Custom AI Agent / Automation Blueprint
Not one of our off-the-shelf agent SKUs — this is a bespoke build for whatever YOU have in mind. Tell us exactly what you want the AI agent or automation to do, the platform you plan to run it on, and the apps involved, and we produce a complete build blueprint tailored to your request. For automation platforms you get BOTH an n8n and a Make.com build spec (node-by-node / module-by-module), the exact paste-ready AI prompt the flow uses, a field-map and a 5-minute test checklist. For prompt platforms (Custom GPT, Claude Projects) you get the full system prompt plus the playbooks to run it. You build and run it once on YOUR OWN accounts, with YOUR OWN keys — it lives on your infrastructure forever. ECTD hosts nothing, stores no data, needs no ongoing access and provides no backend, database, login or live integration; the blueprint is yours to own and operate.
The Custom AI Agent / Automation Blueprint starts from your description rather than our catalogue. You say what the agent or automation should do, the platform you plan to run it on, and the apps it has to touch; the build is designed against that and arrives as a Markdown download on your order page in about 15 minutes. A$149, one-off.
What you are buying is a specification, not a service. You build it once on your own accounts with your own API keys, and it stays on your infrastructure. ECTD hosts nothing, stores no data, needs no ongoing access to your systems, and provides no backend, database, login or live integration — no recurring fee, no dependency. The flip side is plain enough: nobody from here logs into your n8n workspace and wires the nodes up for you.
Four free AI revisions come with it, which matters more than it sounds. A blueprint is only as good as the brief behind it, and a first pass will sometimes read a request slightly differently to the way it was meant. Say what missed, and it gets reworked.
Two automation specs, or one system prompt
Nominate an automation platform and the blueprint arrives as two parallel builds: an n8n spec written node by node, and a Make.com spec written module by module. Same logic, two ecosystems, so the platform decision can be made after reading both rather than before. Nominate a prompt platform instead — a Custom GPT, a Claude Project — and there are no nodes to specify, so the deliverable becomes the full system prompt plus step-by-step playbooks for running it. Either route includes the exact paste-ready AI prompt the build depends on.
Knowing which field feeds which step
Specs describe the steps; the field-map describes the data. It is a data-flow table showing which field in which app feeds which step, so connecting things up is a matter of reading across rather than guessing. Paired with it is a checklist built to take you from finished build to a working test in five minutes — enough to see the trigger fire and the output land where it should before real work goes through it.
Writing a brief for someone who cannot see your screen
The document is written from your description of your setup, not from your setup. It does not connect to your accounts, so it cannot confirm that your Make.com plan includes the modules it calls for, that your API keys carry the right scopes, or that the app you named still exposes the trigger you want. Those stay yours to check as you build. So name the apps precisely, say what should happen when a step fails, and say where the output needs to land. The revisions are better spent refining a build than re-explaining the original job.
Built for a job that is not on the shelf
The other agent SKUs here are defined in advance — a known job, in a known shape. This one is not. It exists for the build that is specific to how you work, where the useful version is the one shaped around your terms, your categories, your product. You describe it, and the specs, the prompt and the test checklist are personalised to that request rather than adapted from something pre-built.
What you get
- A complete build blueprint tailored to exactly what you described
- Automation platforms: node-by-node n8n spec + module-by-module Make.com spec
- Prompt platforms: full system prompt + step-by-step playbooks
- The exact paste-ready AI prompt your build uses
- Field-map / data-flow table for the apps involved
- "Test it in 5 minutes" checklist
- Data-safety note — your accounts, your keys, nothing hosted by ECTD
- Self-contained note: hosting, backend & live integration remain yours to run
- 4 free AI revisions included
Turnaround: ~15 minutes. Delivery: Custom build blueprint (Markdown download — self-contained, run on your own accounts).
Frequently asked questions
Will you build the automation and connect it to my apps?
No. This is a build blueprint you run yourself. You get the node-by-node or module-by-module specification, the prompt, the field-map and the test checklist, and you do the building on your own accounts with your own keys. ECTD provides no hosting, no backend and no live integration, and needs no ongoing access to your systems.
I haven't picked between n8n and Make.com yet — do I need to decide before ordering?
No. For automation platforms the blueprint covers both, with an n8n spec written node by node and a Make.com spec written module by module for the same build. Read them, then choose.
What if the approach turns out not to suit my setup once I start building?
Four free AI revisions are included at this price. Tell us what did not fit — a trigger your plan does not support, a field that does not exist, a step that should behave differently — and the blueprint is reworked. It remains an AI-written specification, so anything touching billing, customer data or contractual output is worth reviewing before you switch it on.
Is A$149 the whole cost?
A$149 is a one-off charge for the blueprint, with no subscription and nothing recurring from us. What the build itself consumes on your side is separate and yours: your n8n or Make.com plan, your API usage, your own app subscriptions. Those are worth checking against the apps and modules the spec names.