Develop the model layer
Advance an in-house model program alongside the company's agentic software portfolio.
All innovations
Trunnion LLM: Trunnion AI's in-house model layer
Trunnion LLM is the company's in-house model layer, currently in development. It is a model-building program - not a new label for third-party model routing.
AI-generated illustration · Not a company or customer location
Built forModel & application teamsProduct direction · in development
See the work in context
Explore the product direction.
An illustrative workspace for model & application teams. Explore the work and the decision behind it.
Illustrative work queuePrepared for review
Task definitionSelected example
Evaluation caseRelated work
Release reviewRelated work
The next decision
Define an evaluation case
An intended product task needs explicit evaluation criteria and a documented release boundary.
- Decision owner
- Model evaluation team
- Action boundary
- Review before action
Illustrative decision reviewHuman decision required
Proposed next step
Define an evaluation case
Review behavior and limitations before making a release decision.
Responsible reviewerModel evaluation team
- 01Check the context
Review the material behind the proposed action.
- 02Resolve the boundary
Confirm scope, permissions and consequences.
- 03Decide what proceeds
Approve, revise or decline through the defined workflow.
Concept only. Features, controls and records are confirmed for the proposed release.
Why Trunnion LLM exists.
An enterprise model needs to be evaluated in the context of the work it will support. Capability, limitations, provenance and operating conditions all matter. The LLM program creates a dedicated place for that development within Trunnion.
Built by Trunnion AI, a wholly owned Duskbridge company. This page presents the direction of a product in development; it does not promise release timing or available features.
An in-house model development program.
Evaluate against real tasks
Define the evidence needed to assess usefulness and limitations for intended product work.
Prepare for responsible release
Establish the release boundary before describing model capabilities as available to customers.
From a defined task to a release decision.
An illustrative sequence, not a product screenshot or a guarantee of released behavior.
Define the task
Identify the intended work and evaluation context.
Develop
Build and assess the model layer.
Evaluate
Review behavior, limitations and the proposed operating conditions.
Set the release boundary
Publish supported capabilities only when established for a release.
Make the evaluation specific.
Start with the workflow you need to improve. Then confirm what the proposed release can actually do.
- What task and model scope will a future release address?
- What evaluation evidence will be available?
- What licensing and deployment conditions will apply?
- Status
- Building
- Portfolio
- Innovations
- Technology owner
- Trunnion AI, LLC · Duskbridge
- Release scope
- Confirmed in the written agreement
What to know
about Trunnion LLM.
Is a Trunnion model publicly available?
This page does not announce a public model endpoint, weights or model license. The program is building.
Is this the same as Core's model routing?
No. Core is the orchestration runtime and developer platform. Trunnion LLM is the in-house model layer.
Are there published benchmark claims?
No performance or benchmark claim is established here. Such claims need release-specific evidence.
Related products & technology.
Portfolio relationships do not establish an included integration or entitlement.
Help shape the right product conversation.
Start with a non-sensitive overview. Product fit, release scope and next steps belong in a direct conversation.