Appian vs Camunda (2026)

Agents "perform actions such as routing cases, updating records, or starting processes" through tools; process models configured as tools create or update records through a Write Records node; the Query Records tool reads synced record types; and "AI agents can also use external tools hosted on MCP servers".

The platform and pricing pages list 200+ prebuilt connectors, REST and SOAP, and ODBC, JDBC and message queue integration, with Snowflake and Kafka on Advanced and Premium. Process models, expression rules and other agents serve as custom tools. Query Records is read only and does not apply record security expressions, and writes go through process models.

Sourcedocs.appian.com 26.8 create-and-configure-ai-agent, ai-agent-reference; readread 2026-09-17

Agents act through the BPMN activities a builder places in an ad hoc sub process, so any service task, connector call or sub process becomes a tool the model can choose. Camunda offers more than 400 prebuilt connectors, including Salesforce, SAP OData and RFC, HubSpot, Slack, Microsoft Teams, GitHub, Twilio, WhatsApp, Google Drive and Sheets, Amazon S3, DynamoDB and Lambda, plus generic REST, SOAP, GraphQL and SQL. Credentials sit in secrets, a connector SDK builds custom connectors, and MCP client connectors let an agent discover and call tools on outside MCP servers.

Sourcedocs.camunda.io/docs/components/connectors/out-of-the-box-connectors/available-connectors-overviewread 2026-10-12

Appian combines deterministic process steps with agent steps. "Process agents run inside a process model using the Execute AI Agent smart service", where the agent interprets the prompt, selects tools and executes each step.

The configuration guide says: "Keep deterministic or rule-based logic (such as conditional flows, validations, or calculations) inside the process model itself, and let the AI agent handle only the reasoning or decision steps." Process models serve as agent tools, and "agents can be added directly as tools" for multi agent collaboration up to two levels deep. A flow can therefore sequence, branch, retry and route around autonomous agent steps, and agents are versioned.

Sourcedocs.appian.com 26.8 agent-studio, create-and-configure-ai-agent, ai-agent-reference; readread 2026-09-17

Processes run on the Zeebe engine as BPMN, with gateways, timers, events, sub processes and DMN decisions. An AI agent is an addressable loop inside that process. The model picks the next activity from an ad hoc sub process, the engine runs it as a governed step, and the result goes back to the model until the goal is met. Several agents can share one process, agents built in LangGraph, Amazon Bedrock or custom code can be connected and tracked as agent instances, and predictable steps run straight through while agents handle the judgment calls.

Sourcedocs.camunda.io/docs/components/agentic-orchestration/ai-agentsread 2026-10-12

Process agents run as a step inside a process model through the Execute AI Agent smart service, so they start whenever the process does, and the AI Agents overview says agents "trigger workflows automatically" and are used to automate case triage.

The pricing entitlements list consuming events from external Apache Kafka topics on Advanced and Premium as a way to start a process, so an outside event can reach an agent with nobody asking. Chat agents respond when an end user messages them.

Process start events such as timers, record events or messages are not described for agents, and many runs in the agent summary's Initiator field are started by people.

Sourcedocs.appian.com 26.8 agent-studio; appian.com/products/pricing (read 7 August 2026); readread 2026-09-17

Processes start or move on without anyone pressing a button. Inbound connectors listen on an HTTP webhook, Kafka, RabbitMQ, Amazon SQS and SNS, Amazon EventBridge, Slack, email, Microsoft 365 mail, GitHub and Twilio, and BPMN timer and message events start or resume instances on a schedule or when a message arrives. An MCP start event registers a deployed process as a tool that outside AI clients can call, and each call starts a new instance.

Sourcedocs.camunda.io/docs/components/connectors/out-of-the-box-connectors/available-connectors-overviewread 2026-10-12

Agents query record data with the Query Records tool, which can join across related record types, over synced record types: Appian's data fabric keeps a synchronized copy of the customer's source data that stays queryable, so new knowledge is available without retraining.

Agents also connect to enterprise data and documents, using document resources to retrieve context and AI DocCenter models for trained extraction, and the same SQL API infrastructure backs the data fabric tools on the Appian MCP Server. Unsynced and legacy record types, smart search and record security expressions are not available to the Query Records tool.

Sourcedocs.appian.com 26.8 ai-agent-reference, create-and-configure-ai-agent; readread 2026-09-17

Agents can draw on company knowledge kept in a vector store. The Vector Database connector splits documents into chunks, embeds them with Amazon Bedrock Titan, Azure OpenAI, Google Vertex AI or OpenAI models, and writes them to Amazon OpenSearch, OpenSearch, Elasticsearch, Azure AI Search or Azure Cosmos DB. At run time it retrieves the most relevant passages by semantic similarity, so an agent can look up policies and procedures, search a large product catalog, reuse answers to past support questions or quote the exact policy behind a decision. The store grows as new documents are embedded.

Sourcedocs.camunda.io/docs/components/agentic-orchestration/rag-ai-agentsread 2026-10-12

Conversation persistence "stores chat interactions between an end user and a chat agent across sessions". Each conversation is stored as a session tied to a specific user and chat agent; users can return to and switch between earlier sessions without losing context, and the agent keeps prior messages within a session.

Each session belongs to a single user (other users cannot view, access or share another user's sessions), conversations are retained for audit purposes and end users cannot clear them, and sessions are tied to the environment where they were created. Users cannot delete their history, process agents carry no memory beyond the process, and learning agents have been announced but not shipped. Process and case state belongs to the application's own records.

Sourcedocs.appian.com 26.8 conversation-persistence-chat-agents; readread 2026-09-17

Each agent keeps a short term conversation inside its process instance, passing the last 20 messages to the model by default while always keeping the system prompt. Those messages can live in a process variable, in Camunda document storage with a time to live set to the process's expected lifetime, in Amazon Bedrock AgentCore Memory, or in a custom store on self managed setups. Recall from one process instance to the next relies on an outside service such as AgentCore Memory, which AWS runs.

Sourcedocs.camunda.io/docs/components/connectors/out-of-the-box-connectors/agentic-ai-aiagent-subprocessread 2026-10-12

Oversight comes from the process around each agent. Process agents run as one step inside a process model, and the documentation tells designers to keep conditional flows and validations in the process and let the agent handle only the reasoning or decision step. The human task, review and escalation nodes of the surrounding process then decide what happens to the agent's output, so approvals can be placed at the node level around any agent step.

The AI Agents overview names "combining AI reasoning with optional human oversight" as a core use, agents fail to start when the person starting them lacks access to a required tool, and Appian's AI agents page describes actions and escalations to people managed in the same workflow. No approval tool specific to agents or per action approval setting is described, so approvals sit in the process design around the agent.

Sourcedocs.appian.com 26.8 agent-studio, create-and-configure-ai-agent; readread 2026-09-17

People sit in the process as user tasks, and Tasklist shows each assigned task whether the step before it ran automatically or an AI agent produced the result a person now acts on. A user task can be one of an agent's tools, so the agent asks for help or sign off inside its loop. For MCP tools, a gateway in front of the client lets listing and read operations through and routes selected calls to a user task for confirmation before they run. Task work can also reach people in Microsoft Teams, and teams can build their own task apps on the API.

Sourcedocs.camunda.io/docs/components/connectors/out-of-the-box-connectors/agentic-ai-mcp-client-human-in-the-loopread 2026-10-12

Access controls apply to every agent action. "When an AI agent invokes a tool, user access determines whether the process continues or stops", an agent "fails to start if the initiator doesn't have access to required tools", and a sub agent "must respect all access controls defined for that sub-agent". Chat sessions are scoped to the user who started them.

On the FedRAMP Marketplace Appian holds two authorizations, High for Appian Government Cloud and Moderate for its Moderate edition, Appian Government Cloud also holds a DoD Impact Level 5 provisional authorization from DISA, and Appian is SOC 1, SOC 2, SOC 3, ISO 27001 and HITRUST CSF certified. Appian's private AI commitment is never to train on customer data.

Sourcedocs.appian.com 26.8 ai-agent-reference; appian.com/support/resources/trust/compliance; read 2026-09-17; FedRAMP Marketplace and appian.com/industries/public-sector/appian-government-cloud, readread 2026-10-12

Camunda runs two identity layers, the Orchestration Cluster admin for runtime work and the management identity for Hub and Optimize, and both connect to an outside OpenID Connect provider such as Entra ID for single sign on, with LDAP and Active Directory on the licensed editions. Authorizations, roles, groups, mapping rules and tenants decide who can start, see and change processes, tasks and other resources. The company holds SOC 2 Type II, ISO 27001 and TISAX and is listed with the Cloud Security Alliance on its trust center.

Sourcetrust.camunda.comread 2026-10-12

Each agent run can be inspected afterward, step by step. The run history "uses a timeline layout that displays each step chronologically, including tool calls, AI agent reasoning, and outputs".

A Monitor view logs every agent request with run ID, agent version, status, request, AI actions consumed, tool call count, initiator, duration and start time, per agent and across agents, and active runs can be stopped. System level audit logs for AI agent executions track all AI interactions and can optionally capture prompts and responses, separate from the runtime traces. Feedback and LLM grading are part of testing, and Process HQ's process mining covers processes.

Sourcedocs.appian.com 26.8 monitor-ai-agents, ai-agent-reference; readread 2026-09-17

Operate shows each agent instance's state, such as thinking or calling a tool, its usage metrics and its conversation history grouped by loop iteration, including the tools it picked and the results it got back, for native and connected external agents alike. Health indicators and limit warnings flag an agent that keeps calling the same tool or nears its configured limits. Every process and decision instance can be inspected step by step, audit log APIs record user and user task actions, and Optimize builds reports, dashboards and KPI alerts.

Sourcedocs.camunda.io/docs/components/agentic-orchestration/evaluate-agents/monitor-ai-agentsread 2026-10-12

Appian runs in several named environments. Appian Cloud is managed SaaS across multiple cloud regions. Appian Government Cloud runs in an AWS GovCloud virtual private cloud with full commercial feature parity, including AI. Customers can also self manage on Kubernetes (with RPA included from release 24.2), deploy hybrid on AWS, Azure or Google, or run on premises. Agent Studio is included in the Advanced and Premium tiers, and whether agents are available on self managed installs is not stated. Appian Cloud has been reported to run in 24 regions.

Sourcedocs.appian.com 26.5 appian-government-cloud-overview; appian.com/products/pricing; read 7 August 2026, datedread 2026-09-17

SaaS customers choose the region for each cluster. On Google Cloud the choices include Belgium, London, Iowa, South Carolina, Montréal, Toronto, Singapore and Sydney, and on AWS they include Frankfurt, Paris, Ohio, Oregon, Virginia and Singapore, each paired with a secondary backup region. Enterprise customers can ask for custom regions. Self managed Camunda runs the whole stack on the customer's own infrastructure, and agent model calls can route through a private endpoint inside the customer's network.

Sourcedocs.camunda.io/docs/components/saas/regionsread 2026-10-12

AI skills come as packaged automation for defined work (classification, extraction and summarization), alongside packaged industry solutions and an application template library, and the 26.8 release adds AI DocCenter extraction and classification models that agents call through a process model. Each can be adopted on its own. No catalog of prebuilt AI agents is offered beyond those skills and solutions. Appian also promises a first application in eight weeks, delivered as a service.

Sourceappian.com/products/platform/artificial-intelligence (read 7 August 2026); docs.appian.com 26.8 create-and-configure-ai-agent; datedread 2026-09-17

Camunda Marketplace carries model blueprints and connectors from Camunda, partners and the community, installed straight into a Hub project or downloaded for the desktop modeler. AI agent blueprints such as AI Agent Chat Quick Start and AI Agent Chat With Tools run out of the box on the Camunda provided LLM in SaaS. Partner and community connectors sit outside Camunda's commercial support, and administrators of a self managed Hub can switch the Marketplace off.

Sourcedocs.camunda.io/docs/components/hub/workspace/modeler/modeling/camunda-marketplaceread 2026-10-12

Designers choose the model for each agent. "AI agents use the model configured in the Model Selection section of the Advanced tab." By default new agents use Auto, "which uses the latest model from the configured provider with automatic fallback", and designers "can also select a specific model or route execution through your own provider account". Which models and providers are on that list is not stated.

Sourcedocs.appian.com 26.8 create-and-configure-ai-agent; readread 2026-09-17

The builder picks the provider and the backend for each agent. Anthropic Claude models run directly, through AWS Bedrock or through Microsoft Foundry, OpenAI models run directly, through Foundry or through any OpenAI compatible endpoint, and Google Gemini runs directly or through Google's Enterprise Agent Platform, with custom endpoints for private network setups and a custom provider on self managed installs. SaaS adds a Camunda provided LLM with a free budget for testing, so teams can switch between models such as Claude Sonnet 4.6 before moving to their own account.

Sourcedocs.camunda.io/docs/components/connectors/out-of-the-box-connectors/agentic-ai-aiagent-model-providersread 2026-10-12

Outside agents and systems can call Appian, and Appian agents can call out. The Appian MCP Server's data fabric tools share the agents' SQL API infrastructure, so outside agents can reach Appian. Organizations can administer their own Appian MCP servers so third party agents reach Appian business rules, processes and data, alongside Appian's REST and SOAP web APIs. Appian's own agents also consume external MCP tools, so MCP runs in both directions. Appian also describes Composer, an MCP representation of the entire application estate.

Sourcedocs.appian.com 26.8 ai-agent-reference; Appian World 2026 release (read 7 August 2026); datedread 2026-09-17

The Orchestration Cluster REST API drives processes, instances, user tasks, decisions, identities and audit logs, with SDKs for Java, TypeScript, Python, C# and PHP and the c8ctl command line tool. The Orchestration Cluster MCP Server opens operational work such as incidents, user tasks and process instances to AI clients under the same authentication and authorization model as the REST API. Older Operate and Tasklist REST APIs remain available for existing integrations.

Sourcedocs.camunda.io/docs/apis-tools/orchestration-cluster-api-mcp/orchestration-cluster-api-mcp-overviewread 2026-10-12

Evaluation covers agents before and after deployment. Agents are tested in a test pane (process agents) or a chat test experience before deployment, and the AI Agent Reference lists three feedback sources for both agent types: developer feedback at design time in the test pane, an automated LLM grader, and dev and ops feedback in production, with chat agents appearing in the same quality metrics as process agents. The AI Agents overview lists "test and monitor other AI agents to ensure reliable performance" as a supported use, and agents are versioned.

No test suites built on fixed datasets are documented.

Sourcedocs.appian.com 26.8 ai-agent-reference, agent-studio, monitor-ai-agents; readread 2026-09-17

Camunda Process Test runs an agent process as an integration test, keeping the real model in the loop while mocking outside tool calls. Conditional behavior reacts to whichever tools the agent activates instead of one fixed order, and judge assertions send an output and a plain language expectation to a model that rates the match against a set threshold, alongside semantic similarity checks. Tests run through a Spring Boot starter inside the project's own build.

Sourcedocs.camunda.io/docs/components/agentic-orchestration/evaluate-agents/test-ai-agentsread 2026-10-12

Real application interfaces are controlled through Appian RPA. RPA is a licensed part of the platform, sold by bot count per tier and available to self managed Kubernetes customers from release 24.2, and Appian RPA bots operate the interfaces of applications that expose no programmatic route. These are deterministic bots that work application screens. The agents' tools, which are process models, expression rules, other agents, system tools and external MCP tools, include no computer use tool, so an AI agent does not choose and drive a screen session itself.

Sourceappian.com/products/pricing, /industries/public-sector/appian-government-cloud (read 7 August 2026); docs.appian.com 26.8 create-and-configure-ai-agent; datedread 2026-09-17

Camunda RPA runs Robot Framework scripts on a standalone RPA worker to operate legacy mainframe applications, desktop applications such as ERP systems and websites that expose no API. Each script runs as a step in a process, so screen work on an old system sits in the same flow as API calls, agent steps and user tasks, and scripts are built and tested before they move to production workers.

Sourcedocs.camunda.io/docs/components/rpa/overviewread 2026-10-12