@Chris Jadama I provide personalized brochures that can be sent out that contain all the information including stats + samples. The sales person can make use of it. From experience, if the average salesperson can't book a client on the first touch, prospects always ask for more information - that's when you sent them that (but only if they are really interested, otherwise they are time wasters). Current sales script (can be adjusted) requires good understanding of B2B sales and delicate touch (I know it works), and I have been booking 3-4 meetings per day without sending anything. I am open to other sales approaches - I have only cold called since then.

100% agree. I cringe when I see most automation builds. Security, compliance, protocols for when things go wrong/leaked/abused/... Organizations already realizing they still need to hire tech people for maintenance. It's funny because they laid-off the tech staff in favor of AI, then as they grow, they need to re-hire the same experts again. Same tech, different tools.

Good point. Short answer: Kinda. IMO, It still does not completely replace fully agentic workflows. You definitely don't need to be tech-savvy and understand n8n on an expert level, most agents can hook up to n8n and build the template for you - you just customize some nodes and set up credentials. I find it very useful though when shipping to clients. Shipping agentic workflows, markdowns, etc. feels kinda loose and with the insane growth speed of many models, outputs can (and will) vary in the future. Running Python scripts is recommended, but increases the complexity for debugging for others. I run workflows inside n8n and I have access to it at any time to configure and fix stuff for my clients. For my own work, locally, it's definitely easier and better to just do all with Claude Code + custom builds. I get the best of both worlds. Deliverability and maintenance for others is a deciding factor.

You are your first client. Make something more easier for yourself by automating. Does not need to automate 100%, manual input or review is often times needed to ensure quality. You do this a couple of times, and talk about what you build, share it with other people for free (if you really think it has helped you - chances are, people might have similar problems) > People will start liking, people will ask for business. Even better if you share it on YouTube, and dumb down your explanations so that a 5th grader can understand. Keep doing it. How many hours does it need to get proficient in playing an instrument? Correct... a lot. Don't expect that things work out after short feedback-cycles. I challenge you to invest as many hours, as you would learning a new language, a new instrument, a new skill, etc.

I am kinda both - tech person and "not". I also taught school students aged 10 to 18. Reality is that since popular AI adoption around 2023, even techy people had to learn how "transformers" and [put in every techy modern AI word] things work, but generally it's not needed to have a tech background anymore, even people at Anthropic stopped coding themselves (but it's good to know to understand how to debug and for other niche reasons). What matters is to understand how to realize an idea by properly constructing. If you can develop a sense of clean organization / role delegation, I think it's getting much easier: - Context: The buzzword - but rightfully so. I will skip the explanation what it is and get down to the gist of it: You have one agent doing one thing in a category - A teacher does not know it all, nor does an AI agent. Avoid having one agent memorize all the context from everything - because it can't and will hit his imaginary head against a wall eventually. What to do with limited context? In this case, setup an agent who teaches Math, an agent who teaches a specific language, an agent who teaches History, and so on - it's much easier to set up these agents with their own respective system prompts (what you tell the agent to be before any real work). - Orchestrator: The planner. Can communicate to the different agents (in your case "teachers") and has understanding of learning progress and can rate students ability to transfer knowledge / cross-context thinking. - Both teachers (your worker Agents) and orchestrator usually pull from a common database (could be a Google Sheet/Docs that both are synced to). That allows progress and notes to be tracked and planned and every teacher, at the beginning and end of a lesson, logs the progress in there. The orchestrator analyzes what needs to be done > sets up the plan in the shared document > the teacher agents know what to teach by pulling that from the document > conducting lessons at a set time > by end of lesson, teacher logs everything relevant back to the document.

@Joyce Lui sure whatever comes to your mind