Deployable AI Assistant Kit
A self-contained code scaffold for an AI chat assistant, pre-loaded with your business knowledge and system prompt. You (or your developer) drop your own API key in, deploy it to your own Cloudflare, Vercel or Netlify account, and paste the embed snippet onto your site. ECTD hosts nothing, holds no key and stores no chat data — you own and run the whole thing. Ships as a single index.html with a commented serverless-function stub for your chosen AI provider, an inline README, a .env.example and clear add-your-key → deploy → embed instructions.
The question this kit settles is whether your website chat assistant is something you rent or something you own. Rented widgets run on the vendor server, usually under the vendor API key, with the conversation logs sitting on their side and a monthly invoice keeping the lights on. The Deployable AI Assistant Kit takes the other side of that arrangement: A$199 one-off for a self-contained code scaffold, built in around 15 minutes and delivered to your order page as a single index.html chat widget pre-loaded with your business knowledge, a system prompt written from it, and a commented serverless-function stub for the AI provider you name. ECTD hosts nothing, holds no key and stores no chat data.
When the assistant does not know, it hands over
The system prompt generated from your knowledge carries three standing constraints. It identifies itself as an AI rather than passing as a person on your team. It does not invent prices. And when a query goes past what it knows, it hands the customer to the escalation contact you nominate at intake, which is the one field worth thinking hardest about before you fill it in.
Be clear about what those constraints are. They are prompt instructions shaping a language model, not a compliance layer or a hard-coded rule engine, and the assistant answers from the knowledge you supplied and nothing else. It has no line into your booking system, stock levels, calendar or CRM. Thin knowledge produces thin answers, so read what it says to real customers in the first week. Four free AI revisions are included at this price and they regenerate the kit, which is how you tighten a gap once you have found one.
Somebody has to add the key and push it live
This product ends at a file. You, or a developer you have access to, add an API key to an environment variable, push the scaffold to a Cloudflare, Vercel or Netlify account you control, and paste the embed snippet into your site HTML. The .env.example ships with a placeholder where your key goes, never a real secret and never one of ours, and the inline README and code comments walk the add-key, deploy and embed sequence in order. For anyone who has put a static site live before this is a modest afternoon. It is still a job, and it is yours. If nobody in the business is comfortable doing it, buy something else, because a scaffold sitting undeployed in a downloads folder is worth nothing.
The other half of that independence is what we give up. Chat traffic travels from your visitors through your serverless function to your AI provider and back, with no ECTD infrastructure anywhere in the path, so we cannot monitor whether the widget is up, cannot see what anyone asked it and cannot log in to fix a deploy that has gone sideways. Your AI provider bills your account directly for usage at their own rates and your host bills you under whatever plan you are on. Revisions cover the code we generate, not work inside your accounts.
What you get
- Self-contained index.html — chat widget, ready to deploy
- Commented serverless-function stub for your chosen provider
- System prompt pre-built from your knowledge (identifies as AI, never invents prices, hands off uncertain queries)
- Inline README + .env.example (your key placeholder — never a real secret)
- Paste-ready embed snippet + add-key → deploy → embed comments
- You host it on your own account, your own key — ECTD holds no key and no data
- 4 free AI revisions included
Turnaround: ~15 minutes. Delivery: Self-contained HTML scaffold (index.html — deploy it yourself).
Frequently asked questions
Do I need a developer, or can I do this myself?
If you have deployed a static site to Cloudflare, Vercel or Netlify before and know how to set an environment variable, you can do it yourself, and the inline README and code comments cover the add-key, deploy and embed sequence. If neither of those sentences means anything to you, you will want someone technical to handle it. ECTD does not deploy it for you as part of this A$199 purchase.
What does it cost to run after the A$199?
The A$199 is a one-off for the kit, with no subscription attached. After that, your AI provider bills your account directly for the API usage the assistant generates, and your hosting platform bills you under whatever plan you are on. ECTD sits outside both of those relationships, does not resell or mark up either, and cannot estimate your usage, because it depends entirely on your traffic and the provider you pick.
I nominated the wrong AI provider or deploy target at intake. Am I stuck with it?
No. The serverless-function stub is generated and commented for the provider you name, and the instructions are written for the Cloudflare, Vercel or Netlify target you chose, so a change means a rebuild rather than an edit. That is what the four free AI revisions are for: one of them can regenerate the kit for a different combination.
Can it take bookings, look up an order or check my stock?
No. It answers from the knowledge you supply at intake and nothing else. There are no integrations to your booking system, inventory, payment platform or CRM in this product, and building those is not part of it. Treat it as a well-briefed front-of-house responder that knows your business and passes anything beyond that to your escalation contact.