The paperwork between the print and the part doesn't have to be a human job. The old config-era way — macros, brittle scripts, a clipboard — is dead. The new paradigm is AI. AI automation is InTouch AI.
RFQs, drawings, outside processes, CoCs, shop status — you describe the workflow in your own language, AI runs it against your Gmail and Sheets. AI isn't a feature bolted onto this; it's the architecture. The vault, the scheduling, the audit log all sit behind it.
InTouch AI is a general-purpose automation platform. This page points it at small-shop manufacturing using example skills and jobs from the InTouch Hub. The same platform runs in every other industry and handles workflows we haven't published examples for.
The examples below kill the spec/quote/release/ship paperwork that lives between your front desk and the shop floor.
These are examples, not turnkey products. Every shop has its own quote log, traveler flow, and CoC template. But you don't start from a blank page — you start from something that works. Find it in the Hub, point it at your sources, run it. Read the README and YAML for each, set your placeholders, and you're live.
Email comes in from 15 customers a day. Some are real RFQs with prints attached, some are status checks on quotes you sent two weeks ago, some are change orders. Triaging by hand is a half-hour every morning.
Customer sends rev D on Tuesday. You're already running rev C since Monday. By the time anyone notices, you've got 40 parts you can't sell. Extracting the part number and rev from each PDF title block doesn't have to be a human job.
Heat treat lot at the vendor since the 14th, supposedly back yesterday. You call to check. Spreadsheet says nothing. The job needs them for the assembly tomorrow.
Every shipment needs a Certificate of Conformance. Same template every time, but each customer has tweaks: their PO number, your traveler number, the material lot, the heat-treat cert reference, the inspector who signed off. 15 minutes per shipment that doesn't have to be human.
Drawing calls out AMS 2750. Your QE manager knows it; the lead programmer doesn't. A fast structured summary of what the spec requires, what's typical, and what shops get wrong saves a half-hour Google search.
How many open quotes? Rush queue depth? Stale RFQs to chase? Shipments waiting on CoC? You know it by feel. A printed one-pager at 7 AM saves you a brain dump.
Reference and analysis tools that already speak aerospace, medical, and general-machining. Ask in plain words; get the answer.
Paste an AMS, ASTM, MIL-STD, NAS, ASME, or ISO spec number. Get a 60-second orientation: title, scope, key requirements, who invokes it, common pitfalls, verification trail. Refers you to the actual document for exact numbers (specs are copyrighted).
Paste a GD&T callout or describe a feature; get a plain-English breakdown of what the tolerance is asking for, how it's measured, and what often gets confused with what.
Customer calls out AMS 5604 (17-4 H1075). Stockless. What are the practical substitutes, what's the customer-approval risk, and what cert paperwork changes? Answers with sourcing context.
Sanity check a quote. Given the part scope and your hourly rates, returns hours/labor/material/markup breakdown plus risks that could blow the price. Pairs with your real estimating process.
Scheduled or trigger-driven. They read your Gmail labels and Google Sheets, do the work, and digest to your channel. You describe it once; it runs without you. This is the contract, now intelligent: tell it what to do, when, what to do when it works, what to do when it doesn't, and who to notify. The "doesn't work" clause stopped being a dumb retry — it reads the failure, refreshes the expired token, and tells you the one sentence that matters.
Polls Gmail for new RFQ emails. Claude classifies each one: customer, project, parts, rush vs standard, complexity. Logged to a QuoteLog Sheet, digest to the front office.
Watches Gmail for drawing-revision emails. Extracts part number + revision from each PDF title block via Claude, looks the part up in DrawingsTracker, archives the old drawing on rev change, alerts production. Catches rev changes mid-production before parts get made wrong.
When a ShipLog row flips to "ready-for-coc", generates the Certificate of Conformance PDF from a template + the row's data (PO, traveler, lot, inspector). Customer-specific tweaks supported via per-customer template overrides.
Tracks lots out for heat treat, plating, anodize, NDT. Alerts on lots stale past the promised return date. Pairs with your traveler workflow.
Once-a-day plain-text digest of open quotes, rush queue, stale RFQs to chase, shipments ready for CoC, CoCs sent today. Goes to the owner subscriber via the channel you've configured.
Weekly digest of quotes you sent 7+ days ago without a response. The owner decides who to call; the job is the nag.
Download the free Personal edition of InTouch AI. Browse the InTouch Hub, pick the jobs and skills you want, click Install. Or use the REST API.
Each job has placeholders (Sheet IDs, Gmail labels, credential names, recipient publishers). Read the README, set the placeholders, point at your actual sources. Verify locally before scheduling.
Attach a daily/hourly schedule and walk away. InTouch AI runs the job, sends the digest to your channel (email, Slack, Discord, Teams), and logs every run. When something breaks: "It broke. Here's why. I fixed it." And as a job earns your trust, turn the determinism dial — graduate it from AI-driven to fully deterministic, zero AI cost, identical every time, fully audited.
Straight talk. No surprises, no fine print:
Personal edition is free. Install on your shop-office PC, paste your service account JSON, point it at your QuoteLog and ShipLog Sheets. The morning triage is done before you walk in. A general AI-native engine does what a stack of specialized config-era tools does — and the reverse is impossible, because you can't grow an AI core onto a clipboard. Find it. Run it. Stop triaging.