demeyer1

I’ve been thinking about going that route.

I’ve seen some good research where xx% of the time, all LLMs essentially just get lazy and won’t actually look things up.

Been working on an informational retrieval system similar, but endless Nerfs combined with non deterministic and finicky behaviors make it hard to reliably depend on.

How’s your luck been with LLm-wiki?

It’s called Autobot, it’s designed to get as close as I could reasonably get to a real world Iron Man Jarvis.

WHAT IT DOES

OSS/MIT Harness that helps agentic tasks run for up to 4 days using without performance degradation. Designed to work really well in native, conversational voice mode. Does a good job with knowledge work, managing computer use “leases” in a way that avoids conflicts across sub agents.. or deploying an entire AWS infrastructure pattern from zero and launching 100 different VM images.

HOW IT DOES IT

Uses a canonical event ledger, heartbeat (for persistence by low cost orchestration), extends foundation memory to encrypted disk storage, and manages its own scheduling system (easier to switch to another platform).

Installs as a project, so very easily. Zero config. No special app, access needed.

Accompanying repos have the skills to disable approvals inherent to the foundation models (use with caution, not advised - separate repo under parent).

BENCHMARKS

Currently top of AssistantBench Leaderboard; and, benchmarked at the top of OS World 2.0, the hardest knowledge work computer use benchmark I could find, using a model that is one generation behind. All benchmarks in repos with cryptographic seals.

Entirely free, nothing to sell - it’s been a game changer for me, so I’m just putting it out there.

Hoping to find others also working on pushing this particular dimension of harnesses forward.

CHECK IT OUT

https://github.com/demeyer1/Autobot/graphs/traffic

Yeah, that is a perfect use case for commuting, going for a walk in between meetings, etc. Same concept here.

But then the next annoying thing is the endless approval and safety nags. So there is a somewhat hacky skill I posted on my GH that is designed to trick the models into approving each other (use cautiously, it can go sideways fast in full computer use mode).

Good share!

I believe Zack started Gridfinity, but I don’t know who the very first designer was that built a modular storage system for the 3D printing community.

I’ll say this - the Gridfinity community is, from my POV, unique in how they’ve taken this and really run with it as a crowd sourced concept.

Agreed. The shear volume of permutations out there now may be adequate for many beginners that aren't yet comfortable with 3D design clients.

Super excited to see Gridfinity on HN, it introduced so many new people to 3D printing as well as introductory 3D design.

It’s simple, modular storage that is practical and really grounds 3D printing in a genuinely practical use case.

To provide anecdata on how big gridfinity has become - it’s the #1 search, every week on Thangs.com, which is the largest 3D model search index available (over 14M models, and search volume is ~ 1 search every 1-1.5 seconds on Thangs). (1)

Gridfinity is literally more popular than Christmas in this community. And it’s been the top search term for a long time. (2)

There are entire subreddits dedicated to it. (3)

Zack, the guy who started it, is a genuine treasure of a human as well. Check out his YouTube channel (4), he recently built a Google Glass like device - and then shared all the designs, openly.

Disclaimer: CTO, CPO, co-founder at Thangs.com - total 3D design and 3D printing zealot

(1) https://twitter.com/Thangs3DSearch/status/160156271186268160...

(2) https://docs.google.com/presentation/d/1hv4FBexbf6isW57j063H...

(3) https://www.reddit.com/r/gridfinity/

(4) https://www.youtube.com/c/ZackFreedman

Couldn't agree more with your points, and I agree that this clash is going to happen.. and that will be a great thing if it does!

If you are interested in connecting - I would be happy to trade observations over a video meeting and coffee/beer. [email protected]

Disclaimer: I don't know if we are allowed to post this on HN, off the official posts, but I'm also hiring. We do 3D search, are growing rapidly, and have some excellent investors: Sequoia, Google Ventures, Tiger, etc.

Well said. What I suspect we are seeing is the beginning of a trend where the rate of change begins to accelerate.

If nothing else, the growth in new 3D creators (be they mechanical, design/sculpt, AR/VR/MR/XR, printing/additive) will bring in new individuals less beholden to the way things have been done in the past. This is happening today.

If we look at those with some of the highest 'change cost', we can look to CAD and PLM within manufacturing. Most of the tooling in that category (design clients through to PLM) hasn't even come close to keeping pace with modern conventions. I'm as excited as you, even within that relatively small, change averse context.

We have a few million unique users, which is small within the broader context of 3D creators and consumers - so small N, but I'll share a couple observations at that volume.

Adoption

STLs remain the most popular, today, on our platform. 3MF has been growing, but only in the past quarter did our users tell us in meaningful volumes that we need to support it - which we did, happily. The push to 3MF seems to be gaining momentum.

Expected benefits vs current state (again, small N and biases in my data)

1. Color/texture is a benefit, though creators include it to varying degrees based on their use case and the intended consumption. We support printers, general 3D designers/sculptors, and mechanical engineers - so our audience is broader than primarily/solely 3D printing. As such, adoption varies considerably depending on the user and use case.

2. Often, the model has more metadata (though rarely includes color, photorealistic renders, etc) and a smaller storage footprint. Benefits creators, consumer and platforms. Though again, adoption lags on the more advanced capabilities inherent to the format.

3. Support for assembly like models is a huge benefit! However, the average 3D printing oriented creator still prefers collections of STLs. In user interviews, I often hear the rationale for this is that there are (perceived) tooling and general workflow changes that seem time consuming - so the usual.. change/learning takes time.

Timing

Creators will drive adoption, as if often the case. We are seeing this increase every month on our platform (thangs.com), but the relative share of uploads still skews towards STL and other formats.

Platforms will need to adapt and take advantage of the format benefits to help consumers see the value.

Over the mid term, I am really excited about 3MF. The fact that there is a spirited discussion on HN about it, if nothing else, strikes me as a great sign!

Well said. In some respects (not all), on our teams that manage geometric indexing, we've been comparing this to Office document formats a long time back.

Your take is similar to mine. Three core components to the metaverse: 3D creators/content, lower cost hardware, and social/distribution networks.

The 3D creator population is expanding. The tools are better and the use cases are expanding. So yes, to your point, existing assets.

Then there is the hardware, as you've pointed out. This makes all that content slightly more user friendly and available at a lower barrier of entry to the average user.

Then there must be a network that brings users together. Certainly seems like Facebook had capability for this area.

Will it be controlled by Meta? Too soon to say IMO, but they have capacity and expertise for two of the three requirements in my (overly simple) mental model. I suspect the Web3/DAO crowd has a pretty strong opinion here. My two cents is we'll have at least two primary metaverse-like networks of users (one more closed, one more open).. but likely more.

Disclaimer: I work at Thangs.com, which is a 3D search and collaboration platform. So I have obvious biases which lead me to assume there will be a more open metaverse in the future.

Appreciate the kind words - if you ever have any feedback, please just send me an email ([email protected]) and I'll be happy to either jump on a call or take the feedback async by email. We are genuinely trying to build a different type of platform, and we highly prioritize listening to our users. We can't build everything everyone wants immediately, but it's a rare product change we make that isn't aligned to user feedback.

I love this comment! Thank you for saying that!

Indeed, as the other individual commented - there are compliance obligations which are more difficult/costly/time_consuming (not impossible) to satisfy if we have non US citizens working on the product. We are going to hire outside the US, but at our current stage there is a cost we can't easily absorb - right now.

We do. It works for us because we are aggressive on salary (for a startup at our stage) and very aggressive with benefits and flexibility.

In an aggressive hiring market for engineers, this is a decision Google is making which will increase potential churn risk for some engineers. As a Xoogler myself, I know how much they invest into thinking about talent acquisition and retention and would venture to say they are aware of this. I suspect none of this is as "hot take" to them.

At the same time, Google's stock is up massively over the last two years. Google has been relatively aggressive with RSU allocations. That appreciation in RSU value will act as an offset for a subset of those who would churn. This must also factor into their current policy decisions. The team's at Google try very hard to optimize for retaining their best people, but they are dealing with very large numbers of individuals during an unusual time. Further, they have traditionally been leading the charge for geographic proximity and density for product teams. Lots of change for them to manage through right now.

That said, as a CTO at a startup (which counts Google Ventures as an investor) I have hired many engineers from Google over the past 18mo. Approximately half our engineers are ex Google, Amazon, Microsoft. Beyond our product and ENG culture, our policy is fully distributed work (within the US - compliance reasons) and that is the one benefit ex FAANG engineers seem to value the most, currently (small N, selection biases acknowledged). Over all roles, I would estimate more than half the company consists of ex FAANG employees. Many were tired of being locked into a geography which did not make them happy.

So I view this as a change in the market dynamics which is allowing startups like mine to help make these engineers happier. We have engineers that have joined, and immediately moved to Hawaii, or purchased a rural farm in the Midwest - and we are 100% supportive of them having a work/life that makes them happiest. Also, we've removed policies of paying differently based on the traditional tier 1, 2, and 3 geographies.

This is a market advantage that favors engineers, and as such - I'm very supportive :)

I don't think you are off entirely here. The 3D creator community is growing very quickly with new tools, lowering the barrier of entry to 3D design, emerging consistently. That provides the content. Hardware, to your point, is getting better, cheaper and more accessible to more content consumers. The last component, as I have been thinking about it, is then the network that brings users together. As other posters have already mentioned - the user experience is an improvement vs the original Second Life.. which may be enough, over time, to become a new inflection point.

I do think there is more hype than substance right now around the metaverse, but I don't have to squint too hard to see a world where this becomes real relatively soon.

It‘a most often used as an auto-complete - of sorts.

For example, upload an engine piston and it will find other pistons, then all the engines that the piston will also fit within. Handy for designers and mechanical engineers.

But, in the 3D printer use case it is most often used to find similar objects (it does both). For example, upload and a Squid Games mask to print for Halloween and find other Squid Games masks. It’s entirely geometric in that use case, and math doesn’t care if you have different languages or people add the right tags and categories.

Hopefully that makes more sense.

In general, that feature is used less than 1% of the time vs text search. We have a powerful text search engine which we’ve built and customized on top of a lucene implementation. We are generally improving CTR by a percent or two every week.

They are shockingly useful for real world, functional needs.

If your interested in digging in more within some generally positive and positive communities, I'd recommend r/3dprinting and r/functionalprint.

Good list! Couple thoughts - if you are using some of these items with kids (the replacement part example) it's worth considering the types and materials of the filament you are using. Some are safer than others, depending on the application. If you are using for food storage, there are ways to take otherwise unsafe materials and make them safer for storing edible/consumable materials.

If there is something out there you need, I'd recommend searching Thingiverse, Prusa, and Thangs.com as they all have large libraries of free 3D printable objects.

Disclaimer: I'm the CTO at Thangs.com and also a bit of a 3D printing nerd ;-)

Agree with the fundamental premise of the article.

We started Thangs.com to allow for actual 3D search, eg. search by uploading a 3D cylinder and find other cylinders as well as engines it fits within.

The growth has been mostly quite strong since we launched a year ago, but it’s a very capital intensive product to build - massive volumes of data.

Our monetization strategy is to skip the ads and instead use our 3D indexing to allow us to build a 3D, visual, revision control system - also free. However, we plan to charge enterprises for the standard advanced SaaS capabilities - which we then integrate with our current enterprise product.

We have a single tenant 3D search product that is doing well in enterprise, without the revision control capabilities yet integrated. Beyond search and reuse, there are massive opportunities in supply chain and many other valuable problem categories where true geometric search is a uniquely valuable solution.

It’s not always the case that ads and web3 principles are required to build commercially viable, yet somewhat narrow search products. The world of search is large and there is ample room for commercially viable specialization.

I've had the opportunity to know Kelsey for a number of years, and we've worked together closely on occasion. He is the genuinely good human the article portrays him as, and unlike some who evangelize - Kelsey understands his technical area (k8s) deeply and is on the CLI daily. Top notch human, and I thought I'd add a POV from a regular HN reader.