Frank Lindseth retweeted
Head of Claude Code:
"I'm not prompting my agents anymore, I'm building graphs and loops so they can build the agents for me."
In 10 minutes he shows exactly where agentic engineering is going and how not to get left behind.
Definitely something you cannot afford to skip.
Watch it, then read the step-by-step guide below on how to build a system that improves itself.
Frank Lindseth retweeted
A scan gives you a picture. A simulation needs objects.
We can output Revit files, USD, IFC. Anything needed for BIM/Simulation workflows.
This is a real process plant in Rotterdam. On the left is the raw LiDAR, about 5 million NavVis points. On the right is our reconstruction, filmed from the same camera.
Most scan-to-3D pipelines stop at one fused mesh, which looks right but behaves like a single rock. A robot can't open, avoid or reason about something it can't tell apart from the floor.
So after we rebuild the geometry, we separate the scene into objects. In this plant that's 89 of them: pumps, vessels, heat exchangers, stairs, guardrails, pipe networks, foundations. Each one:
→ is its own mesh with its own pivot point, so you can select it, move it or swap it
→ is labelled by type, like "pump assembly" or "vertical heat exchanger"
→ traces back to the scan panoramas and points it came from
→ is flagged where we had to infer hidden geometry rather than measure it
Frank Lindseth retweeted
Cauliflower detection with size estimation using @ultralytics YOLO26 ☘️🪴
This example showcases the ability to analyze the size of individual cauliflower plants cultivated in the field. Such insights can help farmers optimize planting strategies and monitor crop health.
#agriculture #MachineLearning #Research
Frank Lindseth retweeted
In July we open-sourced a dynamic graph of everything a child learns in primary school. 1,590 concepts, 3,221 dependencies.
Now we’re turning every concept into interactive experiments kids can play and learn from.
First one, age 7: heat and water.
Everything that used to be a boring textbook can now be an engaging experience, potentially tailored to every child.
Kids education just got a massive upgrade.
Frank Lindseth retweeted
Upgrading an intersection used to mean trenching the pavement for loops, closing lanes, and calibrating a stack of separate sensors. Ouster BlueCity mounts to the existing signal pole in as little as 3 hours, with no trenching, no new mast arms, and no complex field calibration.
Once it's mounted, the traffic engineer draws virtual detection zones directly on a live 3D view of the intersection. BlueCity connects to the existing signal controller, so actuation, safety alerts, and analytics run from the same data on day one.
Learn more about Ouster BlueCity: ouster.com/products/software…
Frank Lindseth retweeted
InfiniSplat v2: our follow-up to InfiniSplat is out.
🤗 Multi-view images in, clean high-fidelity Gaussians out — unbounded free-viewpoint rendering.
👉 Try it now: huggingface.co/spaces/Edison…
💻 Code: github.com/zju3dv/InfiniSpla…
Frank Lindseth retweeted
RustGIS v0.3.0 is here!
RustGIS is a fast, free and open-source GIS in pure Rust for mapping, analyzing and sharing geospatial data. It runs natively on Windows, macOS and Linux, and in your web browser. Your data stays on your machine unless you send it to an AI model you choose, and every action is a command you can script.
This release puts the world's open data a search away, draws maps on the GPU, and brings RustGIS to Python and Jupyter. It is our biggest release yet: 150 pull requests from 16 contributors, 15 of them contributing for the first time. Thank you to everyone who helped!
What's new in v0.3.0
- Open data, built in: search STAC catalogs, Planetary Computer, Overture Maps, Source Cooperative, OpenStreetMap, geoBoundaries, data.gov and other open-data portals, and browse S3 buckets.
- Cloud-native by address: GeoParquet, FlatGeobuf and cloud-optimised GeoTIFFs open by their web address and are read in place, only the parts you need.
- More formats and services: live PostGIS tables, DXF, LiDAR (LAS, LAZ, COPC), MBTiles, Zarr, WCS, OGC API Tiles and many more, plus an SQL pane powered by DuckDB Spatial.
- A faster map: GPU rendering (beta), smooth zoom and pan, and a globe with your map draped over it under the stars.
- Python and Jupyter: pip install rustgis to run the engine in Python and open the app in a notebook with rustgis.Map.
- Process plugins: plugins that run as programs, with their own Python environments, in the background with progress and Cancel.
- AI Assistant: sign in with ChatGPT to use your plan, and let external AI clients drive the app.
Try it out
- Launch the web app: web.rustgis.app
- Download: downloads.rustgis.app
- Documentation: rustgis.app
- GitHub: github.com/opengeos/RustGIS
- Release notes: github.com/opengeos/RustGIS/…
#GIS #Rust #OpenSource #Geospatial #OpenData #Python #CloudNative
Frank Lindseth retweeted
New tutorial | Click to track & crop objects with Ultralytics YOLO26 🎯
Learn how to build interactive object tracking with YOLO26 + @OpenCV. Click an object, follow its tracking ID, and display a live cropped view.
Watch here ➡️ youtu.be/hPXsZ_fVz6M
#Ultralytics #OpenCV #ObjectTracking
Frank Lindseth retweeted
@Google's Gemini 4 Argon is now available in Ultralytics Platform! 🚀
In our latest internal benchmark of 26 AI models across 36 real-world object detection datasets, Gemini 4 Argon ranked #1 in auto-annotation accuracy with 0.565 mAP50, outperforming Claude Opus 5.5 and GPT-6 Astra.
From medical imaging to retail, traffic & industrial inspection, developers can use Gemini 4 Argon to generate bounding boxes, review predictions & annotate datasets at scale, reducing manual labeling effort.
Read more ➡️ ultralytics.com/blog/gemini-…
#Gemini4 #Ultralytics #AI
Frank Lindseth retweeted
Great news for creators using Gaussian Splats.
Not only does Blender 5.3 add native GS support, but depth of field also works 🤩
We tried using an orthographic camera and super shallow depth on a KIRI Engine scan, the result = instant micro worlds
Frank Lindseth retweeted
🤯 a year ago I would say photogrammetry will always be better than 3DGS to mesh. But now I might have to talk that back.
I still think lidar/photogrammetry fusion is the ultimate winner though.
Thoughts?
Frank Lindseth retweeted
RootSLAM: Square Root Bundle Adjustment for Visual-Inertial SLAM
📄 arxiv.org/abs/2610.15157
⚡ State-of-the-art SLAM accuracy, 3–14× faster!
🤝 A collaboration with Tomás Santucci and Prof. Taihú Pire (@taihu_pire) at the UNR🇦🇷, and Prof. Daniel Cremers at the TUM🇩🇪.
Frank Lindseth retweeted
Gemini 4 Argon taught a robot to play cornhole just by watching a video.
It mapped the physics, coded its own simulation, and taught @Apptronik's Apollo humanoid to sink 7 throws in a row.
Frank Lindseth retweeted
AI agents are moving from engineering workflows to the fab floor.
Announced at #SEMICONWest, NVIDIA cuPotential joins the CUDA-X portfolio, helping developers build AI agents that simulate atomic and molecular interactions so engineers can model semiconductor materials and manufacturing processes faster.
Frank Lindseth retweeted
RustGIS is a fast, free and open-source GIS in pure Rust for mapping, analyzing and sharing geospatial data. Its a really good fit for running in agent bots, its fast and well organized with builtin mcp/cli. rustgis.app/
Here's a @grok @bot template to install it in GrokBot.
example:
"I asked my Grok Bot to pull today's live wildfire perimeters from a public ArcGIS service and check which counties are exposed. In seconds RustGIS buffered them by 10 km, overlaid counties, and made the map + report: 40 Pacific NW counties, 7.0M residents."
Frank Lindseth retweeted
This is our biggest ever update just 1 week in.
We are now on par with companies that have 100+ employees and raised tens of millions.
I'm 1 dude with a @claudeai subscription and a dream.
Go test it out, if this is what I can build in a week, image where it is by the end of the year.
Shoutout @levelsio hoping for a retweet 😀
Live now, you can add furniture, fittings, flooring, change cabinet layouts and so much more.
If you want to have a free play around:
homistry.ai/demo/try
Prices double in 2 hours.
Frank Lindseth retweeted
We just open-sourced an entire 3D team for Blender.
Introducing the Scenario plugin.
Use it inside Blender, or let Claude Code run it through MCP.
📁/scenario-plugin
┃
┣ 📁 Concept Artist
┃ ┣ 📁 Concept Art From a Prompt
┃ ┗ 📁 References From Your Viewport
┃
┣ 📁 3D Artist
┃ ┣ 📁 Meshes From Text or a Photo
┃ ┗ 📁 Retexture or Remesh the Selected Mesh
┃
┣ 📁 Animator
┃ ┣ 📁 Rig the Selected Character
┃ ┗ 📁 Motion From Text
┃
┣ 📁 Surfacing Artist
┃ ┣ 📁 PBR Materials Onto Any Slot
┃ ┗ 📁 Maps From a Photo
┃
┣ 📁 Layout Artist
┃ ┣ 📁 Greybox From a Description
┃ ┗ 📁 360 Worlds and Skies
┃
┣ 📁 Camera Operator
┃ ┣ 📁 Camera Path From a Sentence
┃ ┗ 📁 Orbits, Dollies, Cranes and Flyovers
┃
┣ 📁 Render Artist
┃ ┣ 📁 Viewport to Finished Still
┃ ┗ 📁 Playblast to Finished Shot
┃
┗ 📁 Sound Designer
┣ 📁 Music, Voice and SFX
┗ 📁 Dropped on Your Sequencer
Comment "BLENDER" and I'll send you the GitHub repo.
Frank Lindseth retweeted
NVIDIA × NV-Reason-CT：胸部CTから放射線レポートを生成するオープンAI
🔹約7万件のCTスキャンで学習。3D胸部CT全体を解析
🔹NVIDIA A100 GPU 1台で、30秒未満にレポートを生成した実証例
🔹Hugging Faceでモデル公開。院内GPUで動かせば、CT画像を外部クラウドへ送らずに処理可能。
情報元はリプ欄
Frank Lindseth retweeted
Anthropic CEO Answers the Question Everyone Cares About Most: Should You Still Learn Programming Now?
1. Programming will be the first to be eliminated.
AI models are already doing this. Broader software engineering tasks will take a bit longer, but they're also receding. If you're learning programming just to keep your job, you're heading in the wrong direction.
2. Even if you only do 5% of the work, it still has value.
AI does 95%, you do 5%, and your productivity can increase 20-fold. The power of comparative advantage is astonishing, even with huge gaps.
3. The most promising careers are human-centered.
Jobs that combine interpersonal skills, the physical world, and analytical abilities. He cited radiologists as an example: doctors who understand patients and context, not just those who can read scans.
4. Critical thinking may be the most important skill in the next decade.
When AI can generate anything, the ability to discern truth from falsehood becomes scarce and precious. You don't want to hold false beliefs or be deceived. This is his practical advice to 25-year-old young people.
5. If you don't use AI well, it will make you dumber.
Anthropic has done related research. Depending on how you use the models, the decline in coding ability is measurable and real. The problem isn't the tool—it's misuse.
6. He's bullish on semiconductors as capitalism's victory in the next decade.
The physical world, traditional engineering, direct AI dividends. Not software, but chips.