Multiple classification methods
Use the method appropriate to each hazard class rather than forcing every decision through one generic scoring model.
- ATEmix for acute toxicity
- Cut-off thresholds
- Summation and M-factor logic
CLP classification
Classification you can explain line by line.
Calculate mixture hazards from structured substance data and preserve the evidence behind every triggered or non-triggered classification.
SDS Labs
Structured workspace
Classification result
Acute toxicity · Oral
ATEmix
1,640 mg/kg
Threshold
≤ 2,000
Contributors resolved3 substances
Rationale stored
0 LLM
Calls inside classification
3 routes
Oral, dermal, inhalation ATEmix
1 rationale
Stored with every result
A calculation engine, not a compliance chatbot.
SDS Labs applies encoded CLP methods to the actual composition, using the actual concentration when known and the maximum concentration as the conservative fallback. Results remain reproducible and reviewable.
Complete hazard communication output
Resolve H-statements, derived P-phrases, GHS pictograms, and the signal word from the final classification.
- Pictogram precedence
- Danger before Warning
- Deduplicated precautionary phrases
Rationale for every hazard class
Record contributors, concentrations, intermediate calculations, the applied threshold, and why the class did or did not trigger.
- Audit-ready evidence
- Intermediate calculation values
- Explicit trigger decisions
How it works
A repeatable path from composition to outcome.
The engine resolves source data first, applies hazard-specific methods, and then assembles the final communication set.
-
01
Resolve inputs
Load substance hazards and concentration values.
-
02
Apply methods
Run ATEmix, cut-off, and summation calculations.
-
03
Apply precedence
Resolve pictograms, signal word, and phrase sets.
-
04
Persist rationale
Save the result and the complete audit trail.
Deterministic means deterministic
Same structured input. Same classification output.
The classification layer makes no external HTTP calls, performs no language-model inference, and introduces no randomness into regulatory calculations.
Keep exploring
Related workflows and guides
Build with confidence
Connect the regulatory decision to the product work that follows.
Bring mixture data, classification, labels, SDS documents, and market context into one traceable workspace.
CLP Annex I · REACH Annex II · 24 EU languages · Audit-ready rationale