Build 3D spatial AI that actually ships.
Clarity on what actually matters, and the capacity to build faster than you ever have, guided by a mentor who has shipped 3D spatial AI for fifteen years, with his Brain-to-Deploy (B2D) method.
For professionals and teams inside companies, researchers, and innovators who want the capability in-house, not rented.
The shift to physical AI, in four minutes.
A direct look at what you build inside the program, and why I built it the way I did.
The problem
Spatial AI is moving fast. The path through it is the hard part.
The opportunity is real, and it is growing. The hard part is seeing clearly what spatial AI actually is, where it is heading, and how to use it end to end. Without that map, an idea never gets scoped into something real.
Making it work means owning the whole chain: the right acquisition (LiDAR, photogrammetry, the sensor that fits), then the data, the segmentation and identification, the domain and the algorithms, and finally orchestrating all of it, by putting your hands in the mud where it matters, and leaning on AI where it lifts your ROI.
This is where it stalls. A surveyor needs feature extraction that runs at scale, real maps and plans with every feature on them, not a demo on half a dataset. A founder needs a result they can sell, not an idea on a canvas. A researcher needs a result solid enough to publish and build a program on. All of them need the same thing: clarity, and a way to actually build it.
The gap was never your ambition. It is a clear, end-to-end path, and the capacity to build it.
The mentored path
So we close that gap, together.
From the first session you get clarity: on what spatial AI really is, where it is heading, and the end-to-end path from sensor to shipped result. Then the capacity to build it, at a pace you have never had. And you are not doing it alone. I walk it with you, using my Brain-to-Deploy (B2D) methodBrain-to-Deploy (B2D) is the exact loop I run on client work for the world's largest companies, including BMW and Airbus. Scope the real problem, design the pipeline from sensor to result, build it (by hand where it matters, with AI where it speeds you up), then verify on real data until it ships. Other training stops at theory. B2D is what gets you to a system that runs, and it is yours..
Your idea, sharpened. We start from what you want to build, shaped by the opportunities and openings we map together, so you aim at something that actually matters.
Grounded in 3D data. Point clouds, LiDAR, photogrammetry, 3D Gaussian splatting. This is not generic AI, it is built for the messy reality of three-dimensional data.
You write the code. Python first, and the right language when it matters. You understand every line, instead of pasting what you cannot trust.
AI as leverage. Agents that read, plan and act on your 3D data, a planner, your tools, a verifier, so you move faster without losing control.
A complete tool you own. A real application with a GUI and advanced functions, packaged, deployable, and yours to keep, sell, or build on.
By the end, this is the level you reach: a real project, even a startup, built in a single weekend.
"Think of a coach. You could learn tennis from videos, or now by asking an AI. Or you train with someone who played at the top for fifteen years, who tells you in one sentence what would have cost you two years on your own. That is what I do here. AI can generate code all day. The fun part, the part that is yours, is building the thing. I do it with you and for you, so your time goes into building, not into figuring out how."
Everything inside the program.
Every member gets the CORE. Professional adds the four Operating Systems I use on real client work. Architect adds me.
3D AI Accelerator i17 episodes across 6 Acts, each ending in software you keep:- 1. 3D Python Systems, a viewer .exe
- 2. World Reconstruction, a reconstruction engine
- 3. Smart Point Clouds, a labelling system
- 4. 3D Deep Learning, a classifier
- 5. Spatial Agents, an agent app
- 6. Digital Twin, a twin engine
Under 20 hours to complete, and you are ready to tackle complex projects end to end.
From raw scan to deployed app, the spine of the whole program.
Course Library iEvery standalone specialty course, with 12 months of continuous updates. Open the Library ↗
Gaussian splatting, PointNet, change detection, feature extraction, and more.
Neurones 3D software iThe desktop app my consulting runs on. Forkable, no watermark, no rent.
Load, label, segment, reconstruct, agent-query. Yours to keep.
Monthly briefing iA short video plus written report on what moved in 3D AI, and what to ignore. See the reports ↗
Stay current without drowning in papers and noise.
Job & market board iCurated 3D AI roles and tenders from around the world. See open roles ↗
Curated roles and tenders that actually need 3D spatial AI skills.
Community
Share work, get feedback, build alongside other architects. Async-first.
Not slideware. Four full production stacks I use with consulting clients, each a complete specialty in its own right.
Reconstructor OS€1,497 value
Turn images and scans into clean, metric 3D.
+ What's inside
- Photogrammetry automation (COLMAP, UAV)
- 3D Gaussian splatting, trained on your own images
- NeRF and multi-view stereo pipelines
- Metric scaling and control-point discipline
- Meshing and a one-click reconstruction engine
Segmentor OS€1,497 value
Give every point a name, at scale.
+ What's inside
- Classical segmentation (RANSAC, region growing, DBSCAN)
- Supervised and unsupervised pipelines
- Foundation-model-assisted labelling (SAM, CLIP for 3D)
- Human-in-the-loop correction that stays sharp
- Change detection and instance extraction
Deep Learning OS€1,497 value
Train your own models, from scratch.
+ What's inside
- PointNet and PointNet++ implemented line by line
- KPConv, sparse-conv networks, GrowSP, RandLA-Net
- Custom training pipelines and calibrated confidence
- Model compression: pruning, quantization, distillation
- Deployment: ONNX, TensorRT, edge inference
Spatial OS€1,497 value
Make your 3D data answer questions.
+ What's inside
- Scene graphs and spatial reasoning
- Agents: planner, tool-use, verifier loops
- LLM and VLM integration for 3D scenes
- Browser-deployable spatial AI web apps
- Digital twin engine, queryable and defendable
See the complete program breakdown+
Foundation
- 3D AI Accelerator, 17 episodes
- Full Course Library
- Neurones 3D software
- Monthly briefing
- Job & market board
- Community access
- 12 months of updates
- 90-day guarantee
Professional adds
- 4 Operating Systems
- Reconstructor OS
- Segmentor OS
- Deep Learning OS
- Spatial OS
- Priority support
- Deployment playbooks
- Monthly Intel Drop
Architect adds
- Onboarding + strategy calls
- Private channel with me
- Co-built custom solution
- Portfolio endorsement
- Priority in sprints
They chose the architect side.
A slice of what working pros build, across 80 countries.
One of my students walks through an idea he had and the system he built for it inside the program. Not a testimonial, the real thing, shipped.
"I built a spatial agent that walks our scan-to-BIM pipeline end to end. The tool-calling chapter clicked in a way nothing else ever did."
"Scan-to-BIM used to be a three-week manual job. It is a two-day automated workflow now. Three new heritage contracts the same quarter."
"A surveyor with 20 years in the field. This gave me the Python and AI to modernise our entire workflow. Best investment of my career."
"We replaced a brittle classical pipeline with a sparse-conv network. Accuracy jumped from 78% to 94% on the same data."
"Florent treats agents as engineering, not magic. The evaluation harness alone changed how my team measures whether an agent is useful."
"Two factories, one onboarding script, identical deliverables. It turned our scan-to-FM process from artisan craft into a repeatable product."
Architect Spatial Intelligence.
The Brain-to-Deploy methodology. From first principles to production-grade 3D AI.
Master the core 3D AI stack. For innovators building a strategic edge in spatial intelligence.
- Spatial Accelerator (17 Episodes) The 17-episode deep-dive on the Brain-to-Deploy methodology. From mental models to shipped production systems.
- Full 3D Course Library (20+ Courses) The complete curriculum. Point clouds, meshing, segmentation, deep learning 3D, spatial reasoning.
- Neurones 3D Software Suite The software suite I built to unify 3D reconstruction, segmentation, and spatial analysis in one stack. Standard commercial license included.
- Monthly Spatial AI Nuggets A monthly briefing on what moved in 3D AI. Research, code, and market signals I think you should know.
- Private Job & Market Intelligence
Scale from prototype to industrial-grade deployment. For founders architecting proprietary spatial systems.
- Everything in Foundation
- 4 OS Deep-Dive Production Tracks Complete tracks: Spatial Reconstructor, Segmentor, Deep Learning 3D, and more. Each ships with 12 months of active updates and support, with optional annual renewal. Each track valued at €1,497.
- 5 Forge .exe Apps + AI Agent Toolkit Five ready-to-run Windows apps plus the AI agent toolkit. Run 3D AI pipelines without touching infrastructure code.
- 12-Month Strategic Production Briefs Every month I spot a real market opportunity and hand you the full step-by-step blueprint to build it. Twelve briefs, twelve shipped tools or software over the year.
- Monthly Live Q&A Sessions Monthly live calls. Ask anything technical or strategic and get a direct answer from me.
Elite 1-on-1 advisory. I personally review your architecture and deliver custom Brain-to-Deploy blueprints.
- Everything in Professional
- Onboarding + Annual Strategy Sessions A kickoff session to map your system, plus recurring strategy calls where I pressure-test your architecture and roadmap.
- Private 48h Priority Channel A direct line to me for architecture reviews and technical unblocks. Guaranteed response within 48h on business days.
- Co-Built Custom 3D AI Solution Not built for you, but with you. I code alongside you to architect and ship a custom 3D AI solution for your specific use case.
- Portfolio & Project Endorsement
Not an idea on a canvas. A system that works.
You walk out with results you can use, automate, sell, or publish. The ROI is clear from month one.
Automate at scale
Auto-extract maps and plans with every feature, from LiDAR or photogrammetry. Production scale, not a half-dataset demo.
Build something you can sell
A working system, not a slide. Sell it, raise on it, or publish it.
Own it, stop renting
Build exactly what you need and own it. Stop paying rent for software that never quite fits.
Local-first, zero cloud bills
Runs on your machine and local AI. No cloud bills, no lock-in, faster iterations.
And because what you build is this advanced, you stop feeling stuck, and you become the person teams, labs and companies want involved in their project.
Find Your Perfect Fit
From free exploration to enterprise transformation.
| Feature | Free Experience | Cherry Pick (Library) | Foundation (€1,997) | Professional (€2,997) ★ | Architect (€4,997) | Enterprise (Custom) |
|---|---|---|---|---|---|---|
| Spatial Accelerator | Act I only (Ep 0–3) | ✕ | All 19 episodes | All 19 episodes | All 19 episodes | All 19 + custom content |
| Course Library + OS Tracks | ✕ | Pick any (from €99) | All 20+ included | All 20+ included | All 20+ included | All + custom modules |
| 3D AI Architect Ebook | First chapters | ✕ | ✓ Full | ✓ Full | ✓ Full | ✓ Full |
| Neurones 3D | ✕ | ✕ | Standard license | Full license | Full license | Full + team licenses |
| Code Templates & AI Agents | Starter only | Snippets only | Starter only | 50+ production + 4 AI agents | 50+ production + 4 AI agents | Production + custom agents |
| Monthly Drops (Briefing, Briefs, Templates) | ✕ | ✕ | Briefing only | Briefing + briefs + templates | Briefing + briefs + templ. | Briefing + custom briefs |
| Forge .exe Apps | ✕ | ✕ | ✕ | ✓ 5 apps | ✓ 5 apps | ✓ 5 + custom |
| Spatialetics Community | ✕ | ✕ | ✓ Access | ✓ Priority | ✓ VIP Private | ✓ Team channel |
| Job & Market Board | ✕ | ✕ | ✓ | ✓ | ✓ | ✓ |
| Sprint Slots (1-on-1 Florent) | ✕ | ✕ | ✕ | ✕ | 4 / year (w/ Florent) | Dedicated weekly |
| Services Access (R&D, software) | ✕ | ✕ | ✕ | ✓ R&D, consulting | ✓ Priority access | ✓ On-site integrated |
| Onboarding + Strategy Calls | ✕ | ✕ | ✕ | ✕ | ✓ 1h + 1h per year | ✓ Team onboarding |
| Private Channel | ✕ | ✕ | ✕ | ✕ | ✓ 48h response | ✓ Dedicated |
| Certificate | ✕ | ✕ | Completion | Professional | Architect | Team certs |
| Renewal (monthly content) | ✕ | Per course | €697/yr (fdg: €497) | €697/yr (fdg: €497) | €497/yr (always) | €497/yr (always) |
Everything You Need to Know
Do I need a PhD?
No. You need intermediate Python and basic 3D knowledge (point clouds, meshes). If you can write a for-loop and know what a point cloud is, you’re ready. I’ve trained surveyors with zero coding background and PhD researchers. Both succeed.
How is this different?
Three things. First, it’s 100% focused on 3D spatial data, not generic computer vision repackaged for point clouds. Second, you build everything from scratch. No black boxes. You understand every line. Third, production deployment is included. Real systems, not notebooks that only work on clean demo datasets. And in 2026, there’s a fourth: I show you how to use AI tools from a position of understanding, so they accelerate your work instead of replacing your thinking.
How long does it take to complete the Program?
The Spatial Accelerator gets you building from day one. Each system’s main track is 15-20 hours. Students using the accelerator path complete their first production system in under a week. Full program completion is realistic in 3-4 weeks at a focused pace. Lifetime access means you can go at your own speed. No pressure, no expiration.
Can I use this commercially?
Yes. Full commercial license. Use all code in products, consulting, or internal systems. Only restriction: can’t resell the course as training material.
What hardware needed?
Most systems run on laptops (16GB RAM minimum, 32GB recommended). For deep learning, you’ll want a GPU—but free Colab/Kaggle work fine for training. No $10K workstation needed.
What if I get stuck?
Complete documentation. Private community. Monthly office hours. Email support. Average response under 24 hours.
Refund policy?
90-day money-back guarantee. Go through the material, apply it to your data. If you don’t get real, measurable results after actually doing the work, email me for a full refund. I don’t want your money if you don’t get value. Simple as that.
Why trust you?
I have walked the whole path: field LiDAR engineer, then a PhD and postdoc in 3D spatial sciences, then professor, alongside fifteen years advising startups and consulting for teams at companies like Airbus, BMW, CNRS and Meta. Author of “3D Data Science with Python” (O’Reilly). I mentor this because the niche is hard, and I would rather build beside you than watch you give up.
