Every Model, Under Your Governance

More than 16 foundation models sit behind one governed interface, and your administrators choose, restrict, or replace any of them by policy. The model is chosen per workload, with your own keys or through Sprinklenet's managed service, and it can change without a rebuild while the access rules, retrieval limits, guardrails, and audit trail stay the same.

Request an Integration Review

Bedrock

Model Terms and Portability

For regulated and enterprise workloads we route to models under no-retention, no-training terms, and confirm the exact model terms as part of your deployment. Each answer records the provider and model version that produced it. You change models by configuration, and your data, application code, and evidence export in non-proprietary formats, with knowledge transfer as a scheduled deliverable.

What Reaches the Model

Conversation audit in Knowledge Spaces: a retained record with its session detail, retention setting, and the provider and model version that produced the answer

Connectors to the Systems You Already Run

Knowledge Spaces connects to your data sources and business systems through documented APIs and connectors, and your systems of record stay where they are. Each connector documents its authenticated identity, permitted records, allowed actions, and credential owner, and connectors ship in phases, so each source connects when you authorize it.

Evaluation jobs, model versions, and serving health
Asset registration, risk rating, and gate decisions in the governance catalog
CI status read, and gate decisions written back as commit status
Microsoft SharePoint

Current evidence documents through Microsoft Graph
Microsoft Entra ID

Owners, approvers, and governance roles
Microsoft 365

Approval requests and gate outcomes in Outlook and Teams
Remediation tasks opened, commented on, and closed
Approved policy and template versions published
Salesforce

CRM records as governed sources
HubSpot

CRM records and tickets
PostgreSQL

Live operational data through tables and custom queries

AI Integration Services How Connectors Ship

Merlin's connected platforms: Databricks, Collibra, GitHub, Jira, Confluence, Microsoft 365, and Entra ID, with the operations and endpoints each one calls

Minimum-Necessary Retrieval

For each request, Knowledge Spaces retrieves only the minimum necessary context from approved sources, isolated per tenant and logged. The model never receives your whole database or document set.

APIs We Make and Connect

The platform is API-first: bots, sessions, connectors, and governance are callable under organization-scoped keys. If the API you need does not exist, we design and build it, scope what it may access, and log every call.

Knowledge Spaces console: organization API keys, each scoped to what it may call

The first governed API call in Python from the public developer quickstart

Start at Developers

Map One Workflow Across Your Systems

Send the systems involved and the workflow you want mapped. We will produce a map of systems, data, permissions, and success measures, and a fixed scope for the build.