US BUYER’S GUIDE · 6 PRODUCTS

Workflow automation

Which automation platform can run our real workflows reliably at an understandable cost?

Connect apps, orchestrate processes, operate automations, and govern AI actions.

Research desk · Oct 20, 2026 · Methodology

YOUR SHORTLIST STARTS HERE

Compare the options.

Sort by

Highest documented score first. Fit adds up the 28 feature checks below, including AI, across documented plans. Entry price does not include every capability. Unknown evidence appears as a range.

Product / good fit for	Pricing context	Documented fit	Next step
Microsoft Power Automate Organizations building workflows across Microsoft services and applications with Power Platform administration.	$15.00user/month · billed annuallyPremium	58–100/100 documented fit 58% evidence coverage	Full research →
Make Operations teams that want visible branching and transformations across cloud applications.	Usage basedcreditsFree / paid credit plans	47–100/100 documented fit 48% evidence coverage	Full research →
Zapier Business teams connecting familiar cloud applications without maintaining an integration server.	See pricingtask allowanceFree / Professional / Team / Enterprise	42–100/100 documented fit 42% evidence coverage	Full research →
Workato Teams that want production integration controls and a path from self-service to enterprise orchestration.	$75.00month for 2,500 creditsWorkato Pro	40–100/100 documented fit 40% evidence coverage	Full research →
n8n Technical operations teams building multi-step workflows and AI integrations with deployment flexibility.	See pricingworkflow executionsCloud / self-hosted editions	38–100/100 documented fit 38% evidence coverage	Full research →
Rippling Companies that need workflows and reporting across several employee-related systems.	Custom quoteper employee/month plus base and modulesConfigured platform	22–99/100 documented fit 23% evidence coverage	Full research →

BEYOND A FEATURE CHECKMARK

How deep does the capability go?

Choose up to 3 for a closer look →

28 weighted checks: core capabilities contribute 32 points and workflow depth contributes 68. AI is included at the weight shown for this category. Expand Evidence for scope and sources, or ask for the same demonstration from each supplier.

Show checksShowing fewer rows does not change the score.

Workflow automation features for all 6 products
Feature / base weight	Microsoft Power Automate 58–100/100 documented fit Core capabilities32 / 32 pts10/10 checks with established status Workflow depth25.6–68 / 68 pts5/18 checks with established status	Make 47–100/100 documented fit Core capabilities24–32 / 32 pts7/10 checks with established status Workflow depth23.5–68 / 68 pts6/18 checks with established status	Zapier 42–100/100 documented fit Core capabilities25.6–32 / 32 pts8/10 checks with established status Workflow depth16.8–68 / 68 pts4/18 checks with established status	Workato 40–100/100 documented fit Core capabilities20.8–32 / 32 pts6/10 checks with established status Workflow depth19.2–68 / 68 pts4/18 checks with established status	n8n 38–100/100 documented fit Core capabilities20.8–32 / 32 pts7/10 checks with established status Workflow depth17–68 / 68 pts5/18 checks with established status	Rippling 22–99/100 documented fit Core capabilities11.6–30.8 / 32 pts4/10 checks with established status Workflow depth10.4–68 / 68 pts2/18 checks with established status
Core capabilityApplication connectors4 points · Use supported connector operations to connect business applications.	Supported4 / 4 pts Evidence Certified connectors are documented. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Standard application connectors are documented. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Supported app connectors are documented. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Recipes connect supported applications. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Plans include application integrations. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Connected actions are limited to supported integrations. Official documentation · Oct 15, 2026Source
Core capabilityCustom APIs or webhooks4 points · Connect a custom HTTP API, webhook or custom connector.	Supported4 / 4 pts Evidence Custom connectors are documented. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Custom API connections are documented. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Webhooks are available on paid plans. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence API-based operations are documented. Official documentation · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed documentation does not establish custom apis or webhooks at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed documentation does not establish custom apis or webhooks at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source
Depth checkConnection identity and access scope5.6 points · Configure or inspect the identity/credentials and scope used by a connected application. Ask for a demonstration Show which account a workflow uses and which data/actions its connection can access. A scenario to run with the supplier; not a completed product test.	Not verified0–5.6 / 5.6 pts Evidence The reviewed evidence does not establish this advanced behavior: connection identity and access scope. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–5.6 / 5.6 pts Evidence The reviewed evidence does not establish this advanced behavior: connection identity and access scope. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–5.6 / 5.6 pts Evidence The reviewed evidence does not establish this advanced behavior: connection identity and access scope. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–5.6 / 5.6 pts Evidence The reviewed evidence does not establish this advanced behavior: connection identity and access scope. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–5.6 / 5.6 pts Evidence The reviewed evidence does not establish this advanced behavior: connection identity and access scope. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Supported5.6 / 5.6 pts Evidence The API action explicitly configures the external method, authentication and payload. Official documentation · Oct 15, 2026Source
Depth checkAuthenticated webhook or API trigger6.4 points · Protect an inbound webhook/API trigger with authentication or documented verification controls. Ask for a demonstration Call a test endpoint without valid credentials and show it being rejected. A scenario to run with the supplier; not a completed product test.	Not verified0–6.4 / 6.4 pts Evidence The reviewed evidence does not establish this advanced behavior: authenticated webhook or api trigger. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–6.4 / 6.4 pts Evidence The reviewed evidence does not establish this advanced behavior: authenticated webhook or api trigger. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–6.4 / 6.4 pts Evidence The reviewed evidence does not establish this advanced behavior: authenticated webhook or api trigger. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–6.4 / 6.4 pts Evidence The reviewed evidence does not establish this advanced behavior: authenticated webhook or api trigger. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–6.4 / 6.4 pts Evidence The reviewed evidence does not establish this advanced behavior: authenticated webhook or api trigger. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–6.4 / 6.4 pts Evidence The reviewed evidence does not establish this advanced behavior: authenticated webhook or api trigger. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source
Core capabilityConditional workflow paths4 points · Branch or filter workflow execution based on conditions.	Supported4 / 4 pts Evidence Condition actions run different steps according to true/false results. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Scenarios support routing. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Multi-step workflows and their paths require Professional or above. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Conditional recipe steps are documented. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Multi-step workflow logic is documented. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Employee-data triggers drive onboarding workflows. Official documentation · Oct 15, 2026Source
Core capabilityData transformation steps4 points · Map, format or transform data between workflow steps.	Supported4 / 4 pts Evidence Data operations compose, select, join, filter arrays and parse JSON. Official documentation · Oct 15, 2026Source	Supported4 / 4 pts Evidence Data transformation modules are documented. Official documentation · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed documentation does not establish data transformation steps at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed documentation does not establish data transformation steps at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed documentation does not establish data transformation steps at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed documentation does not establish data transformation steps at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source
Depth checkIteration with conditional paths5.6 points · Process a list with loops/iteration and branch according to each item’s result. Ask for a demonstration Process three records, route one exception separately and show the per-record results. A scenario to run with the supplier; not a completed product test.	Supported5.6 / 5.6 pts Evidence Conditions provide true/false paths; data operations filter and transform arrays. Official documentation · Oct 15, 2026Source	Not verified0–5.6 / 5.6 pts Evidence The reviewed evidence does not establish this advanced behavior: iteration with conditional paths. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–5.6 / 5.6 pts Evidence The reviewed evidence does not establish this advanced behavior: iteration with conditional paths. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–5.6 / 5.6 pts Evidence The reviewed evidence does not establish this advanced behavior: iteration with conditional paths. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–5.6 / 5.6 pts Evidence The reviewed evidence does not establish this advanced behavior: iteration with conditional paths. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–5.6 / 5.6 pts Evidence The reviewed evidence does not establish this advanced behavior: iteration with conditional paths. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source
Depth checkExplicit error branch or fallback6.4 points · Define what should happen when an intermediate action fails. Ask for a demonstration Fail the second step and show the configured fallback rather than a silent incomplete run. A scenario to run with the supplier; not a completed product test.	Supported6.4 / 6.4 pts Evidence Run-after settings and grouped scopes implement try/catch, error logging and termination. Official documentation · Oct 15, 2026Source	Supported6.4 / 6.4 pts Evidence Error-handler routes can retain failed bundles and decide whether to retry rather than stop the whole scenario. Official documentation · Oct 15, 2026Source	Not verified0–6.4 / 6.4 pts Evidence The reviewed evidence does not establish this advanced behavior: explicit error branch or fallback. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Supported6.4 / 6.4 pts Evidence Recipes support error handling steps and try/catch-style recovery paths. Official documentation · Oct 15, 2026Source	Not verified0–6.4 / 6.4 pts Evidence The reviewed evidence does not establish this advanced behavior: explicit error branch or fallback. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–6.4 / 6.4 pts Evidence The reviewed evidence does not establish this advanced behavior: explicit error branch or fallback. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source
Core capabilityExecution history and monitoring3.2 points · Inspect workflow runs, job history or execution status.	Supported3.2 / 3.2 pts Evidence Flow run history and error diagnosis are documented. Official documentation · Oct 15, 2026Source	Supported3.2 / 3.2 pts Evidence Scenario execution history and usage monitoring are documented. Official documentation · Oct 15, 2026Source	Supported3.2 / 3.2 pts Evidence Run history and held runs are documented. Official documentation · Oct 15, 2026Source	Supported3.2 / 3.2 pts Evidence Job monitoring and debugging are documented. Official documentation · Oct 15, 2026Source	Supported3.2 / 3.2 pts Evidence Execution history and insights are plan-dependent. Official documentation · Oct 15, 2026Source	Not verified0–3.2 / 3.2 pts Evidence The reviewed documentation does not establish execution history and monitoring at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source
Core capabilityFailure handling or replay3.2 points · Handle errors or retry/replay a failed execution.	Supported3.2 / 3.2 pts Evidence Retry policies are documented. Official documentation · Oct 15, 2026Source	Not verified0–3.2 / 3.2 pts Evidence The reviewed documentation does not establish failure handling or replay at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source	Supported3.2 / 3.2 pts Evidence Held runs can be replayed. Official documentation · Oct 15, 2026Source	Supported3.2 / 3.2 pts Evidence Error-handling blocks and alternative recovery paths are documented. Official documentation · Oct 15, 2026Source	Not verified0–3.2 / 3.2 pts Evidence The reviewed documentation does not establish failure handling or replay at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source	Not verified0–3.2 / 3.2 pts Evidence The reviewed documentation does not establish failure handling or replay at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source
Depth checkPer-step input/output diagnostics4.8 points · Inspect individual step results and errors in a workflow execution. Ask for a demonstration Open a failed run and show which step failed and the input/output relevant to repair. A scenario to run with the supplier; not a completed product test.	Supported4.8 / 4.8 pts Evidence Run history identifies a failed step and its detailed error message. Official documentation · Oct 15, 2026Source	Supported4.8 / 4.8 pts Evidence Incomplete executions retain the error, mappings and data for diagnosis. Official documentation · Oct 15, 2026Source	Supported4.8 / 4.8 pts Evidence Held-run diagnostics identify the failed connection, policy or data condition for an individual run. Official documentation · Oct 15, 2026Source	Supported4.8 / 4.8 pts Evidence Job history shows each executed step and the data passed through it. Official documentation · Oct 15, 2026Source	Not verified0–4.8 / 4.8 pts Evidence The reviewed evidence does not establish this advanced behavior: per-step input/output diagnostics. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Supported4.8 / 4.8 pts Evidence The developer workflow guide directs users to execution history to inspect the API action and returned response. Official documentation · Oct 15, 2026Source
Depth checkControlled retry or failed-run replay4.8 points · Configure or initiate retries/replay with visible execution boundaries. Ask for a demonstration Retry the failed work and explain which completed actions will or will not run again. A scenario to run with the supplier; not a completed product test.	Supported4.8 / 4.8 pts Evidence Failed runs can retry with the same input; individual action policies support retries with backoff. Official documentation · Oct 15, 2026Source	Supported4.8 / 4.8 pts Evidence Saved incomplete executions can be retried manually or automatically; retries use the original configuration. Official documentation · Oct 15, 2026Source	Supported4.8 / 4.8 pts Evidence A held run can be replayed after fixing its disconnected account, permission or capacity problem. Official documentation · Oct 15, 2026Source	Supported4.8 / 4.8 pts Evidence A failed job can rerun with retained trigger data using the latest recipe; source changes require a new trigger. Official documentation · Oct 15, 2026Source	Supported4.8 / 4.8 pts Evidence Failed runs can retry with the saved workflow and previous execution data. Official documentation · Oct 15, 2026Source	Not verified0–4.8 / 4.8 pts Evidence The reviewed evidence does not establish this advanced behavior: controlled retry or failed-run replay. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source
Core capabilityWorkflow or team permissions2.4 points · Limit access to workflows, teams or shared connections.	Supported2.4 / 2.4 pts Evidence Cloud flows distinguish owners from run-only users; sharing requires an eligible license. Official documentation · Oct 15, 2026Source	Supported2.4 / 2.4 pts Evidence Team-role management requires an eligible subscription. Official documentation · Oct 15, 2026Source	Supported2.4 / 2.4 pts Evidence Shared connections and team controls vary by tier. Official documentation · Oct 15, 2026Source	Not verified0–2.4 / 2.4 pts Evidence The reviewed documentation does not establish workflow or team permissions at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source	Supported2.4 / 2.4 pts Evidence SSO and administrative controls depend on edition. Official documentation · Oct 15, 2026Source	Supported2.4 / 2.4 pts Evidence Existing platform permissions govern workflow/AI actions. Official documentation · Oct 15, 2026Source
Core capabilityDeployment or connection policy controls2.4 points · Use environments, managed secrets, connector policies or comparable deployment controls.	Supported2.4 / 2.4 pts Evidence Data policies restrict connector use and can affect running flows. Official documentation · Oct 15, 2026Source	Not verified0–2.4 / 2.4 pts Evidence The reviewed documentation does not establish deployment or connection policy controls at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source	Not verified0–2.4 / 2.4 pts Evidence The reviewed documentation does not establish deployment or connection policy controls at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source	Not verified0–2.4 / 2.4 pts Evidence The reviewed documentation does not establish deployment or connection policy controls at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source	Supported2.4 / 2.4 pts Evidence Environments and external secrets depend on edition. Official documentation · Oct 15, 2026Source	Not verified0–2.4 / 2.4 pts Evidence The reviewed documentation does not establish deployment or connection policy controls at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source
Depth checkWorkflow version or change history3.2 points · Inspect or restore earlier workflow definitions or a change history. Ask for a demonstration Change a workflow, then show the previous version and restoration/audit options. A scenario to run with the supplier; not a completed product test.	Not verified0–3.2 / 3.2 pts Evidence The reviewed evidence does not establish this advanced behavior: workflow version or change history. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–3.2 / 3.2 pts Evidence The reviewed evidence does not establish this advanced behavior: workflow version or change history. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Supported3.2 / 3.2 pts Evidence Published versions and drafts are tracked separately; configured publishing restrictions require approval. Official documentation · Oct 15, 2026Source	Supported3.2 / 3.2 pts Evidence Users with the appropriate recipe edit role can restore previous recipe versions. Official documentation · Oct 15, 2026Source	Supported3.2 / 3.2 pts Evidence Source-control environments preserve workflow changes in a private repository. Official documentation · Oct 15, 2026Source	Not verified0–3.2 / 3.2 pts Evidence The reviewed evidence does not establish this advanced behavior: workflow version or change history. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source
Depth checkRestricted shared credentials or secrets4 points · Control which collaborators can use or administer shared credentials/secrets. Ask for a demonstration Give a colleague workflow access while showing the boundary around connection credentials. A scenario to run with the supplier; not a completed product test.	Supported4 / 4 pts Evidence Run-only users and owners have different rights; an owner cannot change another owner’s connection credentials. Official documentation · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed evidence does not establish this advanced behavior: restricted shared credentials or secrets. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed evidence does not establish this advanced behavior: restricted shared credentials or secrets. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed evidence does not establish this advanced behavior: restricted shared credentials or secrets. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed evidence does not establish this advanced behavior: restricted shared credentials or secrets. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed evidence does not establish this advanced behavior: restricted shared credentials or secrets. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source
Core capabilityAI workflow-building assistance2.4 points · Use natural language or AI assistance to construct or change a workflow.	Supported2.4 / 2.4 pts Evidence Copilot creates and refines supported cloud flows within environment settings. Official documentation · Oct 15, 2026Source	Not verified0–2.4 / 2.4 pts Evidence The reviewed documentation does not establish ai workflow-building assistance at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source	Supported2.4 / 2.4 pts Evidence Copilot assists workflow creation. Official documentation · Oct 15, 2026Source	Not verified0–2.4 / 2.4 pts Evidence The reviewed documentation does not establish ai workflow-building assistance at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source	Supported2.4 / 2.4 pts Evidence Workflow Assistant use has hosted-plan allowances. Official documentation · Oct 15, 2026Source	Not verified0–2.4 / 2.4 pts Evidence The reviewed documentation does not establish ai workflow-building assistance at this feature level. This is an evidence gap, not a finding that the capability is absent. Not verified · Oct 15, 2026Source
Core capabilityAI or agent steps at runtime2.4 points · Run a model, agent or tool-enabled AI step inside an operational workflow.	Supported2.4 / 2.4 pts Evidence AI Builder prompt actions generate text inside a flow; region, capacity and licensing conditions apply. Official documentation · Oct 15, 2026Source	Supported2.4 / 2.4 pts Evidence AI Agent knowledge and tool calls consume operation and sometimes token credits. Official documentation · Oct 15, 2026Source	Supported2.4 / 2.4 pts Evidence AI model/tool steps use model-tier task multipliers and additional tool consumption. Official documentation · Oct 15, 2026Source	Supported2.4 / 2.4 pts Evidence AI and agentic orchestration consume shared platform credits. Official documentation · Oct 15, 2026Source	Supported2.4 / 2.4 pts Evidence AI nodes and agents run with provider-specific model charges and limits. Official documentation · Oct 15, 2026Source	Limited1.2 / 2.4 pts Evidence AI-proposed actions require confirmation; scheduled AI automation is waitlisted. Official documentation · Oct 15, 2026Source
Depth checkAI step uses configured business context3.2 points · Provide an AI step with explicit workflow data, knowledge or connected business records. Ask for a demonstration Show the exact source records passed into an AI step for a sample execution. A scenario to run with the supplier; not a completed product test.	Not verified0–3.2 / 3.2 pts Evidence The reviewed evidence does not establish this advanced behavior: ai step uses configured business context. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–3.2 / 3.2 pts Evidence The reviewed evidence does not establish this advanced behavior: ai step uses configured business context. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–3.2 / 3.2 pts Evidence The reviewed evidence does not establish this advanced behavior: ai step uses configured business context. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–3.2 / 3.2 pts Evidence The reviewed evidence does not establish this advanced behavior: ai step uses configured business context. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–3.2 / 3.2 pts Evidence The reviewed evidence does not establish this advanced behavior: ai step uses configured business context. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–3.2 / 3.2 pts Evidence The reviewed evidence does not establish this advanced behavior: ai step uses configured business context. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source
Depth checkHuman approval before AI actions4 points · Insert an approval or review boundary before AI-generated output triggers a consequential action. Ask for a demonstration Let AI propose a record change and show the run waiting for human approval. A scenario to run with the supplier; not a completed product test.	Not verified0–4 / 4 pts Evidence The reviewed evidence does not establish this advanced behavior: human approval before ai actions. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed evidence does not establish this advanced behavior: human approval before ai actions. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Supported4 / 4 pts Evidence Human in the Loop can pause a workflow for approval, with explicit decline and timeout outcomes. Official documentation · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed evidence does not establish this advanced behavior: human approval before ai actions. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source	Supported4 / 4 pts Evidence The Gmail approval operation pauses execution while waiting for a human response. Official documentation · Oct 15, 2026Source	Not verified0–4 / 4 pts Evidence The reviewed evidence does not establish this advanced behavior: human approval before ai actions. Request the demonstration above; absence of public evidence does not establish absence of the feature. Not verified · Oct 15, 2026Source
Depth checkDetect prompt-injection attempts before agent tool use2.5 points · Demonstrate: Detect prompt-injection attempts before agent tool use. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. Ask for a demonstration Demonstrate: Detect prompt-injection attempts before agent tool use. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. A scenario to run with the supplier; not a completed product test.	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source
Depth checkOmit sensitive payloads from stored execution logs2.5 points · Demonstrate: Omit sensitive payloads from stored execution logs. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. Ask for a demonstration Demonstrate: Omit sensitive payloads from stored execution logs. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. A scenario to run with the supplier; not a completed product test.	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Supported2.5 / 2.5 pts Evidence Confidential mode omits payloads from execution logs. Keep data confidential setting; removes stored payloads and restricts troubleshooting options. Global capability evidence; local plan availability and interface language are not inferred. Official documentation · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source
Depth checkPreserve event order across successive executions2.5 points · Demonstrate: Preserve event order across successive executions. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. Ask for a demonstration Demonstrate: Preserve event order across successive executions. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. A scenario to run with the supplier; not a completed product test.	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Supported2.5 / 2.5 pts Evidence Sequential mode preserves execution order. Process data in order; unresolved incomplete runs block following executions. Global capability evidence; local plan availability and interface language are not inferred. Official documentation · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source
Depth checkChoose pause or discard behaviour when failure storage is full2.5 points · Demonstrate: Choose pause or discard behaviour when failure storage is full. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. Ask for a demonstration Demonstrate: Choose pause or discard behaviour when failure storage is full. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. A scenario to run with the supplier; not a completed product test.	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Supported2.5 / 2.5 pts Evidence Overflow settings select pausing or discarding data. Incomplete-execution storage allowance depends on subscription; discarding cannot be undone. Global capability evidence; local plan availability and interface language are not inferred. Official documentation · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source
Depth checkStop a workflow after a configured execution timeout2.5 points · Demonstrate: Stop a workflow after a configured execution timeout. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. Ask for a demonstration Demonstrate: Stop a workflow after a configured execution timeout. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. A scenario to run with the supplier; not a completed product test.	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Supported2.5 / 2.5 pts Evidence Workflows can stop after a configured time limit. Timeout Workflow setting; Cloud plans enforce different maximum timeouts. Global capability evidence; local plan availability and interface language are not inferred. Official documentation · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source
Depth checkSuspend execution and resume from an external callback2.5 points · Demonstrate: Suspend execution and resume from an external callback. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. Ask for a demonstration Demonstrate: Suspend execution and resume from an external callback. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. A scenario to run with the supplier; not a completed product test.	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Supported2.5 / 2.5 pts Evidence Paused executions can resume on an external webhook. Wait node with per-execution webhook URL; authentication and timeout should be configured. Global capability evidence; local plan availability and interface language are not inferred. Official documentation · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source
Depth checkRestrict workflow requests to approved destinations2.5 points · Demonstrate: Restrict workflow requests to approved destinations. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. Ask for a demonstration Demonstrate: Restrict workflow requests to approved destinations. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. A scenario to run with the supplier; not a completed product test.	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source
Depth checkRoll back supported transactional changes on failure2.5 points · Demonstrate: Roll back supported transactional changes on failure. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. Ask for a demonstration Demonstrate: Roll back supported transactional changes on failure. State the plan, limits and behaviour on failure. Use a sample event, a repeated event and a failing destination. A scenario to run with the supplier; not a completed product test.	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source	Not verified0–2.5 / 2.5 pts Evidence This specific check has not been verified. Missing evidence does not mean the capability is absent. Ask for a demonstration in the proposed plan. Not verified · Oct 21, 2026Source

WHAT WE MEASURE

The criteria behind the score.

The established 20 checks contribute 80 points. Eight new operational checks contribute 20 points. AI checks are included in their own group and are excluded when AI priority is set to zero. These assess documented capability scope, not measured performance. Adjust group priorities in the comparison tool.

See the complete calculation →

Connectors & APIs 20%
Actual supported operations matter more than logo counts.
Workflow logic 20%
Branches and transformations determine process coverage.
Monitoring & recovery 16%
Failures need visible, recoverable handling.
Governance & access 12%
Shared credentials and changes require ownership.
AI workflow capabilities 14.5%
AI agents need tool boundaries, controls, and cost visibility.
Operational detail 17.5%
Specific workflow, integration and control checks beyond the established rubric.

THE PRACTICAL BUYER’S GUIDE

How to choose workflow automation.

Automation platforms connect applications and coordinate work, but their operating models differ. A simple lead handoff can suit a business-owned no-code workflow; a process spanning finance, employee access, and private infrastructure may need stronger engineering and governance. Start with a written process, its source of truth, expected volume, and the person who owns failures. Then compare connectors, logic, operations, access controls, and AI. An integration count is only a starting point: the necessary trigger, action, fields, and authentication must be supported. A successful demo run also says little about duplicate events, expired credentials, rate limits, or recovery after an upstream system changes.

Document the workflow and its failure paths

Write the trigger, required inputs, decision rules, actions, and expected outcome before selecting a platform. Add the cases people currently fix manually: missing fields, duplicate submissions, rejected approvals, and unavailable systems. Ask each finalist to run those cases and show the execution history. Inspect what can be replayed and how duplicate side effects are prevented. Assign a business owner who can decide what the workflow should do and an operator who can diagnose failures. Start with a bounded process that can be reversed or reviewed, then expand once your team understands its behavior and costs.

Normalize billing using the same sample workload

Count completed business outcomes and the steps required for each, including loops, retries, polling, AI calls, and file processing. Then apply each vendor's current billing rules: a task, credit, execution, and bot are different units. Include external model charges if you bring your own AI provider, plus hosting and administration for self-managed deployments. Model a normal month and a burst month, and ask whether reaching a limit pauses work or creates overage charges. Record which administrative features require a higher tier. A low subscription price can still be expensive if the workflow processes many records or requires constant maintenance.

Give agents narrow tools and visible approval points

AI can help build a workflow, classify information inside a fixed process, or select actions dynamically as an agent. Evaluate these separately. Give the agent only the data and tools needed for its task, and require review before consequential external actions. Test ambiguous instructions and untrusted content entering the workflow. Check logs for what the model saw, which tool it called, and why the run stopped. Ask how credentials are shared, how development differs from production, and how a departed employee's access is removed without breaking every workflow. An agent is an operating system component that needs ownership, not a substitute for one.

What changes the total cost?

Tasks, credits, executions, bots, concurrency, and overages.
External model tokens, hosting, connectors, and premium capabilities.
Monitoring, maintenance, credential management, and implementation effort.

Where purchases go wrong

Comparing unlike billing units without a representative workflow.
Counting an app logo as proof that the needed action is supported.
Letting a single employee's personal credentials own critical automations.

Bring these questions to the demo

Run duplicate, malformed, and failed inputs and show recovery.
What does this exact workflow cost at normal and peak volume?
How do we separate development, review, and production access?
Which AI actions need approval and what is recorded in the execution log?

Questions buyers ask

Is a workflow execution the same as a task?

No. Vendors measure different units. Reprice the same sample process under each platform's rules instead of comparing raw allowance numbers.

Is self-hosting automatically cheaper?

Not necessarily. Include infrastructure, patching, backups, monitoring, and engineering time alongside any software license.

Should every step use AI?

No. Use deterministic rules where the outcome is known, and reserve AI for work that benefits from interpretation or generation with appropriate review.

Workflow automation

Compare the options.

Microsoft Power Automate

Make

Zapier

Workato

n8n

Rippling