AI skill · Volume 1 · Workflow W3
How do you qualify an inbound TIC enquiry in minutes?
It reads an inbound enquiry into a fixed format, with a value band and a fit score, and drafts an acknowledgement in your approved language.
It is workflow W3 of the twelve in The AI-Augmented TIC Sales Machine, sold on its own as one AI skill file (.zip) with a two-page start-here guide. Everything it writes is a draft for a person to check.
What you give it, and what comes back.
You attach: the enquiry and your approved language blocks.
You get back: An enquiry in a fixed format with a fit score, and an acknowledgement reply.
It does not: reply to the customer. You send the acknowledgement.
The enquiry, the teams you route enquiries to, your value bands and your ideal-customer criteria. For the acknowledgement, also your approved language blocks, the clarifying questions and the response time you commit to.
One run, start to finish
What it looks like when you use it.
Run in Claude (Opus 5, high effort), in our own Claude account with the skill installed on 20 September 2026. The firm, the routing rules and the enquiry are invented for this example, and the run is shown as it came back, unedited. Your assistant's wording will differ: AI answers change between assistants and between runs.
input · what was given to the skill
Our firm. Harbourline Testing Services. Teams we route to: electrical safety, EMC, food contact, key accounts. Value bands: A above EUR 25,000; B EUR 5,000 to 25,000; C below EUR 5,000. Ideal customer: household appliance makers selling into the EU and UK, repeat testing, 50 or more staff. Response time we commit to: one working day.
The enquiry, invented. "Hello. We are a small kitchen appliance start-up, four people. We have a prototype toaster and want to know what we need before we can sell it in Germany next year. We do not have a budget yet. Can someone call me this week? — Product lead, ToasterExample"
output · what came back
Draft, for review before any reply goes out.
what to notice
It refused to write the reply. Prompt 3.2 drafts an acknowledgement only from your approved language blocks, and none were supplied, so the skill stopped and asked rather than inventing a house style. That is the difference between a workflow and a chatbot.
The scoring is honest in both directions. It scores the enquiry low against your own criteria, then tells you not to bin it, and suggests the cheap way to serve it. Every empty field says "not stated" rather than a guess, and the one judgement it makes, the value band, is labelled as a judgement with the reason it could not be calculated.
The skill, or the report?
This workflow is not sold on its own. It comes with The AI-Augmented TIC Sales Machine, which sets out the method and the evidence for all twelve of its workflows, and every licence includes all twelve as ready-made skill files at no extra cost.