Personal AI Agents with OpenClaw and Codex is a hands-on course in setting up an always-on personal AI assistant that you message through Telegram or LINE, with scheduled routines, email and calendar access, Codex for file work and sensible security controls. It suits professionals, business owners and power users, and you leave with a working assistant for your own routine tasks.
Chat-based AI assistants answer questions well, but they still wait for us to open the app and type a request every time. Routine work such as summarising the morning's email, reminding us of appointments, following news on topics we care about or tidying the weekly report files stays on our plate. A new generation of personal AI agents such as OpenClaw runs on our own machine around the clock, takes instructions through the chat apps we already use, remembers the context of our work and acts on a schedule by itself, while OpenAI's Codex works methodically on files, documents and scripts on the computer.
This course has learners set up a working personal AI assistant in a single day. They install OpenClaw, choose a model that fits their budget, connect it to Telegram and LINE, define its memory and role, set up routine jobs with cron and webhooks, and connect email and calendar through skills. They then use Codex for file work and scheduled automations, and learn the security measures that an always-on agent with access to personal data requires. The course closes with each learner designing an assistant for their own work. (1 day, 6 hours, Beginner to Intermediate level.)
Course Details
The course runs for 1 day, 6 hours (09:00-16:00), with learners building their assistant step by step alongside the trainer. Beginner to Intermediate level. Unlike AI Agents for Business, which designs agents for business processes, this course focuses on a personal assistant that learners install and run themselves. OpenClaw is free and open source, but learners pay for model usage through their own ChatGPT plan or API key. Codex is available according to the learner's ChatGPT plan. Running the assistant around the clock needs a spare machine or a VPS at the learner's own cost. Test email and calendar accounts are recommended during the course. Learners take home the course handbook, sample configuration files and prompts, and a security checklist.
Day 1: Setting Up a Personal AI Assistant That Really Works
No public rounds are open right now. Join the waiting list and we will contact you first when the next round opens, or ask us on LINE. Or call 02-570-8449 or 088-807-9770
A chatbot answers when you ask and stops at the end of the conversation, while an AI agent takes a goal and carries out several steps on its own, using tools such as email, calendars, files or the web. A personal agent goes further: it runs continuously on a computer or server, remembers its owner's context, performs scheduled routines and takes instructions from your phone through chat apps such as Telegram or LINE.
Cron runs tasks on a schedule, such as summarising your email every morning or reporting outstanding work every Friday: you write the task as a clear instruction and set the timing. A webhook runs a task when something happens in another system, such as a new form submission or an alert; that system sends data to your assistant's URL and the assistant acts straight away. Test with trial accounts and have results sent back to chat before relying on them.
Prompt injection is hiding instructions inside content an AI reads, such as an email, website or document, to trick it into doing something its owner never asked for, like sending data out or deleting files. Reduce the risk by giving the assistant only the permissions it needs, for example reading email but not sending it, requiring confirmation before important actions, installing skills only from trusted sources and keeping test accounts separate from real ones.
It suits professionals with repetitive routine work, business owners and freelancers juggling email and appointments, IT teams and power users who want to trial a personal agent before using one at work, and ChatGPT users who want an AI that takes action. It is beginner to intermediate level, so you should be comfortable with computers, able to install software yourself and use the command line at a basic level.
You will be able to install and configure OpenClaw and choose a model, instruct your assistant from your phone through Telegram or LINE, set its memory and role, schedule routines with cron and webhooks, connect email and calendars with only the permissions needed, and use Codex for file and script work, with safeguards against prompt injection. Participants receive a Certificate of Completion with its own number that can be verified online.
Bring a Windows, macOS or Linux laptop on which you can install software, a ChatGPT plan that includes Codex or an API key from an AI model provider, a Telegram or LINE account, and email and calendar accounts for testing, ideally test accounts. OpenClaw is free and open source, but you pay for model usage yourself, and keeping the assistant running around the clock requires a spare machine or a VPS.
Organizations have mostly used AI in a question-and-answer style: a person asks, AI answers, and the person then does the follow-up work every time. Today AI has moved beyond an assistant waiting for commands to an AI agent that takes a goal in human language…
Traditional automation follows conditions written in advance, so it stalls on a message that does not fit the pattern or a question that needs data from several places. Agentic AI changes this by letting an LLM interpret the request, plan, and choose which…
Almost every Thai organisation uses LINE as its main communication channel, from team group chats to LINE Official Accounts, yet that communication is still almost entirely manual: posting announcements, summarising meetings from group chats, chasing tasks…
