name to something many of us building with AI agents have felt but rarely measured: the harness tax. The harness is the agent wrapper around the model: Claude Code, Codex CLI, Pi, and others. It turns out the harness can change your inference costs as much as the model you pick. A few findings stood out: → Claude Fable 5 solved ~97% of tasks in Claude Code, Codex CLI, and Pi. But Claude Code cost $1.33 per rollout, while Pi cost $0.67. → On SWE-bench Lite, Claude Code cost about 2x more than Pi and 1.6x more than Codex. Success rates were within 2 percentage points of each other. → Much of the gap appears before the agent does anything. Claude Code starts with 27,000+ tokens of context, compared with ~2,000 for Pi. → Models don't always perform best inside their own vendor's harness. Simple setups are often surprisingly competitive. The most interesting part for me: even when two setups post identical benchmark numbers, they often fail on completely different tasks. A leaderboard score won't tell you which one fits your codebase. The researchers' advice is practical: 1- Test a few model and harness combinations on your actual engineering workload 2- Measure cost per solved task, not raw cost per run 3- Pick the cheapest option that clears your reliability bar 4- Retest whenever the model or harness changes If you're running agents at scale, whether for your own product or for clients, this is the difference between a margin and a leak. The best agent setup isn't the most elaborate one. It's the one that solves your problems at a cost you can sustain.

I would not sell founder weekly operations as 'AI automation.' For Brendan's AI Community, whose audience is AI voice agents, Claude Code, and n8n; 26.9k members, I would center the lesson on implementation boundaries and recovery evidence. I would sell a specific outcome: a founder operations queue covering follow-ups, calendar conflicts, forms, and approval-ready actions. The operating workflow would review connected inbox and calendar context, prepare a weekly action queue, continue longer research in the background, and request approval before sensitive actions. The reason Muse fits this test is a dedicated secure VM, connected apps, background work, sensitive-action approval, and an audit trail. The implementation question is whether state, credentials, retries, and the failure path remain observable. A sensible starting offer is a one-time setup and operating-playbook package; commercial operating terms must be verified before sale. This is pricing logic, not a guaranteed result. Human control remains visible: Access is least-privilege, sensitive actions require approval, and the audit trail is reviewed weekly. What would an agent need to prove before you connected your inbox or calendar?