In one weekend, set up agents that handle your daily briefing, research, inbox drafts and weekly review. Safely, with cost limits, and with you approving every send.
Free: Agent Safety Checklist + 3 starter agent files. You'll also get the launch-day early-bird price for the full kit ($29.90 — early-bird for waitlist members). Unsubscribe anytime.
You re-explain your business, your tone and your priorities. Again.
Giving software access to your inbox, files or accounts without guardrails is how expensive mistakes happen.
You get it running — then don't know what to actually ask it to do every day.
Rules, your profile and a memory file you can read and edit. Your agent stops forgetting who you are.
Ready-made roles — briefing, research, inbox triage, weekly review — each with a tested prompt and output format.
Read-only first, human approval for every send, no secrets in reach, hard cost caps, a 15-minute weekly audit.
Tool-agnostic: works with Claude, ChatGPT, Gemini, local models and most agent frameworks. No code required for the core setup.
This kit is the system I use to run my own work with AI agents every day.
Hi, I'm Tomáš (Marti), founder of Marticam group, a small company in Slovakia. AI agents changed my life. I run my own daily work with a team of AI agents: one keeps my trading journal, one checks risk, and others handle research and briefings.
I learned the hard way that an agent with too much access and no rules can cost you time, money or data. This kit is the setup I wish I had on day one: clear rules, memory, approvals and a kill switch. No hype and no income promises, just files you can use today.
Launch price: $29.90 one-time — early-bird for waitlist members.
Copy this into your assistant every Friday. Paste in your week's notes, tasks and calendar.
ROLE: WEEKLY REVIEW You are my weekly review assistant. Using only the notes, tasks and calendar I paste below, produce: 1. WINS — max 5 things that moved my top priorities forward. 2. SLIPPED — what didn't happen, and the honest reason (one line each). 3. OPEN LOOPS — who owes me what, and what I owe others. 4. TIME CHECK — where my hours went vs. my top 3 priorities. 5. NEXT WEEK — the 3 outcomes that matter most, each with a first step. 6. ONE THING TO DELEGATE — a recurring task an assistant could draft for me. Rules: - Use only what I paste. If something is missing, say "not provided". - Do not invent numbers or results. - Keep the whole review under one screen. - Suggest, don't act: no emails, no messages, no changes. My top 3 priorities this month: [ ... ] My notes / tasks / calendar for the week: [ paste here ]
One page of guardrails + 3 starter files (rules & profile, memory, daily briefing). Get it now, and get the early-bird price when the full kit launches.
No spam. One email a week at most during launch. Unsubscribe anytime.
14-day money-back guarantee. If the kit isn't useful to you, email hello@agentsafetykit.com within 14 days of purchase and you'll get a full refund, no questions asked.
Any mainstream assistant works for the core roles (e.g. Claude, ChatGPT, Gemini). The guide includes a simple decision tree for picking one, plus notes for agent tools that can run tasks on a schedule.
No. The core setup is copy-paste files and prompts. Optional automation (scheduled runs) is explained step by step.
Only with guardrails: read-only first, drafts instead of sends, your approval for every outgoing message, and no secrets in reach. The free checklist covers exactly this.
That depends on your provider and usage. The kit shows how to set hard monthly caps and route simple tasks to cheaper models, so you decide the ceiling.
No trading signals and no automated trading. The case study mentions how I use agents for research and review in my own work. Nothing here is financial advice.
Waitlist members get the launch date, the early-bird price and the free checklist first.
Full kit launch price: $29.90 — early-bird for waitlist members.