Use multiple AI vendors interchangeably

Montag.AI supports multiple AI vendors, including open source, and allows you to switch between them at will, dynamically.

For example, for highly-complex answers, you may wish to use GPT-4, however for sensitive tasks, you may want to direct a task at an internal open-source model.

Alternatively, you may want to ensure your chatbot, or smart-application, can handle long inputs, and so you may want to use a model that is optimised for long inputs when these arise, such as Anthropic Claude-2, without having to build your entire stack on a sinlge Vendor.

Montag makes switching between models while using the same prompt easy, and allows you to manage your AI models in one place without exposing credentials to your team.

Build and launch chatbots quickly

Montag features a built-in slack integration and OpenAI chat API shim. We can have you set up with a GPT-powered chatbot infused with knowledge of your documentation, manuals and other data in minutes rather than weeks.

Smart bots, out of the box

Montag.AI comes with it's own web-scraping engine, that can ingest any content on the web, give your bot access to content repositories such as ZenDesk, JIRA, and Confluence in a secure and self-managed way.

Upload a zip-file of your content, and Montag handles ingesting, chunking, vectorising, and enhancing your content for you, making it quick and easy to create AI-powered applications that are informed by your specialist knowledge domain.

Use Montag's powerful ingestion pipline and webhook support to send content into your vector database - e.g. as Github Actions, or CI/CD pipeline ops.

AI-as-APIs

Montag.AI enables your team to create REST APIs that feed into pre-programmed AI prompts, like stored queries in a database and expose them as REST APIs. Make common AI use-cases as easy as a REST call.

Go even further with Scripting

Cmbine multiple LLMs, content repositories, and external content into more capable and specialised applications. Scripting enables your team to call on all the functionality within Montag from inside a robust and fast scripting language.

Scripts can be called from REST endpoints, intercept chatbot inputs, or intercept LLM outputs to enhance and enrich them. Scripts can also be used to create custom chatbot responses, or to create custom chatbot commands.

Build multi-modal, multi-vendor chatbots that can call on multiple content repositories to provide expert functionality to your users in a matter of minutes.

Security and Observability

Montag.AI has a granular Access Control system system with support for OpenID-compatible external IDPs (SSO). Ensuring you can integrate Montag into your organisation quickly and securely without new user repositories.

Montag's granular Access Control capability enables you to create Workspaces for teams, and secure upstream AI vendor credentials and API secrets away from your AI Developers using per-object-type read, write, delete, and listing permission scopes.

Montag has full analytics support, and records all inbound and outbound data in it's database with full chat logs. This ensures you have full visibility of what is being used, when, and by whom.