rebuilt.studio
§ The makers’ moment  ·  April 29, 2026

AI just made designing real things ten times easier.

The shops AI search can read get the work. The shops it can’t, don’t.

1 in 4
B2B buyers now prefer GenAI to traditional search for vendor research.
Digital Commerce 360, Oct 2025
82%
Perplexity citation rate for content updated in the last 30 days. Older content earns 37%.
Qwairy AI Citation Study, Q3 2025
1,000+
machine shops worldwide already running AI inside their CAM software.
CloudNC, October 2025

The thesis

Four things shipped between October 2024 and this week: ChatGPT Search, Google AI Overviews, a generation of text-to-CAD tools (Zoo, Backflip, Onshape AI, Adam), and — yesterday — Anthropic’s Claude–Fusion connector, which puts text-to-CAD directly inside the chat window an engineer is already using to think. Together they collapsed the design barrier for physical objects.

A founder describing a custom enclosure walks out of the conversation with a STEP file. A homeowner sketching a railing has a 3D rendering and a price quote in the same hour. A product engineer iterates on a sheet-metal bracket between meetings instead of between vendors. Hardware design used to be a week of CAD-designer time. Now it is a single afternoon, sometimes a single prompt.

The bottleneck moved. It used to be: can I afford a CAD designer. Now it is: can I find a shop that can make this on time.

The shift is not theoretical. CloudNC CAM Assist runs inside more than 1,000 machine shops worldwide. Frerotech, a Dutch precision shop, measures CAM-programming time savings of 60–90%, with peaks at 95%. Hadrian raised $260M in July 2025 to build AI-staffed factories in Arizona; their software, Opus, reads legacy hand-drawn blueprints and generates digital part programs in minutes. The plumbing is in place.

What is missing is the way in — the website that makes a shop legible to the buyer who started the conversation in a chat window.

Tools that moved the bottleneck
Claude + Autodesk Fusion connector
Built by Anthropic
Plain English inside a Claude conversation, geometry inside Autodesk Fusion, STEP out. Shipped April 28, 2026.
Visit →
Claude Design
Built by Anthropic Labs
A visual canvas inside Claude. Prototypes, mockups, and one-pagers from a prompt. Research preview, April 17, 2026.
Visit →
Zoo Text-to-CAD
Built by Zoo.dev
The reference implementation for the category. Prompt to STEP. Open-source CLI for embedding in any toolchain.
Visit →
Backflip
Built by Markforged founders, NEA + Andreessen Horowitz ($30M)
Text, sketch, or phone photo to a 3D-printable or CNC-ready model. Onshape native, STEP export.
Visit →
Adam
Built by Y Combinator
AI CAD agent for mechanical engineers. Sits inside the engineer’s existing CAD platform.
Visit →
CloudNC CAM Assist
Built by CloudNC
AI plug-in inside Fusion 360 that writes CAM toolpaths. Now used by 1,000+ machine shops worldwide.
Visit →
AI has 10–100x’d the productivity of software development but mechanical design is brutally tedious, slow, painful. Zoo can do this for hardware.
Andrew Côté, former CAD engineer, on X, January 2025

The catch

The new buyers only reach the shops AI search can find.

A buyer asks ChatGPT, “find a CNC shop in Indiana that does small-batch aluminum prototype work to ±0.0005,” and ChatGPT names three shops with inline citation links. If your shop isn’t one of those three, the buyer never knows you exist. The quote request goes to one of them. You don’t hear about it. You can’t bid on it. You don’t lose the work — you were never in the room.

Most small manufacturer and fabricator websites cannot be parsed by those engines. No schema markup. No FAQ block. Capabilities buried two clicks under About Us. Equipment lists locked inside PDFs. Certifications shown as image badges in the footer (AI parsers read text, not pixels). Mobile load past three seconds. The engines see thousands of these shells and trust none of them.

Hadrian’s Factory 3 opens in Mesa, Arizona in early 2026. Founded in 2020 by Chris Power and backed by a $260M Series C from Founders Fund and Lux Capital, Hadrian builds AI-staffed factories for aerospace and defense parts; their Opus platform turns legacy blueprints into digital part programs in minutes. The competitor for first-time custom-parts work is no longer the shop down the road; it’s an AI factory in the desert. The shops that win are the ones AI search can find first. TechCrunch, July 2025.

A 2025 study of 118,000 AI answers found Perplexity averages 21.87 citations per question, ChatGPT 7.92, and only 11% of cited domains appear on both engines. Two parallel indexes, both growing. A well-structured manufacturer page can show up in both; an unstructured one shows up in neither.

4.4× how much more valuable an AI-referred site visitor is than a traditional organic search visitor.
HubSpot State of Marketing, 2026

What changed, in order

Here is the eighteen-month run-up. Each entry is verifiable in primary sources. Read top to bottom; the inflection point is in the last three rows.

01 · Sep 2024

ChatGPT Search ships. Inline citations land inside the generated answer. Buyers click through to vendors directly.

02 · Dec 2024

Backflip launches text-to-3D, $30M from NEA and Andreessen Horowitz. The Markforged founders re-up on hardware AI.

03 · Jan 2025

Andrew Côté, a former CAD engineer, posts what becomes the line of the year: AI has 10–100x’d software productivity but mechanical design is brutally tedious. That ends now.

04 · Mar 2025

Google AI Overviews rolls out widely in the United States. The featured snippet is now an AI-generated paragraph, not a blue link.

05 · Jun 2025

Fortune profiles MSP Manufacturing in Bloomington, Indiana. Johnny Goode says programming an hour-and-a-half part dropped to seven minutes. First mainstream press for an AI-adopting small CNC shop.

06 · Jul 2025

Hadrian raises a $260M Series C from Founders Fund and Lux Capital. Factory 3 in Mesa, Arizona begins construction. Defense and aerospace start routing first-time custom work to AI factories.

07 · Oct 2025

CloudNC CAM Assist crosses 1,000 active machine shops. Frerotech in the Netherlands measures programming-time savings of 60–90%, with peaks at 95%.

08 · Nov 2025

Xometry posts a record Q3: marketplace revenue up 31% year over year, active buyers up 21%. The AI-mediated quote network compounds.

09 · Apr 17, 2026

Anthropic launches Claude Design. The visual canvas comes inside the chat.

10 · Apr 28, 2026

Anthropic ships nine connectors for creative tools. The Autodesk Fusion connector lets a buyer describe a part in plain English and pull manufacturable geometry out the other side. The category is now end-to-end inside one conversation.

11 · Apr 29, 2026

You are reading this page.

← Drag · 1 / 11 · Tap a date →

What to do

Get a website AI engines can read. The anatomy is short and specific:

Rebuilt Studio ships every line above by default. We rebuild your site first — built before you decide — and you only pay if you keep it. The build is scope-priced; most rebuilds land between $1,000 and $5,000. Hosted is $9.99 a month, install bundled.

We’ve taken on mission-critical parts for Fortune 500 companies, as well as the Department of Defense directly, and got them in and out of here in less than two or three weeks — which, prior to AI, was unheard of.
Johnny Goode, President, MSP Manufacturing (Bloomington, Indiana) — profiled in Fortune, June 2025

The shops that fix their site this year compound through the next two. The ones that wait catch up later, expensively, against shops that already own the AI search results in their vertical.

Built for shops like yours

What we’ve already shipped

Three live shop sites, each rebuilt first and now indexed by both Google and the AI engines. Open them on your phone.

Integrity Test Labs
Non-destructive testing
View live →
Allen Fab
Metal fabrication
View live →
Tuff Automation
Industrial automation
View live →

Common questions

Will AI actually drive more demand to small CNC and fabrication shops?

Yes, and it has already started. AI tools collapsed the design barrier for physical objects. Anthropic’s Claude–Fusion connector (April 28, 2026), Zoo Text-to-CAD, Backflip, and Onshape’s AI features all let a buyer describe a part in plain English and walk out with a STEP file. The bottleneck moved from "I need a designer" to "I need a maker." Small shops are the makers. New demand reaches CNC machine shops first, then sheet metal and fabrication, then specialty work — foundries, tool & die, precision machining — over the next 24 months.

How do I get my machine shop to show up in ChatGPT?

Three things, in order. (1) Make every capability scannable in plain HTML text — alloys, tolerances, equipment with make/model/work-envelope, certifications with cert numbers, lead times by job class. AI engines parse text, not image badges or PDFs. (2) Add FAQPage, Service, and LocalBusiness JSON-LD schema so engines can identify your shop as a vendor for specific kinds of work. (3) Keep the site updated within 30 days. A 2025 study by Qwairy found content updated within 30 days earns an 82% citation rate on Perplexity, versus 37% for older content — a 2.2x multiplier. Rebuilt Studio ships all three by default.

What does an AI-search-ready manufacturing website actually look like?

Front-loaded capabilities above the fold, a parsable equipment list with make/model/work-envelope, an alloy/material matrix with typical tolerance bands per material, certifications visible in the hero with cert numbers and expirations, an RFQ form that accepts STEP, IGES, and DWG uploads, separate URLs for each industry served (aerospace, medical, defense, automotive), FAQPage + Service + LocalBusiness JSON-LD covering the questions buyers actually ask, an llms.txt at the root listing the offer in plain English, a sitemap.xml with freshness dates, and sub-2-second LCP on mobile. Live examples: integrity-test-labs.rebuilt.studio (NDT), allenfab-com.rebuilt.studio (metal fabrication), tuff-automation.rebuilt.studio (industrial automation).

I run a fabrication shop with a website built in 2014. Am I about to lose work?

You are already losing work — you just don’t know which work, because the buyers who don’t reach you also don’t tell you. AI engines decide who shows up in answer results based on the structure and freshness of your site, not your domain age or backlink profile. A 2014 brochure site with no FAQ, no schema, capabilities buried two clicks under About Us, equipment lists locked inside PDFs, and certifications stuffed in the footer is invisible to ChatGPT, Perplexity, Claude with web access, and Google AI Overviews. The shops with parsable, recently-updated capability pages are taking the work that would have gone to you ten years ago.

How long does this AI manufacturing shift take to play out?

Eighteen to thirty-six months for the bulk of it. The first move was 2024–2025 with generative CAD and CAM tools (Fusion 360 Generative Design, CloudNC CAM Assist, Zoo, Backflip). The second is now: the Claude–Fusion connector shipped on April 28, 2026 puts text-to-CAD directly inside the chat window an engineer is already using to think. By 2027 most small-to-mid manufacturing buyers will start vendor sourcing in an AI assistant before they ever touch a search engine. Shops with AI-readable websites today compound through that window. Shops without catch up later, expensively, against shops that already own the AI search results in their vertical.

What is the cheapest way to get my shop’s website AI-ready?

For a single-location shop, $1,000 to $5,000 one-time depending on complexity gets a complete AI-search-ready rebuild. Rebuilt Studio quotes per scope and ships sites with the schema, the FAQ, the capability matrix, and the certifications block already wired — built first before you decide. Hosted at $9.99/month includes install and ongoing hosting. Local agencies typically charge $8,000 or more for the same scope and rarely include the AI-readability layer at all.

Why isn’t my shop showing up in ChatGPT or Perplexity searches?

Five overlapping reasons, in order of frequency. (1) No schema markup — the engines cannot identify your site as a vendor offering specific services. (2) Capabilities are not on the page in plain text — they are inside PDFs, image badges, or under nested menus. (3) The site has not been updated in 90 days — freshness is a documented ranking signal in Google AI Overviews and Perplexity. (4) Mobile LCP exceeds 2.5 seconds. (5) The domain is on a CMS that ships generic templates without customization (older Wix, older Squarespace, GoDaddy site builders) — the engines see thousands of similar shells and trust none of them.

What is the difference between SEO and AEO for a manufacturer?

Traditional SEO optimizes a page to rank in Google’s blue-link results. AEO (Answer Engine Optimization) optimizes a page to be quoted in an AI-generated answer where there may be no click at all. Both want fast pages, structured data, and useful content. They diverge on length (AEO is dense and short — 400 to 700 words per answer page; traditional SEO favors long-form), on lead structure (AEO answers in the first sentence; SEO can build narratively), and on which schema types matter most (AEO heavily weights FAQPage, QAPage, Service, and LocalBusiness; SEO weights Article, Organization, and BreadcrumbList). A modern manufacturer site needs both layers; one without the other leaves work on the table.

I have a Google Business Profile. Isn’t that enough?

It is necessary but not sufficient. Google Business Profile gets you into Google Maps and the local Map Pack — both still important and free. But it does not surface your shop inside ChatGPT, Perplexity, Claude with web access, or even most of Google AI Overviews’ generated answers. Those engines pull from your website, not from your Maps listing. Shops with only a verified Maps listing show up for "[trade] near me" geo-anchored queries; they do not show up for capability-anchored queries like "AS9100 shop with 5-axis Inconel capability in the Midwest." The new demand wave is mostly capability-anchored.

How do I check if my shop is showing up in AI search results?

Run the test yourself, every two weeks. Open ChatGPT (logged-in, web search on), Perplexity, Claude with web access, and Google. For each engine, ask 5–10 of the queries your buyers would realistically type — for example, "find a CNC machine shop in [your region] that handles 6061 aluminum prototyping," "AS9100-certified shops near [city]," "fabrication shop for stainless food-grade enclosures." Note which engines surface your shop and which do not. Tools like Otterly.ai automate this; a 30-minute manual sweep every two weeks gives 90% of the signal at no cost. If your shop never surfaces, that is the proof.

Sources

Run a CNC shop, fab shop, foundry, or any other shop that makes real things? Drop your URL. If you’re a fit, we’ll rebuild it first and email you the live demo. No call. No deck. Just the site.

See it on your shop →