TriOngle

AG11Govern AI agent scope

Model migration and routing review

Before you switch models or route between them: evaluation parity, where each provider keeps your data, what each licence requires, and what happens when a provider fails.

Price on requestAsk about AG11How pricing works

What it is for

Changing models is now routine: a cheaper model for one task, an open-weight model on your own hardware for another, a router in front of several. Each change can alter security behaviour, data handling and your legal position. AG11 checks before you switch.

What we look at

  • Evaluation parity: your security and quality scenarios run against the old and the new model, so the change is measured, not assumed.
  • Data: for each provider, where prompts and outputs are processed and stored, how long they are kept, and what abuse monitoring retains. Providers differ, and their terms change.
  • Licences: open-weight licences range from Apache 2.0 to provider-specific licences; for example, Mistral publishes some models under Apache 2.0 and its largest under its own open-weight licence,⁠[1] and AI21 publishes Jamba weights under the Jamba Open Model License.⁠[2] Some licences carry attribution or usage conditions that apply when you deploy.
  • Failover: what your system does when a provider is down, slow or changes a model under you.

Every provider we work with is listed, in plain text, on the platforms page.

What you receive

Parity results, a residency and retention matrix, a licence and attribution check, and a failover plan.

Framework mapping

The frameworks this scope is aligned with, the version we map to, and what we map. Aligned with means we use the framework; it does not mean any body has certified TriOngle.

FrameworkVersionWhat we map
OWASP Top 10 for LLM Applications2025 edition⁠[3]LLM03 Supply Chain
NIST AI Risk Management Framework1.0, 26 Jan 2023 (being revised)⁠[4]Third-party and value-chain risk
GDPR, Regulation (EU) 2016/67927 Apr 2016⁠[5]Processing and transfer of personal data by providers

References

Numbered in the order they are first cited on this page; each entry links back to the places that cite it. Every source was opened on 4 November 2026.

  1. ↑ Models Overview. Mistral Docs.
  2. ↑ Jamba: Hybrid Transformer-Mamba Models. AI21.
  3. ↑ 2025 Top 10 Risk & Mitigations for LLMs and Gen AI Apps. OWASP Gen AI Security Project.
  4. ↑ AI Risk Management Framework. NIST.
  5. ↑ Regulation (EU) 2016/679 (General Data Protection Regulation). EUR-Lex, 27 April 2016.