Model card
all-ian-v40-lts
Long-context judgement model running on a biological substrate with professional, security-aware and charity-sensitive expert modes.
Technical model card
all-ian-v40-lts
| Model | all-ian-v40-lts |
|---|---|
| Category | Human Programming Interface |
| Best at | Messy context, operational judgement, security-aware planning, charity-sensitive delivery |
| Intended uses | Operating model reviews, AI remediation, evidence packs, supplier sanity checks, plain-English briefings |
| Out of scope | Rubber-stamping bad strategy, pretending token pricing is destiny, replacing accountable humans |
| Limitations | Finite biological runtime, bluntness under load, cannot make broken governance become governance by naming it |
| Safety posture | No medical, legal or financial advice without qualified review; no confidential data without an agreed route |
Modes
| Mode | Behaviour |
|---|---|
| professional-blunt | The default. Kind enough to be useful, direct enough to save time. |
| charity-sensitive | Optimises for proportionate risk reduction and limited budgets. |
| broadcast-explainer | Turns technical mess into audience-safe clarity. |
| cleanup-response | For AI, supplier or development-team damage that already happened. |
| sleep-consolidation | Daily memory consolidation during 6-8 hours of biological downtime. |
Limitations
No model should be used outside its operating envelope.
- No confidential data without an agreed route.
- No medical, legal or financial advice without qualified review.
- No pretending that a summary is the same as accountability.
- No replacing humans who own decisions with an output-shaped comfort blanket.