Want to make money and save time with AI? Join here: https://www.skool.com/ai-profit-lab-7462/about
Video notes + links to the tools 👉 https://www.skool.com/ai-profit-lab-7462/about
Get a FREE AI Course + Community + 1,000 AI Agents 👉https://www.skool.com/ai-seo-with-julian-goldie-1553/about
Order AI SEO here → https://go.juliangoldie.com/backlink-portal
Free GEO & AI SEO Checklist! https://orders.goldie.agency/signup
Get a FREE GEO + AEO Strategy Session → https://go.juliangoldie.com/strategy-session?utm=julian
Get 200+ Free AI SEO Prompts → https://go.juliangoldie.com/chat-gpt-prompts?utm=julian
Get our free SEO Link Building Book here: https://go.juliangoldie.com/opt-in?utm=julian
A newer AI model won’t fix a coding agent that keeps breaking things, because the model usually isn’t the problem. Based on Google Cloud’s crash course, this video covers the four parts of the “harness” around the model: memory, tools, sandboxing and evals. It also shows how to set each one up so your agent stays reliable no matter which model you switch to.
00:00 Intro – Why the model isn’t the problem
00:28 The Harness – Model = brain, harness = everything else
00:52 Assistant vs Agent – Why the loop matters
01:29 #1 Context & Memory – Session, state, long-term memory
02:12 agents.md – Stop repeating yourself
02:41 #2 Tools – Python functions & MCP servers
03:50 Workflows – Google ADK graph workflows
04:25 #3 Sandboxes – Contain risky code
04:40 Prompt Injection – Treat outside content as data
04:57 Permissions – Human approval & command hooks
05:33 #4 Evals – The step most people skip
05:48 3 Eval Layers – Code checks, LLM judge, human review
06:43 Key Takeaway – Build a harness that outlasts models
07:00 Start Today – 4 quick wins
Video notes + links to the tools 👉 https://www.skool.com/ai-profit-lab-7462/about
Get a FREE AI Course + Community + 1,000 AI Agents 👉https://www.skool.com/ai-seo-with-julian-goldie-1553/about
Order AI SEO here → https://go.juliangoldie.com/backlink-portal
Free GEO & AI SEO Checklist! https://orders.goldie.agency/signup
Get a FREE GEO + AEO Strategy Session → https://go.juliangoldie.com/strategy-session?utm=julian
Get 200+ Free AI SEO Prompts → https://go.juliangoldie.com/chat-gpt-prompts?utm=julian
Get our free SEO Link Building Book here: https://go.juliangoldie.com/opt-in?utm=julian
A newer AI model won’t fix a coding agent that keeps breaking things, because the model usually isn’t the problem. Based on Google Cloud’s crash course, this video covers the four parts of the “harness” around the model: memory, tools, sandboxing and evals. It also shows how to set each one up so your agent stays reliable no matter which model you switch to.
00:00 Intro – Why the model isn’t the problem
00:28 The Harness – Model = brain, harness = everything else
00:52 Assistant vs Agent – Why the loop matters
01:29 #1 Context & Memory – Session, state, long-term memory
02:12 agents.md – Stop repeating yourself
02:41 #2 Tools – Python functions & MCP servers
03:50 Workflows – Google ADK graph workflows
04:25 #3 Sandboxes – Contain risky code
04:40 Prompt Injection – Treat outside content as data
04:57 Permissions – Human approval & command hooks
05:33 #4 Evals – The step most people skip
05:48 3 Eval Layers – Code checks, LLM judge, human review
06:43 Key Takeaway – Build a harness that outlasts models
07:00 Start Today – 4 quick wins