Strengthen your analyses and recommendations
Manual characterizations take time, vary from one operator to the next, and don’t always deliver the precision needed to back up solid recommendations.
With Akanthas, offer your clients an innovative service: characterizations that are faster, more reliable, and safer.
- Advanced AI-powered waste analysis
- Reduced characterization cost and time
- Automatic regulatory report generation
The challenges engineering firms face
The pressure to innovate
Recruitment difficulties
Sample sizes that are too small
Save time on every characterization
With automatic waste identification, characterization goes from several hours to a few minutes.
Straight on your client’s site, deploy the device effortlessly. Then trigger the analysis of a waste stream in a few clicks — our sensor identifies every visible item.
Save precious time on reporting, too. A full report is auto-generated at every characterization.
Let the AI handle the most time-consuming tasks.
Back your recommendations with reliable data
Our patented AI identifies more than 60 waste categories, guaranteeing standardized, comprehensive, and objective analysis.
By removing physical sorting work, Akanthas doesn’t just reduce human error — it reduces risk to your teams.
Less handling, less exposure to sharps, hazardous substances or undesirable objects: characterization becomes safer for everyone.
From the dashboard, you can complete or adjust the analysis to reach maximum precision.
Stand out from the competition
Gain a competitive edge by offering audits that are faster, more reliable, and safer.
By automating time-consuming tasks, increase your capacity to handle more projects without compromising quality. For example, automatic identification of recoverable materials lets you advise clients more effectively on their recovery strategy.
Let the AI handle data analysis and formatting — and finally focus on your core business.
Features
Features built for engineering firms
Advanced automated characterization
Our sensors with embedded AI run automated characterization on visible waste. Each one can cover up to 4 containers or a 100 m² ground area.
Waste is identified then classified across multiple criteria:
- Mass proportions
- Volume proportions
- Object counting
- EPR classification
- Decree 2021-1199 waste classification
- Classification by local recovery channel
Automated reporting
- Decree 2021-1199
- EPR (extended producer responsibility) channel
- OMR (household waste) approval
Visualization and adjustment
For every characterization, you can visualize the different identified materials separately.
Need to clarify something on a specific item? Annotate the analysis directly by pinning comments on the photo to complement the AI’s work — and sharpen your audit.
Mobile device
Solar-powered and equipped with a 4G antenna, our sensors require no construction work. Our mobile device lets you deploy a sensor directly on the client’s site using a telescopic mast.
Our team trains you to use it so you can operate fully independently — we show you how to:
- Configure the reference zone
- Deploy the sensor
- Capture the image
- Use the dashboard, accessible from any browser.
A complementary
method
| Criteria | MODECOM | Weight-based characterization |
Visual characterization |
|
|---|---|---|---|---|
| Type of characterization | Manual | Manual | Visual | Automatic |
| Frequency | One-off (rare) | One-off (a few times/year) | Very frequent but superficial | Very frequent and automatic |
| Representativeness | High but limited sample | Variable depending on opened batches | Low (subjective, narrow scope) | Very high (analysis of all flows) |
| Time required | About 5 to 7 hours | Over 4 hours | A few minutes | A few minutes |
| Operator safety | High exposure | High exposure | High exposure | Zero risk exposure |
| Objectivity | Standardized method, but handled by humans | Risk of human bias | Highly subjective | 100% algorithmic, defensible |
| Overall cost | Very high | High | Low but limited quality | Low (amortizable, no mobilization) |
| Type of data | Quantitative, standardized | Quantitative but sampled | Qualitative and indicative | Quantitative, continuous, granular |
| Use for operational steering | Low (one-off sample) | Medium (corrected by frequency) | Low (not usable in depth) | Excellent (frequent, comparative, continuous) |
| Regulatory / EPR usefulness | Strong: official standard | Medium | Low | Strong: defensible data, frequent |
| Level of detail | 120 categories | Simplified | Limited | 60 categories and 6 classification types: mass, volume, objects, EPR, recovery channel, particle size |
Process
How does a characterization work?
1
Setup
2
Photo and analysis
3
Data entry
4
Waste labeling
5
Report generation and review
Upgrade your characterizations with AI today
These engineering firmsalready use Akanthas
- Waste Quality
- Sorting & treatment centers
What our customers say
Nicolas Klethi
Simon Loisel
Samuel Chazarin
Everything you need — only what you need
Choose the modules your business needs and never pay for features you won’t use.
A complete, tailored tool that evolves alongside your needs.
- A modular solution adapted to your needs
- A patented AI solution
- Solar-powered, self-sufficient sensors
- A dashboard accessible via the web — or connectable via API to your tools
- An intuitive mobile app
- Support with installation and onboarding
- A dedicated advisor available at all times
- Monthly review meetings
Contact us
Got a question about our solutions?
Blog
Resources & news
Get our brochure
Got a question?
Does your solution replace the MODECOM protocol?
Our solution doesn’t replace MODECOM: it complements it — and sometimes serves as an alternative, depending on the use case. MODECOM and AFNOR standards remain references, but they’re manual, one-off, and costly methods, run on small samples.
On our side, we bring continuous, automated characterization based on larger volumes, with a high level of reliability. For many needs (internal KPIs, EPR obligations, regular monitoring), our approach is already being used as support — or even as the basis for decision-making — because it’s more responsive, less intrusive, and far more economical.
And if needed, our results can be cross-checked with a MODECOM campaign to round out the analysis.
How far can Akanthas' AI go in recognizing objects and materials in waste?
Our AI’s recognition capability depends on the type of object, how often it appears in your flows, and the quality of the images captured.
For the most common objects, our models are already trained. For more specific cases, we gradually enrich our database with images collected on your site and your field feedback.
Our AI identifies material families and standard objects observable to the human eye. On the other hand, it doesn’t perform chemical composition analysis (for example distinguishing PP, HDPE, etc.).
Like any learning system, its performance evolves with exposure: the more it observes, the more it learns — and the more precise it becomes. Some rarely-occurring materials (like asbestos) are still less well-recognized, but will strengthen as real examples are observed. It’s an evolving system that gets more powerful as it observes your flows.
How can I be sure your AI actually works — and isn't just a bunch of promises?
We’re not selling a concept: our algorithms are already deployed in the field, in complex settings, and their effects are measurable.
They translate into fewer reclassifications, better-managed pickups, time saved, and stronger reliability. We understand that AI stays abstract as long as you haven’t seen the results.
That’s why we systematically provide visual evidence, data, and specific examples of what our AI actually detects. You can see for yourself that it’s not a vague promise, but a documented operational reality.
What solutions could we consider with a limited budget?
Akanthas can adapt to different situations. We can prepare the ground together, refine the scope, document potential gains, and be ready for the next budget window.
We can also consider a very limited test or a light first step to start capturing value without committing a large budget — and validate the solution’s relevance right now.
Which companies already use your solution?
Several companies — like Veolia, Spie, Paprec and Suez — as well as local authorities, already use our sensors and our fill, characterization, or estimated weighing solutions (across varied contexts such as bulky waste, drop-off centers, or landfills).
We can share a reference list with their agreement and, if you’d like, connect you directly with one of their managers for a direct exchange.
Our solution for engineering firms
