The add-on has been out a few days now, and the question I keep getting is a
fair one: what do you actually do with it? So here’s the whole thing, from an
empty Blender scene to a coaster with a cup of tea on it, in under seven
minutes and with nothing cut out.

Sizing it to the object

The slate blanks are 100mm square with rubber feet, and I wanted the engraved
area at about 88mm so there’s a margin of honest rough slate around the
pattern. A four-by-four grid at 20mm cells gives 80mm, which sits inside that
comfortably and keeps the arithmetic in my head rather than on paper. That’s
the whole of the setup: pick a cell size, pick a grid.

The part that’s actually design

Then the breaks. A plait is just a plait; it’s the breaks that turn it into a
knot, and where you put them is the design decision. In the video you can watch
me set one, dislike it, clear it and try somewhere else — which is the honest
version of how this goes. When it looked right I checked how many cords it
was, and it turned out to be two interleaved lines rather than one continuous
cord. That’s a real property of the pattern, not a rendering artefact, and it’s
the sort of thing the construction method gives you for free.

Getting it onto the machine

For engraving I switched the strands to relief, which flattens the weave into
outlines the laser can follow, and exported an SVG straight from the add-on in
millimetres.

Now, an honest note. My laser runs on LaserCut 6.1, which is old software tied
to the controller in the machine, and it won’t take an SVG. So I round-trip
through Illustrator and export a DXF. Most software made this decade takes SVG
natively and you won’t have to do this — but if you’re on the same vintage of
kit I am, know that DXF export is coming to the add-on so the detour goes away.

What did survive the trip is worth pointing at: every profile is a single
closed line with one start and one end point. No stacked duplicates, which is
the usual tax on vectors that have been through three programs. The laser
doesn’t retrace anything and you don’t pay for the time.

Eight minutes on the machine.

Try it

The free edition of Celtic3d Knotwork is not a trial. It’s flat panels, borders
and crosses, both break types, weave and relief, every export format, and
commercial use — and it doesn’t expire. No card, no licence key.

Start free →

Pro adds rings, cylinders and toruses and more strand profiles, £25 / $34 for
the first year across three machines.

See Knotwork Pro →

Blender 5.2 LTS, on Windows, macOS and Linux.

The post Designing a Celtic Knot in Blender and Engraving It on Slate appeared first on Celtic3d.

In the first post in this series we described the open-book exam: instead of letting an AI answer your visitors’ questions from memory, you hand it the relevant pages of your own website and ask it to answer from those. That post ended with a promise to explain the next piece of the puzzle — how do you know the AI actually read the book? The answer, when it works, is the little line at the bottom of a reply that says Sources.

This week that line has been on our workbench, because we caught it doing something subtly wrong. So this post is about what a citation actually is, why it matters more than it looks, and two real ways a Sources line can mislead — both of which we found in our own product, in our own chat logs.

What a citation actually is

When a visitor asks a public-facing AI assistant a question, three things happen in quick succession. A search runs over the site’s own documents and fetches a couple of dozen excerpts that look relevant. The AI writes an answer, drawing — if all is well — on a handful of those excerpts. And then, underneath the answer, the assistant lists the documents it drew on, each one a link to a real page.

That list is the citation. And a citation is best understood as a promise: this answer stands on that page — go and look.

Which is why citations matter more than their modest appearance suggests. A chatbot without sources asks to be taken on trust. A chatbot with sources can be checked — by the visitor, who can click through and read the page for themselves, and just as importantly by you, the site owner, reading your chat logs over coffee the next morning. Our working rule for the whole product is never take the AI’s word for anything you can check, and a Sources line is what makes an answer checkable at all.

But a promise is only worth something if it’s kept. Here are two ways we’ve watched it break.

The greeting that arrived with sources

Early in development, a visitor to our demo shop typed one word: “Hello.”

The assistant replied, perfectly reasonably, “Hello! How can I help you today?” And underneath that greeting, it printed: Sources: Hello world!, Clan Name Relationships.

It had cited the default WordPress post — the “Hello world!” page every new site starts with — as a source for saying hello.

The explanation is almost funny once you see it. The document search runs before every message, because at that point nothing knows whether the message is a question or a pleasantry. Asked to find pages resembling “Hello”, it dutifully returned the closest matches it had, and the AI just as dutifully listed them. Every one of those documents was real. The search genuinely fetched them. And the Sources line was still false — because the promise a citation makes is this answer is based on these pages, and the answer was based on nothing at all.

That’s the first lesson about citations: a Sources line can be built entirely from real documents and still be a false statement. The fix, once we understood the problem, was plain: small-talk replies now cite nothing, and our quality scoring stopped awarding marks for citations an answer never used. An empty Sources line under “Hello! How can I help?” is the honest one.

Twelve sources for one clan

The second failure is this week’s, and it’s the opposite of the first: not sources that were never used, but too many of them.

Part of every release for us is firing a battery of test questions at the live demo shop and reading the results back through the chat logs. One of this week’s questions was about Clan Cameron — crest, septs, the usual. The answer was good: grounded in the site’s own pages, correct, every link real and working.

But the Sources line listed twelve documents. And on the follow-up question it listed nine — nine clan-map pages, not one of them about Cameron.

Nothing was invented. Every page existed; every link loaded. What the assistant had done was list a large slice of everything the search fetched, rather than the two or three pages the answer actually stood on. And a citation list like that quietly breaks the promise from the other direction: a visitor who wants to check the answer now has to wade through a dozen links to find the one that matters. Cite everything and you may as well cite nothing — the reader can no longer tell which source is load-bearing.

So we did what we always do with a finding like this: logged the evidence, filed the issue, and designed the fix before touching code. The Sources line should list what the answer used, not what the search returned, and it should be capped at a sensible handful. In the interests of honesty — because honest engineering stories are rather the point of this series — that fix is filed and designed but not yet shipped. If you’re watching our demo, you should see the Sources lines get shorter in a coming release.

And the links must exist

There’s a third way a citation can fail, the bluntest of the three: the AI can simply invent a source — a plausible-looking link to a page that has never existed. We told that story in our guardrails post, where our assistant confidently invented two products, complete with convincing web addresses, on our own domain.

That one we treat as a hard stop. Before any answer reaches a visitor, every link in it is checked against the site’s real pages, and an invented link fails the whole answer rather than being quietly served. In this week’s release checks — umpteen scenario conversations fired at the live site — that check had a quiet day: not one invented link, and every single cited page loaded when we clicked it. Quiet days are what good guardrails sound like.

Three questions worth asking

If you’re evaluating an AI assistant for the front of your own website — ours or anyone’s — the citations behaviour is worth probing, and three questions get to the heart of it:

Does the Sources line show what the answer used, or what the search returned? (Asked of AnswerRails today, the honest answer is “the search, for now” — which is exactly why it’s on our fix list.)

Are cited links checked to actually exist before the visitor sees them? An AI that can print an invented link under the word “Sources” has turned the trust signal into a liability.

Can you, the owner, see the difference in your logs? Our chat logs record what the search offered and what the reply cited, separately — so when a Sources line misbehaves, there’s evidence to read rather than a mystery to shrug at.

A citation is a promise you can check. The engineering job — never quite finished — is making sure it’s a promise that’s kept: no sources the answer never used, no more than the answer needs, and never, ever a link to a page that doesn’t exist.

Next in this series: why your AI should tell you when it’s having a bad day — the story of our assistant’s health dashboard, and the afternoon a provider slowdown taught us what an assistant should say about itself.

AnswerRails is a public-facing AI assistant for WordPress, in development at Celtic3d — built compliance-first, with guardrails as the product rather than an afterthought. Follow its progress at answerrails.com.

The post How Do AI Citations Work? (And Two Ways a Sources Line Can Mislead) appeared first on Celtic3d.

At Celtic3d, we make 3D printing simple. Most of our production is handled through our trusted manufacturing partner, giving us access to industrial-scale capacity, a wide choice of technologies and materials, and consistently high quality. Better still, when you need multiple identical copies of a part, you’ll benefit from generous quantity discounts.

There is one trade-off: because of courier transit times, set-up costs per file and a minimum order value, this service is not the best option for once-off small parts. For larger projects, prototypes or batches of parts, however, it’s an excellent fit.

Our core 3D printing technologies

SLS (Selective Laser Sintering) – PA12 Nylon

Our go-to process. SLS fuses fine PA12 powder layer by layer into strong, detailed parts.

• No support structures needed → allows intricate, interlocking or organic shapes.

• A versatile, durable material with a slightly textured finish.

• Great choice for functional parts, prototypes and assemblies.

SLA (Stereolithography) – Resin

The next most popular option. SLA uses a light-cured liquid resin for superb detail.

• Smooth surface finish, excellent for visual models.

• Can mimic some production plastics, ideal for product design prototypes.

• Wide range of resin properties to choose from.

FDM (Fused Deposition Modelling) – Filament

The odd one out. FDM prints by extruding melted plastic (e.g. PLA).

• Best suited for larger, simpler shapes.

• Limited detail compared to SLS or SLA.

• We don’t use this much, but if it’s needed for a project, we will seek a solution through our manufacturing partners.

To contact us: email sales@celtic3d.com or phone +44 (0)7714 790161 (During business hours – GMT+0)

The post 3d Printing Services appeared first on Celtic3d.

Effective from 10th April 2025, all new digital downloads will include a choice between two types of licence:

Personal Licence

This option is intended for private individuals and hobbyists. It allows the file to be 3D printed for personal use or as a gift. However, prints made under a personal licence may not be sold, and the original digital file may not be shared or redistributed.

Commercial Licence

This option is designed for makers and small businesses who wish to produce physical items for resale. It permits the manufacture and sale of printed products based on the downloaded file, but does not allow digital redistribution or modification for resale.

What About Previous Purchases?

If you purchased a downloadable file from Celtic3d before 10th April 2025, your original licence terms still apply. These new options only affect new purchases made on or after that date.

Why the Change?

We’ve introduced this new licensing structure to:

• Continue to offer an affordable option for personal, non-commercial use

• Support small businesses and home-based makers who want to resell printed products

• Protect the originality and integrity of our 3D designs

If you have any questions about which licence is right for you, or if you need clarification about how you may use your files, please contact us.

The post New Licence Options for 3D Downloadable Files appeared first on Celtic3d.

Using high-resolution digital scans and working with brogenville, a local Ferrari expert and enthusiast, we have been producing parts that are no longer in production and that are very difficult to source.

Producing the parts is made possible by the expertise and hands-on experience brogenville has in converting Ferraris with automatic transmission to manual.   This is a continuation of a fruitful collaboration that started with a clutch pedal for another model of Ferrari.

The original OEM parts are cast in aluminium with selected faces machined flat and threads tapped.   The industrial process to create castings requires minimum production quantities to make it worthwhile.  So, for small batches,  we find it more economical to machine the parts from a solid aluminium block.

Our aim is to produce a part that could simply be bolted in.  To achieve that, we include the press-fit bearings and bushings.  Because the bushings and gearings are already fitted for you,  you don’t have to source and fit these yourself.

We had a number of pre-orders for these parts and, after running a small batch through production, we have some stock that is available for sale and shipping worldwide.

How to buy

If you are interested in purchasing these parts,  contact us here,  or on this thread on the Ferrari Chat forum.

If you are converting your late model Ferrari from automatic to manual, and are looking for parts, get in touch, we may be able to help.

The post Parts for sale- Reproduction right-hand drive clutch pedals for Ferrari 360’s and 430’s appeared first on Celtic3d.

Every now and then, an unusual challenge drops into our inbox.  So it was when we were asked if we could help make a clutch pedal for a Ferrari.    This was needed for a project to convert a rare right-hand-drive Ferrari from an automatic gearbox to a manual stick-shift.

To be clear,  car mechanics are not our strong suit.  So this project would require a lot of input from our client.   Luckily, in this case, our client seemed to know what he was asking for.   His long-term project was to convert a Ferrari “612 Scaglietti” to a manual transmission.  He would be the first person, worldwide to complete this.  Apparently, only 27 right-hand-drive versions of this car were ever made, and whatever stocks of spare right-hand-drive clutch pedal arms were originally manufactured,  none are currently available on the market.

Safety First

Our first discussion was on whether the clutch pedal was a safety-critical part.  If it was,  then a whole lot of testing and assurance would be needed that would have likely made the job uneconomic.  While a clutch pedal failing during operation would be annoying,  it would not affect the safe control of the vehicle or the ability to bring it to a stop.   So, we were happy to proceed.

General approach

In very general terms,  the job split into two phases.  First to recreate the CAD (digital) version of the design,  then Second, to select the most appropriate way of manufacturing the part and get it made.

As with any job that initially looks fiendishly complex,  it’s a case of decomposing it down into small tasks that can be accomplished,  and then make a start and work your way through your list of tasks.  Our assessment of the job was that, although it looked complicated,  we had enough contextual information and known measurements to figure out the design.   It was an interesting challenge in reverse engineering. And, who doesn’t want to work on a Ferrari?

Recreating the design in CAD

We did not have access to the Ferrari CAD files for the clutch pedal.  The reference available to us was some 2d drawings from a maintenance manual,  photos from a car with manual transmission, and the bracket that all the foot pedals fit to.

Using these references, we would need to create our own CAD design.  Our design had to meet some key design criteria:

• The location of the pins and pivot points where the part joins to the bracket needed to match the actual bracket.
• The location in space, angle and range of movement of the socket that accepts the connecting rod from the clutch mechanism needed to be correct – as seen in the 2d drawings in the manual.
• The max and min range of travel for the pedal arm needed to match the drawing in the maintenance manual.
• The location in space of the clutch pedal relative to the other pedals needed to match the photos of existing manual transmission examples.

Although we have never actually seen a clutch pedal arm for a RHD Ferrari, and the only drawing we had consisted of a single side-view, as long as we met these criteria, the bearing points and interfaces were at the correct angles, we were confident that the part would work.

We created a new CAD version in Autocad Fusion 360.    A big advantage of using Fusion 360, is that we were able to give our client online access to the project so that we could both see progress and comment on the design as it developed.  This gave our client some confidence that the part was evolving in the right direction and that progress was being made,  but also allowed us to tap into the knowledge and experience of our client to resolve tricky aspects of the design.

Manufacturing the part

To reduce the risk of wasting money on parts that were not quite right, we ran two iterations where a plastic versions of the pedal arm was 3d printed as a test-fit.  Final tweaks and adjustments from the test-fit fed back into the CAD design.

Having confirmed the correct size, shape, connection points and angles, we were ready for final manufacture.  Three main options were explored to create the final part.  Each with its own pros and cons.

Direct 3d printing in metal.   Pros:  Quick.  Underhangs and voids easily done.  Cons: The size of the part meant that it would have to be 3d printed in a few pieces which would bolt together to form the full-sized shape.  Some machining of bearing surfaces would be needed to finish.  Cost: £££

Investment Casting.   A sand mould would be made around a 3d printed plastic master,  then cast in aluminium.  This is how Ferrari made the original parts.  However,  when trying to find a company to do this, we were not able to find a foundry that could handle the size of the part.  We learned that the larger the wax master, then the hotter the furnace needed to successfully melt it out.  Pros:  Would result in a single piece,  Cons:  Some machining of bearing surfaces would be needed to finish.  Some voids would need to be filled in and cut out later.  Unable to find a foundry that could handle the part.  Cost:  N/A

CNC Machining from metal stock. Pros: Best surface finish.  Bearing surfaces would be finished in a single process.  Cons:  Quite a complex shape – would require advanced tooling. Cost: ££

The final process was to send the CAD files we developed to a CNC machining shop via 3D Hubs.  For this project, our client chose to send the files to 3D Hubs himself.  For other projects, we can take care of this as an end-to-end service.  The key part of the process where Celtic3d can add value is in creating the digital files that can then be sent to manufacturing.

The Results

These images show the CNC machined part received from 3D hubs.    It was a complicated shape to machine.    To be honest, we were prepared for a request to modify the design, perhaps splitting it into a few different parts to make it easier to manufacture.    But, the machine shop found via 3D Hubs, had a 5-axis CNC machine and no problems with it.

“It was a great experience working with Celtic 3d. This is obviously an unusual project, so when I emailed about the project I was pleasantly surprised then that Celtic 3d was willing to have a go. Seeing the design come to life via the web interface was really helpful, and allowed me to make comments as the design evolved. Gavin really couldn’t have been more helpful and accommodating.”

RH, Ferrari Owner

Making your parts

It is always rewarding to see a digital creation you have worked on, produced in physical form.  We are particularly pleased with the way this job turned out.   As an example,  this job perfectly illustrates what is possible if you combine the ability to create 3d content with flexible on-demand manufacturing capability.

If you would like to learn more about our lightweight CAD capabilities, or if you are trying to figure out how best to get something made,  please contact us to discuss your needs.

The post Hard to find Ferrari parts appeared first on Celtic3d.