| Platform approach |
A complete platform for building and operating custom AI agents: centrally manage agent creation, knowledge, models, integrations, governance, and execution. |
An agent platform within the Microsoft ecosystem, connected to Power Platform and Microsoft services; extensively configurable and extensible.[1][2] |
| Internal and external applications |
For internal business processes and external applications, such as customer service and communication. Custom rules and action permissions define how agents can be used. |
For internal and external applications; deployment across multiple channels.[1] |
| Required expertise |
Business users create, understand, and maintain even multistep workflows in natural language. No programming or knowledge of node-based modeling is required. IT is responsible for identities, permissions, and technical integrations. |
Simple agents can be created without programming. For visually modeled workflows, authors need to understand the nodes, conditions, and flows used. Organization-wide operations, policies, and controlled publishing require Power Platform expertise.[1][3] |
| Agent creation |
Creation in natural language: an AI assistant clarifies rules, exceptions, and approvals through a guided interview. This produces an executable “routine” in text form. Tools and knowledge blocks are provided visually and referenced in the text. Schedules and collaboration between multiple agents can also be controlled through natural language. |
Creation through natural language, with AI assistance generating a node diagram. Instructions, knowledge, and tools can be orchestrated; explicit process logic can be modeled using nodes and flows.[1] |
| Process description and maintenance |
The readable routine text is the automation itself and remains the basis for ongoing work. Business teams edit rules and process steps directly in the text, then test and publish changes within their permissions. No additional node diagram needs to be maintained. |
Maintenance involves instructions and the platform components used. Explicitly modeled workflows require changes to nodes, conditions, and flows. The expertise required depends on how the agent is built.[1] |
| Setup and ongoing responsibility |
IT sets up access, system connections, and central policies. Business owners maintain content and test cases; SCALAN can support implementation. |
Organizations or partners configure and maintain environments, roles, data policies, connections, and the publishing process.[3] |
| Expansion to additional departments |
New agents use the same platform framework with their respective permissions. Business teams can implement additional use cases independently or with SCALAN’s support. |
Additional teams are onboarded through roles, groups, access grants, and the chosen environment structure; central policies can be reused.[3] |
| Governance and action permissions |
IT centrally defines permitted models, systems, and individual actions. Business teams cannot independently expand these boundaries. Human approvals can be incorporated into workflows. |
Data policies govern connectors, knowledge sources, and channels, among other things. When integrated accordingly, Agent 365 adds centralized monitoring and policy enforcement.[2] |
| Identities and accountability |
SSO, Microsoft Entra ID, and SCIM are supported. Agents belong to the organization; each agent has a designated owner. |
Roles, teams, and Entra ID groups govern access and collaboration; Agent 365 adds agent identities and ownership information.[2][3] |
| AI models |
Models from OpenAI, Anthropic, and Mistral; centralized approval and selection of a suitable approved model for each workflow. |
Multiple model providers; availability, approval status, and processing terms vary. Certain external models are in preview or experimental.[4] |
| System integrations |
Over 50 integrations in production; custom ERP systems and in-house applications can be connected. |
A broad connector ecosystem for Microsoft and third-party systems, along with custom extensions.[1] |
| Testing and publishing |
Reusable test cases, including document processing, plus versioning and publishing. Quantitative evaluations are in development. |
Reusable test sets, quantitative evaluations, and comparisons of test runs; controlled publishing through corresponding lifecycle processes.[3][5] |
| Traceability and intervention |
Trace execution steps, tools used, results, and consumption; pause or deactivate agents and stop individual executions in progress. |
Activity Maps show inputs, decisions, and outputs. Centralized governance and usage controls complement execution analysis.[2][6] |
| Data location |
Platform and AI model processing takes place in Frankfurt. The processing locations of connected third-party systems must also be considered. |
EU Data Boundary with the appropriate tenant and environment configuration. Data flows also depend on features, models, and connected services.[4][7] |
| EU AI Sovereignty |
A German platform provider, processing in Germany, and a choice of model providers. European models can form part of your AI strategy. |
European data residency is possible; the agent platform remains part of the Microsoft ecosystem. External models are also available.[4][7] |
| Pricing and billing |
SCALAN Business starts at €129 per month with unlimited users; usage-based billing through tiered credits. |
€173.30 per month per pack of 25,000 Copilot Credits; alternatively, usage-based billing with an Azure subscription. There is no blanket requirement for every agent user to have a user license.[8] |
| Included capabilities and additional effort |
A shared platform for creation, governance, and operations. SCALAN can support implementation and agent development; custom services and plan inclusions are specified in the proposal. |
The agent license includes entitlements to premium connectors. Additional services, licenses, and implementation work depend on the scenario.[3][6] |