TECH-IS Model Routes

Give the AI models in your applications a clear structure

Know which models your application is configured to use, how it should respond to eligible failures, and what actually ran. Review model changes, test them in your application and choose when to adopt them.

TECH-IS Model Routes brings those decisions into a consistent workflow. Start with the free Runtime and Assistant. Add Management, including Studio, for a visual workspace and shared configuration, together with practical help applying the approach to your application.

Explore the free Assistant preview · Discuss Model Routes with us

Runtime and Assistant are available in a Windows x64 preview. Management and Studio remain in preview; contact us to discuss availability and suitability for your team.

Make model choices easier to understand and maintain

A Route is a named model configuration: a primary model, its settings and optional backups in a defined order. A Route set groups Routes for an application, environment or purpose.

For example, document extraction and summarization can have separate Routes, each with settings and backups appropriate to the work. Your team can review those choices together and revise them without losing their identity or history.

Model Routes helps you:

• Keep a record of model choices. Inspect configurations and available model facts instead of piecing together scattered settings.
• Plan for eligible failures. Define ordered backups that the Runtime can try within supported execution limits. Fallback does not cover every failure or replay application side effects.
• Change models deliberately. Review a candidate, check compatibility, qualify it in your application and explicitly activate the chosen configuration. Retain compatible previous configurations for rollback.
• Understand configuration across applications. Organize Route sets with meaningful names and inspect shared assignments. Execution receipts provide the separate evidence of what actually ran.
• Review lifecycle and value changes. Use published model facts and reviewed advisories to investigate retirement, capability or pricing changes. Evaluate alternatives against your workload before treating a lower token price as a saving.
• Inspect execution and usage. Use receipts to investigate the selected model, attempts, fallback, outcomes and available usage or estimated cost. Unknown usage or cost remains visible as unknown.

Your application retains its provider credentials, instructions, tools and execution logic. It decides when to adopt a configuration; downloading new catalog information does not change the models it uses automatically.

Start with a useful free workflow

Configure → run → inspect → revise → qualify → activate or roll back.

Model Routes Runtime

Embed structured model execution in a TypeScript/Node.js application using the supplied contracts and supported provider adapters. Runtime handles configured model calls, eligible ordered fallback and execution receipts. Your application can retain its configuration and run without a Management connection.

Model Routes Assistant, including Advisor

Use Assistant to discover an existing integration, create or revise local Routes, review proposed changes, retain history and work through application qualification and adoption. Advisor helps investigate suitable model configurations. Local Route sets and named system instructions are part of the free workflow.

The public Assistant runs in Codex. The Claude Code integration is currently a private preview. Coding-tool access and provider services have their own requirements and possible charges.

The free tools also support compatible published catalog downloads and local matching of supplied reviewed advisories to your primary and backup models. Catalog information helps you investigate choices; it is not a continuous monitoring service or a guarantee that a model fits your application.

You can use the covered free packages in commercial applications under their licences. Buying Management, Studio or professional assistance is optional. Provider inference and your own infrastructure remain separate costs.

Add a visual workspace and shared configuration

Model Routes Management includes Model Routes Studio and the server-side Management package.

Studio provides a visual way to inspect model information, edit Routes and Route sets, review configuration history and examine locally retained execution receipts and usage. It gives leads and reviewers access to that work without requiring a coding-assistant conversation.

Use Studio in a local workspace, or deploy Management to give your team a shared, access-controlled configuration service. The server software is included in the package; you do not need a second software purchase to host it yourself.

History, local authoring and Route sets are already part of the free tools. The paid package adds the supplied visual experience, shared service and practical assistance. Hosted central receipt reporting is not included as an available feature of the current preview.

Apply our experience to your application

The product is shaped by our experience embedding AI in applications. The included expert time helps your team apply that experience to its own model choices, failure handling and change process.

The Management package includes three expert hours. Use them for a session with up to 15 participants, shorter working discussions, a Route design review, evaluation planning, investigation of receipts or agreed integration assistance. A presentation is optional.

Agree the objective and estimate before work begins. Customer-specific preparation and follow-up count toward the three-hour total. The allocation can be used within 12 months of delivery acceptance, under the agreed booking terms. Additional work can be scoped separately.

You receive both the software and access to practical judgment: what to preserve in an existing application, what to check before a model change, and how to make the resulting behavior understandable to your team.

Management package and pricing

Preview package — enquire about availability.

The planned customer-hosted package includes:

• Model Routes Studio for Windows x64.
• The Management server package and deployment documentation for the supported Azure profile.
• Three expert hours, usable flexibly as described above.
• Twelve months of product updates and standard product support from recorded delivery acceptance, within the maintained-release policy.
• Continued use of the supplied software versions under the paid licence after maintenance ends. Further updates and support can be renewed separately.

These are one-time prices per organization for the same package. The introductory window has not opened; its dates will be announced before launch. During that window, TECH-IS's written acceptance of the signed order secures the introductory price. The promotional period does not shorten your software rights, maintenance or expert-time allowance.

Applicable currency, taxes, delivery and service terms are confirmed in your order. The participant limit applies to a session, not to the package price per person. Continued software-use rights do not promise perpetual updates or compatibility with future provider and platform changes.

Ask about the Management package

Choose who hosts Management

Host it yourself

The package includes the server software. The current qualified deployment profile is Azure Functions on Linux with Node.js 24, Entra identity and Azure Blob storage. Your team provides and operates its Azure environment, access controls, monitoring and backups. Studio is a Windows x64 application and can also be used locally without a shared server.

Ask TECH-IS to host it

TECH-IS-operated hosting is a separate optional service. Hosting scope, price and operating commitments are agreed separately; it is not included in the one-time package price. Contact us to discuss the service you need and its availability.

In either arrangement, your application continues to call its providers using its own credentials. Provider inference charges remain yours; Model Routes does not add a per-token toll.

Bring one application, or a team of applications

You do not need a large fleet to benefit from clearer model choices. Start with an application whose models matter to its behavior, understand the existing integration, and keep the choices you intend to preserve.

Create or review its Routes, run representative checks, inspect the results and activate the configuration when you are ready. Add shared Management when coordination becomes useful, and use the expert hours where your team needs judgment or assistance.

Explore the free Assistant preview · Talk to TECH-IS about your application