✅ Full Start To Finish AI Deployment Video - and some updates

𝐓𝐡𝐞 𝐅𝐮𝐥𝐥 𝐂𝐥𝐢𝐞𝐧𝐭 𝐁𝐮𝐢𝐥𝐝𝐨𝐮𝐭 𝐂𝐨𝐮𝐫𝐬𝐞

Okay, I have a video that im going to put out tonight - ill break it down into maybe a 2 / 3 parter & course on Skool. Its probably going to end up being a 2 hr full walkthrough but its an A - Z from scratch enterprise deployment.

Building assistants
Full prompt and refinement process, using the voice lab to test the deployment and making changes using a task / protocol structure
Full 3rd party rag setup via buildship
Full audio to text to PDF and upload to the knowledgebase system
Deployment to a widget and customizing the widget

It literally goes through everything, not for just a conceptual buildout - but for a production grade AI deployment that I am doing for support. Literally everything, I just hit record and built it and tested it all on video.

Hopefully that helpful - its long, so if you dont want to watch thats fine, ill have a TLDR version in a document you can follow.

Along with this is a buildout checklist and worksheet in Google docs you can use like I did in the video.

𝐎𝐫𝐛 𝐅𝐮𝐧𝐜𝐭𝐢𝐨𝐧𝐚𝐥𝐢𝐭𝐲

Other updates, finishing up refining on the Orb - just to make a differentiation, I did just pushing basic tools and booking - it still needs to be refined to be considered prod (I would call it an indepth beta) becasue there are variable we need to account for in the ad blockers / cookie blockers to have errror handling for.

𝐂𝐨𝐧𝐭𝐞𝐱𝐭 𝐈𝐧𝐣𝐞𝐜𝐭𝐢𝐨𝐧 & 𝐁𝐨𝐭 𝐑𝐮𝐧

These tools were temporarily switched from an injected tool call to an injected prompt run for the server migration to maintain proper formatting. This will be patched tonight and with GHL actions you will get access to both - Context Injection (whats in there now) and Bot Run

Whats the difference?

Context injection is what is in there now, permanent memory storage for a conversation and a run of the AI - this will be elaborated on with the ability to not run the AI but still recieve the response in the workflow to use as you would like. There will be a check box - something like "dont respond" that you select and you will just get the response back.
Bot Run is what is actually in there now which is use to just run the AI - you can use this to run prompting via GHL custom value etc etc. Just like running a temporary prompt through an assistant to overwrite it for certain tasks.,

Okay, getting back to it - will send updates later

24 comments