Mendix
Also known as: Mendix, Mendix Agents Kit, Mendix Agent Commons, Mendix Agent Editor, Mendix Cloud GenAI, Mendix, a Siemens business
Siemens' enterprise low code platform, whose Agents Kit builds chat and task agents that call microflows and MCP tools, ground on knowledge bases, pause for human approval per tool, test in bulk, and deploy from Mendix Cloud to air gapped on premises.
Mendix is Siemens' enterprise low code platform, and its Agents Kit turns agents into a native part of the apps teams already build there. Agents are defined in two ways: at design time as version controlled documents in Studio Pro with the Agent Editor, or at runtime in an Agent Builder interface supplied by the Agent Commons module, where an admin can write prompts, pick a model, attach tools and knowledge bases, test, and switch the active version without redeploying the app.
Tools are microflows, the platform's own logic units, plus any tools from an MCP server, so an agent can act on whatever the app is wired to, from SAP and Snowflake to internal REST services. Knowledge bases connect for autonomous retrieval with configurable chunk counts and similarity thresholds. Chat agents serve conversational interfaces, while task agents run background jobs or act as subagents in an agent as tool pattern. Both can be called from microflows, and BPMN workflows include an AI agent task.
Oversight is built into the framework. Each tool can carry human in the loop settings that control whether a call needs the user's approval and whether it is shown, and the Conversational UI module handles those approvals in chat. Agents can be tested with multiple sets of test values run in bulk and compared, exported between environments, and traced through GenAI Commons.
Mendix holds ISO 27001, 27017, 27018 and 27701, ISO 22301 and ISO 9001, SOC 1 and SOC 2 Type II, ISAE 3000 and 3402 Type II, C5 Type II, PCI DSS Level 1 and NEN 7510. It deploys on Mendix Cloud, a dedicated single tenant cloud, Azure, Kubernetes, SAP BTP, on premises, or an air gapped Private Mendix Platform.
Vendor details
Canonical URL
https://www.mendix.com
Category
Agent builder
Subcategory
Enterprise low code platform with an agent framework
Funding status
Owned by Siemens (Mendix, a Siemens business). Not separately funded.
Company status
subsidiary
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Agents use Mendix microflows as tools, so anything a Mendix app can reach becomes an agent action: REST, OData and SOAP services, databases, SAP, Snowflake, AWS services, Teamcenter and other Marketplace connectors. The MCP Client module adds tools from external MCP servers, and the MCP Server module exposes Mendix logic to any agent in the enterprise. Model connectors cover Mendix Cloud GenAI, OpenAI, Amazon Bedrock, Google Gemini, Mistral and Snowflake Cortex, and knowledge bases run on Mendix Cloud GenAI, Bedrock, OpenAI or PgVector.
In practice
A support team ships a conversational agent from the Agent Builder Starter App that looks up orders through microflows and asks the agent user to approve before issuing a refund.
An operations app adds an AI agent task to a BPMN workflow so a task agent classifies incoming claims before a human reviewer picks them up.
A manufacturer exposes Teamcenter backed microflows through the MCP Server module so agents elsewhere in the company can query product data.
Sources & related URLs
Related / legacy domains
Research sources
Agentic Index coverage score
12.5 / 14 capabilities · 89%
| Integrations & Tool Calling | Full |
|---|---|
|
Agents act through tools the admin attaches, microflows that the model calls by function calling with inputs it determines, and tools from any MCP server configured in the MCP Client module, either all of a server's tools or a selected subset with editable fields. Microflows reach REST, OData and SOAP services, databases and Marketplace connectors such as SAP and Snowflake. Sourcedocs.mendix.com/agents/agents-kit-2/reference-guide/agent-commonsread 2026-10-02 |
|
| Workflow Orchestration | Full |
|
Agents work with each other and with process logic. Task agents serve as subagents in an agent as tool setup, chat agents can be called conversationally by another agent, both are called from microflows and workflows by drag and drop activities, BPMN workflows include an AI agent task, and the GenAI Showcase App demonstrates multi agent patterns. Sourcedocs.mendix.com/agents/agents-kit-2/reference-guide/agent-commonsread 2026-10-02 |
|
| Knowledge Grounding & RAG | Full |
|
Knowledge bases attach to agents for autonomous retrieval, with platform supported connectors for Mendix Cloud GenAI, Amazon Bedrock, OpenAI and a PgVector knowledge base; retrieval parameters (number of chunks, similarity threshold) are configurable, and multiple named knowledge bases can be attached for the model to choose between. Sourcedocs.mendix.com/agents/agents-kit-2/reference-guide/agent-commonsread 2026-10-02 |
|
| Human Oversight & Guardrails | Full |
|
Tools carry configurable human in the loop controls for execution approval and visibility, and when a tool call is set for user visibility the agent returns the requested call instead of executing it; the Conversational UI module supports these approvals out of the box. Guardrails are named among the agentic patterns Agent Commons supports. Sourcedocs.mendix.com/agents/agents-kit-2/reference-guide/agent-commonsread 2026-10-02 |
|
| Security, Identity & Governance | Full |
|
Mendix holds ISO 22301, ISO/IEC 27001:2022, 27017, 27018 and 27701, ISO 9001, ISAE 3000 and 3402 Type II, SOC 1 and SOC 2 Type II, and C5 Type II, plus PCI DSS Level 1 and NEN 7510. Agent definition is gated by the AgentCommons.AgentAdmin module role inside Mendix app security, with SSO, access restriction profiles and Control Center administration on the platform. Sourcemendix.com/evaluation-guide/securityread 2026-10-02 |
|
| Observability & Auditability | Full |
|
The GenAI Commons module supports traceability to debug and monitor agents, agent versions are stored as snapshots with an in use version selected per environment, and the platform adds logs, metrics and OpenTelemetry export. Sourcegithub.com/mendix/docs/pull/11285/filesread 2026-10-02 |
|
| Memory & State Persistence | Partial |
|
Conversation state is kept within a chat. Chat agents receive an alternating user and assistant history through the Call Agent w/ History activity and chat contexts, while task agents run without historical context, and prompt variables pull session context at runtime. Mendix does not describe a memory that lasts across conversations. Sourcedocs.mendix.com/agents/agents-kit-2/reference-guide/agent-commonsread 2026-10-02 |
|
| Deployment & Data Residency | Full |
|
Customers choose among Mendix Cloud, Mendix Cloud Dedicated (single tenant), Mendix on Azure, Mendix on Kubernetes, Mendix on StackIT and Mendix on Server, plus SAP BTP and the Private Mendix Platform with an air gapped installation; agents deploy as part of the app model. Sourcemendix.com/pricingread 2026-10-02 |
|
| Prebuilt Agents, Templates & Packs | Full |
|
Starting points come from the Mendix Marketplace. The Agent Builder Starter App ships a working conversational support agent with tools, knowledge base and human in the loop already wired, the Blank GenAI App comes preloaded with model connectors and Agent Commons, and the GenAI Showcase App holds example multi agent, MCP, RAG and function calling patterns. The catalog is a handful of starters. Sourcedocs.mendix.com/agentsread 2026-10-02 |
|
| Triggers & Channel Coverage | Full |
|
Agents run wherever app logic calls them. Task agents are designed for background processes, the Call Agent activities sit in microflows (which the platform starts from scheduled events, published REST operations, business events and page actions) and in BPMN workflows, and chat agents serve Conversational UI chat interfaces. Those start paths are features of the Mendix platform that call the agent. Sourcedocs.mendix.com/agents/agents-kit-2/reference-guide/agent-commonsread 2026-10-02 |
|
| Model Flexibility & Routing | Full |
|
The customer picks the model for each agent version, associating it with a deployed model from any GenAI Commons connector (Mendix Cloud GenAI, OpenAI, Amazon Bedrock, Google Gemini, Mistral, Snowflake Cortex), with temperature, top P and max tokens configurable, and teams can build their own connector. Sourcedocs.mendix.com/agents/agents-kit-2/reference-guide/agent-commonsread 2026-10-02 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
The MCP Server module makes Mendix business logic available to any agent in the enterprise, agents can be created programmatically, Mendix publishes the Platform SDK and Model SDK plus REST APIs for apps, deployments, builds, pipelines and audit logs, and deploy APIs are also offered to AI agents over MCP. Sourcedocs.mendix.com/agents/agents-kit-2read 2026-10-02 |
|
| Testing, Debugging & Optimization | Full |
|
Agent Commons includes logic to define and run tests individually or in bulk with result comparisons, multiple sets of test values can be saved and run together against an agent version in an isolated environment, and versions are compared before one is made active. Choosing the version in use in each environment is always a manual step. Mendix does not describe scoring or an automated judge. Sourcedocs.mendix.com/agents/agents-kit-2/reference-guide/agent-commonsread 2026-10-07 |
|
| Browser & Computer Use | Not documented |
|
Mendix does not describe agents in Agents Kit controlling a browser, desktop or remote computer. Its ATS test automation drives browsers through Selenium, but that is a testing product, not an agent capability. Sourcedocs.mendix.com/agentsread 2026-10-02 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Product updates
Studio Pro 11.16 adds Plan Mode to Maia, the agent built into the Mendix IDE, so it drafts an implementation plan from a prompt that developers can review and refine before any build starts. Studio Pro now includes Playwright, which Maia uses to check the running app in its PDCA cycle. The Changes pane also shows which elements and properties Maia changed inside documents.
Bears on: Human approval / guardrails
View sourceIn Studio Pro 11.15.0, Maia, the Mendix development agent, can start and stop the local runtime, check its status and read its logs, so it can test the changes it makes. The same release has Maia ask before running tools that reach external resources or make changes, with allow once, for the session or always.
Bears on: Human approval / guardrails
View sourcePricing
Standard from $1,090 a month for one app, plus fees per named end user. A free plan is available.
A platform package fee for one app or unlimited apps, plus a monthly fee per named end user per app. Compute and GenAI resource packs are billed separately.
Included quota
Standard includes up to 3 environments per app and business hours support. Premium has flexible environments and 24/7 support.
What is public
Mendix publishes its package names, starting monthly prices for Standard (one app and unlimited apps), environment counts, support levels and deployment options. Per user rates and Premium pricing are not published.
Billing mechanics
Monthly package fee for one app or unlimited apps, plus a per user per app fee for named end users. Anonymous users and developers are not charged.
Cost watchouts
End user fees per app per month sit on top of the package price. Standard and Premium exclude the compute an app runs on. Mendix Cloud GenAI model and knowledge base capacity is sold as separate resource packs whose prices are not published. Other model providers bill directly.
Variable cost rationale
The package fee is fixed per app or for unlimited apps, and three costs sit on top of it. Named end users are billed per app per month, so the bill grows with adoption, though the fee per user falls as volume grows and anonymous users and developers are free. Compute is extra on Standard and Premium, and Mendix Cloud GenAI model and knowledge base capacity is bought as separate resource packs. No per user rate, compute price or resource pack price is published.
Additional watchouts
At enterprise scale, end user fees usually outweigh the package fee.
Overage / add-ons
Additional named end users add a monthly per user fee that falls as volume grows. External users get lower rates.
Sales call required
Mixed (some tiers require a call)
Free / trial
Free plan for building and testing apps, including Agents Kit
Lowest paid plan
Standard from $1,090 a month for one app
Commercial notes
The free plan is self serve. Standard and Premium go through sales, with deployment on Mendix Cloud, Dedicated, Azure, Kubernetes, StackIT or customer servers.
Key ambiguities
Per user rates and GenAI resource pack prices are not published.
Support SLA / resale
Standard has support from 9 AM to 5 PM and a 99.5% uptime SLA, and Premium has 24/7 support and a 99.95% SLA. The free plan carries no support plan or uptime guarantee.
Missing data
Per user rates, GenAI resource pack prices and Premium pricing are not published.
Related vendors
- Lindy — Lindy is an AI teammate that lives in Slack, answers from your tools…
- Activepieces — Open source, MIT licensed automation platform where every one of 700…
- Agenta — Agenta is an open source workspace where teams build AI agents by…
- AgentX — No code platform to build, evaluate, and deploy multi agent…
- AI Library — No code platform for building and running AI agents, with a…
- Aigensei — Enterprise agentic workflow platform that runs business processes…
Alternatives to Mendix
The closest documented capability profiles to Mendix among agent builders tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- OutSystems12.5 / 14Matches Mendix across all 14 documented capabilities
- AgentX12.0 / 14A lighter documented profile than Mendix
- Airia12.0 / 14A lighter documented profile than Mendix
- Dify12.0 / 14A lighter documented profile than Mendix
- LuMay12.0 / 14A lighter documented profile than Mendix
- Moveo.AI12.0 / 14A lighter documented profile than Mendix
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded