{"id":61538,"date":"2026-05-19T18:14:42","date_gmt":"2026-05-20T01:14:42","guid":{"rendered":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/?p=61538"},"modified":"2026-05-19T18:14:42","modified_gmt":"2026-05-20T01:14:42","slug":"agentic-workflows","status":"publish","type":"post","link":"\/us\/en\/blog\/agentic-workflows\/","title":{"rendered":"What are agentic workflows and how do they differ from automation?"},"content":{"rendered":"<p><em><strong>A practical guide to architecture, governance, use cases, and a phased rollout plan for enterprise AI execution.<\/strong><\/em><\/p>\n<p>AI tools are doing more than they were two years ago. They&#8217;re drafting responses, summarizing calls, and suggesting next steps. And yet, many operations teams are still spending hours on structured, repeatable work that no one has fully handed off to a system. Ticket routing still requires a human to confirm. Customer data updates still happen manually after every call. Multi-step processes still stall when one handoff breaks.<\/p>\n<p>That gap is an architectural problem. AI that assists is valuable. AI that executes is transformational. Agentic workflows help close the gap between AI that assists and AI that takes action.<\/p>\n<p>This guide covers everything you need to move from evaluation to deployment with confidence, including how to sequence a rollout that reduces risk while accelerating results.<\/p>\n<h2 class=\"heading h2\">Key takeaways<\/h2>\n<ul>\n<li>Agentic workflows plan, decide, and act across multi-step tasks, going further than standard automation.<\/li>\n<li>Every core component, from memory to orchestration, requires deliberate configuration and guardrails.<\/li>\n<li>Enterprise deployment without governance controls creates real compliance and operational risk.<\/li>\n<li>Contact centers and IT operations can use agentic workflows to reduce average handle time (AHT), improve ticket resolution, and free teams from repetitive work.<\/li>\n<li>A phased rollout reduces deployment risk and accelerates ROI.<\/li>\n<\/ul>\n<h2 class=\"heading h2\"><a id=\"what\"><\/a>What are agentic workflows?<\/h2>\n<p>Agentic workflows are goal-directed, multi-step processes in which autonomous AI agents perceive context, plan a sequence of actions, use tools to execute them, and iterate toward a defined outcome.<\/p>\n<p>Our <a href=\"https:\/\/www.ringcentral.com\/us\/en\/blog\/ringcentral-agentic-ai-trends-2026\/\">2026 Agentic AI Trends report<\/a> found that 97% of organizations have already deployed some form of AI or automation. The question many are now asking is why structured, multi-step work still requires so much human intervention. The answer usually comes down to what their current tools are actually capable of.<\/p>\n<p>Unlike traditional automation, autonomous AI agents can adapt their actions when intermediate steps fail or produce unexpected results. Most <a href=\"https:\/\/www.ringcentral.com\/us\/en\/blog\/agentic-ai-maximize-every-conversation-with-ai-powered-business-communications\/\">agentic AI systems<\/a> still include human oversight, guardrails, and explicit stop conditions.<\/p>\n<p>That&#8217;s a fundamentally different operating model from what most teams have in place today.<\/p>\n<h3 class=\"heading h3\">How do they differ from automation and AI copilots?<\/h3>\n<p>Agentic workflows have key distinctions across four system types:<\/p>\n<table style=\"width: 100%;border-collapse: collapse;border-style: solid;border-color: #000000\" border=\"1\" cellpadding=\"6\">\n<tbody>\n<tr>\n<td style=\"width: 25%\"><strong>System type<\/strong><\/td>\n<td style=\"width: 25%\"><strong>How it works<\/strong><\/td>\n<td style=\"width: 25%\"><strong>Memory-aware?<\/strong><\/td>\n<td style=\"width: 25%\"><strong>Self-correcting?<\/strong><\/td>\n<\/tr>\n<tr>\n<td style=\"width: 25%\"><strong>Rule-based automation (RPA)<\/strong><\/td>\n<td style=\"width: 25%\">Follows fixed, predefined rules triggered by specific inputs<\/td>\n<td style=\"width: 25%\">No<\/td>\n<td style=\"width: 25%\">No<\/td>\n<\/tr>\n<tr>\n<td style=\"width: 25%\"><strong>AI copilot<\/strong><\/td>\n<td style=\"width: 25%\">Responds to user prompts, suggests actions, waits for human confirmation<\/td>\n<td style=\"width: 25%\">Limited<\/td>\n<td style=\"width: 25%\">No<\/td>\n<\/tr>\n<tr>\n<td style=\"width: 25%\"><strong>Standard LLM pipeline<\/strong><\/td>\n<td style=\"width: 25%\">Processes a prompt and returns an output in a single pass<\/td>\n<td style=\"width: 25%\">No<\/td>\n<td style=\"width: 25%\">No<\/td>\n<\/tr>\n<tr>\n<td style=\"width: 25%\"><strong>Agentic workflow<\/strong><\/td>\n<td style=\"width: 25%\">Plans a multi-step sequence, uses tools, tracks state, and iterates toward a goal<\/td>\n<td style=\"width: 25%\">Yes<\/td>\n<td style=\"width: 25%\">Yes<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Memory awareness and self-correction are where agentic workflows stand out. A robotic process automation (RPA) tool can update a field if the field exists and the input matches the expected format.<\/p>\n<p>An agentic workflow can check whether the field should be updated, pull context from a prior interaction, apply the update, verify the result, and flag an exception if the update doesn&#8217;t resolve as expected.<\/p>\n<p><img decoding=\"async\" class=\"alignnone size-full wp-image-61541\" src=\"\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/rc-air-pro-agentic-ai-agents-autonomous-action.png\" alt=\"Agentic AI agents, like RingCentral AIR Pro, use context to take autonomous action\" width=\"1884\" height=\"1800\" \/><\/p>\n<hr \/>\n<p><strong>A note on terminology:<\/strong> If you&#8217;ve encountered &#8220;GitHub agentic workflows&#8221; in your research, that term refers to GitHub\u2019s repository automation feature built on GitHub Actions.<\/p>\n<p>This guide refers to \u201cagentic workflows\u201d as enterprise AI workflows that coordinate structured, multi-step work across business systems and customer operations, while keeping humans in the loop for decisions that require judgment.<\/p>\n<hr \/>\n<h2 class=\"heading h2\"><a id=\"how\"><\/a>How do agentic workflows function?<\/h2>\n<p>Before you can configure, govern, or scale an agentic workflow, it helps to understand what&#8217;s running under the hood. The runtime architecture acts as a set of interacting components, each of which can become a failure point if it&#8217;s underpowered or unmonitored.<\/p>\n<p>Most agentic systems follow a continuous loop. While specific frameworks use different terminology, the stages typically include: perceive, plan, act, evaluate, and repeat.<\/p>\n<p>The agent takes in context from its environment, decides what action to take next, and executes that action using available tools. It then evaluates the result and repeats the loop until it reaches the goal or hits a defined boundary condition.<\/p>\n<p>Because these agentic workflows rely on multiple steps and tool calls, orchestration, guardrails, observability, and explicit exit paths are core design requirements rather than optional extras.<\/p>\n<h3 class=\"heading h3\">Perception and input layer<\/h3>\n<p>Agentic workflows start with a perception or input component. This is how the agent observes its environment, including:<\/p>\n<ul>\n<li>User messages<\/li>\n<li>API events<\/li>\n<li>Workflow triggers<\/li>\n<li>Tickets<\/li>\n<li>Emails<\/li>\n<li>Sensor-like signals<\/li>\n<\/ul>\n<p>This layer decides what context gets gathered and normalized before the planning layer processes it.<\/p>\n<p>This means the agent can pull caller history, account status, and prior interaction summaries before deciding how to respond. In terms of IT operations, it means the agent ingests ticket metadata, system health signals, and user context simultaneously rather than processing each input in isolation.<\/p>\n<p>The quality of what gets perceived directly determines the quality of what gets planned, which is why teams that integrate their perception layer with existing business systems see more reliable workflow outcomes than teams that treat perception as a standalone input handler.<\/p>\n<h3 class=\"heading h3\">The planning and reasoning layer<\/h3>\n<p>The planning and reasoning layer is where the agent decides what to do next by decomposing its goal into sub\u2011tasks, sequencing those steps, and selecting the right tools or actions, often using an LLM as the control logic.<\/p>\n<p>The quality of this layer depends on the following:<\/p>\n<ul>\n<li>How well the agent&#8217;s goal is defined<\/li>\n<li>How much relevant context it has access to<\/li>\n<li>How tightly its action space is scoped<\/li>\n<\/ul>\n<p>An agent with a vague goal, limited context, and an unrestricted action space will produce unpredictable results. In contrast, an agent with a specific goal, rich context from integrated business systems, and a well-defined set of permitted actions will execute reliably.<\/p>\n<p>This is why purpose-built agentic AI solutions for specific domains, like call handling or IT ticket management, outperform general-purpose configurations. The planning layer is pre-optimized for the task domain rather than requiring teams to engineer that precision from scratch.<\/p>\n<h3 class=\"heading h3\">Execution and action layer<\/h3>\n<p>Beyond tools, many architectures treat execution as a distinct layer that manages how actions are queued, retried, timed, and logged. This is where error handling, idempotency, and fail-safe fallbacks typically live. It\u2019s also the layer that prevents a single failed API call from breaking an entire workflow.<\/p>\n<hr \/>\n<p><strong>Terminology:<\/strong> Idempotency means that performing the same action multiple times produces the same result as performing it once.<\/p>\n<hr \/>\n<p>When an agent attempts to update a CRM record and the system times out, the execution layer determines whether the workflow retries immediately, waits and retries with backoff, or escalates to a human.<\/p>\n<p>When a ticket gets closed due to overlapping or conflicting events, idempotency controls prevent the duplicate action from creating data inconsistencies.<\/p>\n<p>Without a properly configured execution layer, your agentic workflows will fail unpredictably under production load, exactly when you need them most.<\/p>\n<h3 class=\"heading h3\">Tools, memory, and orchestration<\/h3>\n<p><strong>Tools<\/strong> are the mechanisms through which an agent takes action in the world. They include API calls to external systems, database queries, message-sending functions, form submissions, and any other interface that lets the agent read from or write to a business system.<\/p>\n<p>The tool set available to an agent defines what it can actually do, and scoping that tool set is one of the most important governance decisions in the deployment process.<\/p>\n<p><strong>Memory<\/strong> gives agents the ability to maintain context across steps within a single workflow run (short-term memory) and, in more advanced configurations, across multiple sessions or interactions (long-term memory).<\/p>\n<p>Without short-term memory, an agent can&#8217;t track what it&#8217;s already done within a workflow. Without long-term memory, it can&#8217;t personalize interactions based on prior history.<\/p>\n<p><img decoding=\"async\" class=\"alignnone size-full wp-image-61542\" src=\"\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/how-air-works-desktop-jpg-rendition-scaled.png\" alt=\"RingCentral AIR Pro uses orchestration to manage customer interaction workflows\" width=\"2048\" height=\"731\" \/><\/p>\n<p><strong>Orchestration<\/strong> is the coordination layer that manages how multiple agents or workflow components interact. In a single-agent workflow, this is relatively straightforward: one agent runs one loop. In multi-agent architectures, orchestration manages task delegation, inter-agent communication, conflict resolution, and state synchronization.<\/p>\n<p>Getting orchestration right is what separates a workflow that scales from one that breaks under load.<\/p>\n<p>In practice, memory and orchestration are often implemented on shared infrastructure, such as a centralized state store or event\u2011driven coordination layer, that both agents and workflows depend on for consistency and observability.<\/p>\n<h3 class=\"heading h3\">Feedback, reflection, and evaluation<\/h3>\n<p>Modern agentic systems include a feedback and reflection layer where agents:<\/p>\n<ul>\n<li>Assess whether actions achieved the intended outcome<\/li>\n<li>Compare results against goals or metrics<\/li>\n<li>Adjust plans or prompts based on what they&#8217;ve learned<\/li>\n<\/ul>\n<p>Some frameworks call this &#8220;refinement&#8221; or &#8220;evaluation&#8221; and treat it as a separate stage in the loop. This layer is what allows agents to improve their performance over time rather than repeating the same mistakes.<\/p>\n<p>This means your AI doesn&#8217;t just execute the same call handling pattern indefinitely. It learns which routing decisions led to faster resolutions, which response templates reduced escalations, and which workflows consistently hit exceptions, then adjusts its approach accordingly.<\/p>\n<p>It also means your ticket resolution workflows get more accurate at classifying requests and more efficient at selecting the right remediation steps as they process more volume. The feedback layer turns your agentic workflows into systems that compound their value over time rather than delivering static performance.<\/p>\n<h3 class=\"heading h3\">Guardrails and governance<\/h3>\n<p>Guardrails form a core <a href=\"https:\/\/assets.ringcentral.com\/us\/whitepaper\/responsible-ai-governance-trust.pdf\" target=\"_blank\" rel=\"noopener\">AI governance<\/a> and policy layer that defines:<\/p>\n<ul>\n<li>Permission boundaries (who can do what, which systems can be modified)<\/li>\n<li>Approval thresholds (when human review is required)<\/li>\n<li>Ethical or compliance constraints<\/li>\n<\/ul>\n<p>This layer is typically surfaced as a separate governance or policy component tied to orchestration and logging. It&#8217;s what prevents agents from taking actions outside their intended scope, even when they&#8217;re technically capable of doing so.<\/p>\n<p>Without properly configured guardrails, you&#8217;re deploying a system that can execute actions across your CRM, ticketing platform, and customer communication channels with no enforceable boundaries.<\/p>\n<p>That&#8217;s a compliance exposure that surfaces the first time an agent updates the wrong record, sends a message to the wrong customer segment, or closes a ticket that requires escalation. The guardrail layer is where you define what \u201csafe to automate\u201d actually means in your environment, and it&#8217;s non-negotiable for any workflow that touches systems of record or customer-facing operations.<\/p>\n<h3 class=\"heading h3\">Integration and connectivity infrastructure<\/h3>\n<p>Agentic workflows are rarely closed-box systems. They sit on top of integration and connectivity infrastructure that connects them to critical systems within your organization: your CRM, messaging platforms, data warehouses, and any other business system the workflow needs to read from or write to. The quality of your integration layer directly affects workflow reliability.<\/p>\n<p>If your integration layer can&#8217;t handle the read\/write volume your workflows generate, or if it introduces latency that breaks time-sensitive processes, your agentic workflows will fail under production load regardless of how well the planning layer is configured.<\/p>\n<p>A workflow that can reason perfectly but can&#8217;t reliably pull customer context from your CRM or update ticket status in your ITSM platform delivers no operational value. This is why teams evaluating agentic platforms need to assess integration architecture with the same rigor they apply to model performance.<\/p>\n<h3 class=\"heading h3\">Human in the loop and collaboration<\/h3>\n<p>Agentic workflows aren&#8217;t fully autonomous. They usually include human-in-the-loop (HITL) components for:<\/p>\n<ul>\n<li>Approvals<\/li>\n<li>Exception handling<\/li>\n<li>Calibration and correction of the agent&#8217;s behavior<\/li>\n<\/ul>\n<p>These components are often treated as a separate collaboration layer, not just an add-on within the planning module. The HITL layer is where you define which decisions require human judgment and which can be safely delegated to the agent.<\/p>\n<p>This matters because it determines where your team spends its time. Instead of handling every routine ticket closure or CRM update manually, your operators focus on the exceptions that actually require expertise: the ambiguous customer situations, the edge cases your workflows haven&#8217;t seen before, the judgment calls that carry reputational or compliance weight.<\/p>\n<p>The human-in-the-loop layer helps teams shift skilled employees away from repetitive execution and toward higher-value decisions.<\/p>\n<h3 class=\"heading h3\">Learning modules<\/h3>\n<p>In more advanced setups, a learning module (sometimes folded into memory) tracks patterns over time so your agents can generalize or optimize workflows. This can include:<\/p>\n<ul>\n<li>Procedural memory (how to execute certain classes of tasks)<\/li>\n<li>Semantic memory (facts, preferences, and domain knowledge)<\/li>\n<li>Reinforcement-style feedback that tunes prompts or tool choices<\/li>\n<\/ul>\n<p>This capability helps workflows improve as they process more volume. Your ticket classification accuracy increases as the system learns which request types map to which resolution paths. Your call routing decisions get sharper as the system identifies which patterns correlate with faster resolutions.<\/p>\n<p>Instead of requiring manual retraining or configuration updates every time your operational environment shifts, the learning module adapts the workflow&#8217;s behavior based on observed outcomes, reducing the ongoing maintenance burden while compounding the efficiency gains over time.<\/p>\n<h2 class=\"heading h2\"><a id=\"enterprise-governance-safety\"><\/a>Ensuring enterprise governance and safety for agentic workflows<\/h2>\n<p>Though they\u2019re separate workstreams, the capability question and the governance question aren&#8217;t separate conversations.<\/p>\n<p>An agent that can update CRM records, send messages, or close tickets has write access to systems of record. It also needs explicit permission boundaries, audit trails, and defined escalation paths, not just capability limits set at the model level.<\/p>\n<p>This could put sensitive data at risk through leaks or unauthorized access, yet <a href=\"https:\/\/www.sailpoint.com\/identity-library\/ai-agents-attack-surface\" target=\"_blank\" rel=\"noopener\">SailPoint\u2019s 2025<\/a> report found that less than half of tech leaders have policies in place to manage agentic workflow governance.<\/p>\n<p>The governance challenge with agentic workflows is that the risks are proportional to the autonomy. The more steps an agent can execute without human review, the more consequential an error becomes.<\/p>\n<p>Enterprise deployment without governance controls creates compliance exposure and operational risk that compounds across every workflow the system touches.<\/p>\n<p>Four governance pillars apply across every enterprise agentic deployment:<\/p>\n<ol>\n<li><strong>Permission scoping:<\/strong> Define exactly which systems, data sources, and action types each agent can access. Scope permissions to the minimum required for the workflow to function. Avoid broad access unless there\u2019s a clearly documented, temporary, and monitored reason.<\/li>\n<li><strong>Human approval gates:<\/strong> Identify the steps in each workflow where an irreversible action, a high-stakes decision, or an ambiguous result requires human confirmation before the agent proceeds. Build those gates into the workflow architecture, not as an afterthought.<\/li>\n<li><strong>Audit logging and traceability:<\/strong> Every agent action should produce a log entry that records what the agent did, what context it used, what tool it called, and what the result was. Traceability is the foundation of both compliance reporting and post-incident investigation.<\/li>\n<li><strong>Compliance and data handling:<\/strong> Map your agentic workflows to your existing SOC 2 and ISO 27001 controls. Data the agent accesses, processes, or stores needs to meet the same handling requirements as data accessed by human operators.<\/li>\n<\/ol>\n<p>For teams building toward enterprise deployment, RingCentral&#8217;s agentic AI solutions, including <a href=\"https:\/\/www.ringcentral.com\/products\/air-pro.html\">AI Representative<\/a> (AIR Pro), are designed with these governance requirements in mind. Building voice-first AI agents with AIR Pro gives you access to the platform\u2019s built-in governance, including enterprise-grade permissions, guardrails, observability, and audit-ready controls.<\/p>\n<h2 class=\"heading h2\"><a id=\"key-use-cases\"><\/a>Key agentic workflow use cases for contact centers and IT operations<\/h2>\n<p>The clearest ROI from agentic workflows comes from high-volume, structured processes where the cost of human handling is measurable, and the workflow steps are well-defined. Contact centers and IT operations meet both criteria.<\/p>\n<h3 class=\"heading h3\">Contact center applications<\/h3>\n<h4 class=\"heading h4\">After-call work automation<\/h4>\n<p>After every customer interaction, agents currently spend time writing summaries, updating CRM records, setting follow-up tasks, and closing tickets. An agentic AI agent handles all of that automatically:<\/p>\n<ul>\n<li>Transcribes the interaction<\/li>\n<li>Extracts key details<\/li>\n<li>Updates the relevant records in the CRM<\/li>\n<li>Creates any required follow-up actions, all before the agent&#8217;s next call begins<\/li>\n<\/ul>\n<p>This is one of the clearest opportunities to reduce AHT.<\/p>\n<h4 class=\"heading h4\">Dynamic call routing with intent recognition<\/h4>\n<p>Standard interactive voice response (IVR) systems route based on menu selections. Agentic workflows <a href=\"https:\/\/www.ringcentral.com\/us\/en\/blog\/dynamic-call-routing-the-new-normal-for-telehealth-triage\/\">route dynamically<\/a> based on caller intent, account history, and real-time context pulled from integrated systems.<\/p>\n<p>For example, a caller who&#8217;s had three billing issues in the past 90 days gets routed differently than a first-time caller with the same stated question. The routing decision reflects the full context, not just the current input.<\/p>\n<h4 class=\"heading h4\">Escalation handling and handoff orchestration<\/h4>\n<p>When a customer interaction reaches a complexity threshold that requires human involvement, an agentic workflow manages the handoff:<\/p>\n<ul>\n<li>Prepares a context summary for the receiving agent<\/li>\n<li>Transfers the interaction with full history attached<\/li>\n<li>Updates the case record<\/li>\n<\/ul>\n<p>Now the agent starts the conversation informed rather than asking the customer to repeat themselves.<\/p>\n<h4 class=\"heading h4\">Proactive outreach workflows<\/h4>\n<p>For follow-up communications, appointment confirmations, or renewal reminders, agentic workflows can initiate outbound contact, handle responses, and update records based on outcomes. This <a href=\"https:\/\/www.ringcentral.com\/us\/en\/blog\/ai-outreach\/\">AI outreach<\/a> escalates to a human only when the interaction reaches a decision point that requires it.<\/p>\n<h3 class=\"heading h3\">IT operations applications<\/h3>\n<h4 class=\"heading h4\">Ticket triage and resolution<\/h4>\n<p>The majority of IT help desk volume comes from a predictable set of request types:<\/p>\n<ul>\n<li>Password resets<\/li>\n<li>Access provisioning<\/li>\n<li>Software installation requests<\/li>\n<li>Connectivity troubleshooting<\/li>\n<\/ul>\n<p>An agentic workflow handles intake, classifies the request, executes the resolution steps for known request types, and escalates to a human technician only when the request falls outside the defined scope.<\/p>\n<h4 class=\"heading h4\">Incident detection and response coordination<\/h4>\n<p>When a monitoring system flags an anomaly, an agentic workflow can:<\/p>\n<ul>\n<li>Assess the alert against historical patterns<\/li>\n<li>Pull relevant system context<\/li>\n<li>Notify the appropriate team members<\/li>\n<li>Initiate a predefined response sequence<\/li>\n<\/ul>\n<p>This can happen before a human engineer begins manual investigation. For incidents with well-defined response playbooks, this shortens response time significantly.<\/p>\n<h4 class=\"heading h4\">Access provisioning and deprovisioning<\/h4>\n<p>Onboarding and offboarding workflows require coordinated actions across multiple systems, including identity management, email, application access, and hardware assignment.<\/p>\n<p>An agentic workflow executes those steps in sequence, confirms completion at each stage, and flags exceptions for human review rather than leaving the process to manual coordination across teams.<\/p>\n<h4 class=\"heading h4\">Change management documentation<\/h4>\n<p>After a system change, an agentic workflow can:<\/p>\n<ul>\n<li>Pull the change log<\/li>\n<li>Generate a structured summary<\/li>\n<li>Update the configuration management database (CMDB)<\/li>\n<li>Route the documentation for review<\/li>\n<\/ul>\n<p>This reduces the administrative overhead that often delays post-change documentation.<\/p>\n<h2 class=\"heading h2\"><a id=\"how-to-measure\"><\/a>How to measure agentic workflow ROI<\/h2>\n<p>Metrics without instrumentation are aspirational, not operational. Every KPI you plan to track needs a defined data source, a baseline measurement taken before deployment, and a review cadence established before the workflow goes live.<\/p>\n<p>According to <a href=\"https:\/\/www.mckinsey.com\/capabilities\/quantumblack\/our-insights\/from-promise-to-impact-how-companies-can-measure-and-realize-the-full-value-of-ai\" target=\"_blank\" rel=\"noopener\">McKinsey<\/a>, successful organizations scale beyond AI pilots by doing the following:<\/p>\n<ol>\n<li>Define value up front and link relevant metrics<\/li>\n<li>Build measurement and attribution into rollout<\/li>\n<li>Run AI as a managed investment<\/li>\n<\/ol>\n<h3 class=\"heading h3\">Step 1: Define value and link metrics<\/h3>\n<p>Every agentic workflow deployment should map to a value chain that connects four metric layers:<\/p>\n<ul>\n<li><strong>Technical performance:<\/strong> Model accuracy, workflow completion rate, error rate, and latency. These metrics confirm the system works as designed, but don&#8217;t prove business value on their own.<\/li>\n<li><strong>User adoption:<\/strong> Active usage rate, workflow utilization, and human override frequency. These metrics show whether your team is actually using the system or routing around it.<\/li>\n<li><strong>Operational change:<\/strong> Average handle time (AHT), containment rate, ticket deflection rate, cycle time, and quality assurance (QA) coverage. These metrics demonstrate measurable efficiency gains in the processes the workflow touches.<\/li>\n<li><strong>Financial impact:<\/strong> Cost per resolution, labor cost reduction, capacity freed for redeployment, and total cost of ownership (TCO). These metrics translate operational change into the financial outcomes that justify continued investment.<\/li>\n<\/ul>\n<p>Map each key performance indicator (KPI) to its baseline source, target improvement, and the value chain layer it belongs to.<\/p>\n<p>Establish your baseline for each metric in the two to four weeks before deployment begins. Any metric you can&#8217;t baseline before launch should be removed from your initial KPI set and added in a later phase once instrumentation is in place. The value chain only works if every layer is instrumented and every metric connects to the one above it.<\/p>\n<h3 class=\"heading h3\">Step 2: Build measurement and attribution into rollout design<\/h3>\n<p>Attribution matters because operational environments change constantly. If you deploy an agentic workflow across your entire contact center at once and AHT drops, you won&#8217;t know whether the improvement came from the workflow, from seasonal volume shifts, or from some combination of both.<\/p>\n<p>Two rollout designs solve the attribution problem:<\/p>\n<ul>\n<li><strong>A\/B testing:<\/strong> Deploy the agentic workflow to a randomly selected subset of interactions or users while maintaining the existing process for a control group. Measure performance differences between the two groups over a defined period.<\/li>\n<li><strong>Staggered deployment:<\/strong> Deploy the workflow to one team, region, or process segment first, measure performance, then expand to additional segments in sequence. Compare performance across segments and time periods to isolate the workflow&#8217;s impact.<\/li>\n<\/ul>\n<p>Whichever design you choose, define your measurement window and success criteria before deployment begins. Document your rollout design, measurement approach, and success criteria in a single evidence pack that stakeholders can review.<\/p>\n<h3 class=\"heading h3\">Step 3: Pilot agentic workflows with a fixed review cadence<\/h3>\n<p>Agentic workflows require ongoing performance monitoring, cost tracking, and explicit go\/no-go decisions at defined intervals.<\/p>\n<p>A stage gate structure allows you to deploy workflow pilots as staged investments with clear evidence requirements at each gate. Only use cases that meet defined thresholds advance to the next stage. Use cases that don&#8217;t get paused, re-scoped, or killed.<\/p>\n<p>This staged investment approach gives you the evidence you need to defend continued investment and the discipline to end pilots that don&#8217;t deliver.<\/p>\n<h2 class=\"heading h2\"><a id=\"put-agentic-workflows\"><\/a>Put agentic workflows to work in your enterprise<\/h2>\n<p>Agentic workflows deliver measurable efficiency gains in contact centers and IT operations when you deploy them with the right architecture, governance, and phased rollout approach. The gap between AI that assists and AI that executes closes through deliberate configuration of scoped permissions, defined approval gates, instrumented KPIs, and a sequenced rollout that builds reliability before it builds scale.<\/p>\n<p>The technology creates the capability. The architecture and governance determine whether that capability translates into operational results.<\/p>\n<p>For teams ready to move from evaluation to deployment, RingCentral&#8217;s agentic AI solutions are purpose-built for complex, high-volume workflows. <a href=\"https:\/\/www.ringcentral.com\/products\/air-pro.html\">AIR Pro<\/a> supports agentic workflow execution, advanced call handling, dynamic routing, and customized logic at enterprise scale, with governance controls built into the platform rather than bolted on after the fact. If your call handling environment has outgrown standard AI receptionist tools or rigid IVR configurations, AIR Pro is worth a close look.<\/p>\n<p>Explore AIR Pro or <a href=\"https:\/\/www.ringcentral.com\/products\/air-pro.html#form\">sign up for early access<\/a> to see how agentic workflows can fit your operational environment.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>A practical guide to architecture, governance, use cases, and a phased rollout plan for enterprise AI execution. AI tools are doing more than they were two years ago. They&#8217;re drafting responses, summarizing calls, and suggesting next steps. And yet, many operations teams are still spending hours on structured, repeatable work that no one has fully &#8230;<\/p>\n","protected":false},"author":1252,"featured_media":61539,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[43331],"tags":[7538,12837],"class_list":["post-61538","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-innovation","tag-artificial-intelligence","tag-cloud-industry-insights"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v19.3 (Yoast SEO v27.2) - https:\/\/yoast.com\/product\/yoast-seo-premium-wordpress\/ -->\n<title>Agentic Workflows: 2026 Enterprise Guide<\/title>\n<meta name=\"description\" content=\"Learn how agentic workflows differ from automation, their core architecture, governance requirements, and a phased rollout plan for enterprise AI.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.ringcentral.com\/us\/en\/blog\/agentic-workflows\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"What are agentic workflows and how do they differ from automation?\" \/>\n<meta property=\"og:description\" content=\"Learn how agentic workflows differ from automation, their core architecture, governance requirements, and a phased rollout plan for enterprise AI.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.ringcentral.com\/us\/en\/blog\/agentic-workflows\/\" \/>\n<meta property=\"og:site_name\" content=\"RingCentral Blog\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/ringcentral\" \/>\n<meta property=\"article:published_time\" content=\"2026-05-20T01:14:42+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.ringcentral.com\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/contact-center-agent-using-ringcentral-air-pro-to-set-up-agentic-workflows-scaled.png\" \/>\n\t<meta property=\"og:image:width\" content=\"2048\" \/>\n\t<meta property=\"og:image:height\" content=\"1152\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Wajih Kazmi\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@ringcentral\" \/>\n<meta name=\"twitter:site\" content=\"@ringcentral\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Wajih Kazmi\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"17 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#article\",\"isPartOf\":{\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/\"},\"author\":{\"name\":\"Wajih Kazmi\",\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#\/schema\/person\/d9d0bdb423048421df12e805eec6061d\"},\"headline\":\"What are agentic workflows and how do they differ from automation?\",\"datePublished\":\"2026-05-20T01:14:42+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/\"},\"wordCount\":3674,\"publisher\":{\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#organization\"},\"image\":{\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#primaryimage\"},\"thumbnailUrl\":\"\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/contact-center-agent-using-ringcentral-air-pro-to-set-up-agentic-workflows-scaled.png\",\"keywords\":[\"artificial intelligence\",\"Industry insights\"],\"articleSection\":[\"AI &amp; Innovation\"],\"inLanguage\":\"en-US\"},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/\",\"url\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/\",\"name\":\"Agentic Workflows: 2026 Enterprise Guide\",\"isPartOf\":{\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#primaryimage\"},\"image\":{\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#primaryimage\"},\"thumbnailUrl\":\"\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/contact-center-agent-using-ringcentral-air-pro-to-set-up-agentic-workflows-scaled.png\",\"datePublished\":\"2026-05-20T01:14:42+00:00\",\"description\":\"Learn how agentic workflows differ from automation, their core architecture, governance requirements, and a phased rollout plan for enterprise AI.\",\"breadcrumb\":{\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#primaryimage\",\"url\":\"\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/contact-center-agent-using-ringcentral-air-pro-to-set-up-agentic-workflows-scaled.png\",\"contentUrl\":\"\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/contact-center-agent-using-ringcentral-air-pro-to-set-up-agentic-workflows-scaled.png\",\"width\":2048,\"height\":1152,\"caption\":\"Contact center agent using RingCentral AIR Pro to set up agentic workflows\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"What are agentic workflows and how do they differ from automation?\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#website\",\"url\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/\",\"name\":\"RingCentral Blog\",\"description\":\"Intelligent Communications\",\"publisher\":{\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#organization\",\"name\":\"RingCentral\",\"url\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#\/schema\/logo\/image\/\",\"url\":\"\/us\/en\/blog\/wp-content\/uploads\/2025\/04\/ringcentral-logo.png\",\"contentUrl\":\"\/us\/en\/blog\/wp-content\/uploads\/2025\/04\/ringcentral-logo.png\",\"width\":2048,\"height\":309,\"caption\":\"RingCentral\"},\"image\":{\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#\/schema\/logo\/image\/\"},\"sameAs\":[\"https:\/\/www.facebook.com\/ringcentral\",\"https:\/\/x.com\/ringcentral\",\"https:\/\/www.linkedin.com\/company\/ringcentral\/\",\"https:\/\/www.instagram.com\/ringcentral\"]},{\"@type\":\"Person\",\"@id\":\"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#\/schema\/person\/d9d0bdb423048421df12e805eec6061d\",\"name\":\"Wajih Kazmi\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/secure.gravatar.com\/avatar\/27f503e82446f4501466877ad804067ddebe18375abef4d17eff2f44fad4b6fb?s=96&d=mm&r=g\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/27f503e82446f4501466877ad804067ddebe18375abef4d17eff2f44fad4b6fb?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/27f503e82446f4501466877ad804067ddebe18375abef4d17eff2f44fad4b6fb?s=96&d=mm&r=g\",\"caption\":\"Wajih Kazmi\"},\"description\":\"Wajih Kazmi is a product marketing leader with a background in AI and a focus on building innovative, customer-centric solutions. He currently leads go-to-market efforts for RingCX at RingCentral, bringing deep expertise in customer experience and agentic AI. Wajih blends technical insight with strategic storytelling to drive adoption and impact in the CX space.\",\"sameAs\":[\"https:\/\/www.linkedin.com\/in\/wajihkazmi\/\"],\"url\":\"\/us\/en\/blog\/author\/wajih-kazmi\/\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"Agentic Workflows: 2026 Enterprise Guide","description":"Learn how agentic workflows differ from automation, their core architecture, governance requirements, and a phased rollout plan for enterprise AI.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.ringcentral.com\/us\/en\/blog\/agentic-workflows\/","og_locale":"en_US","og_type":"article","og_title":"What are agentic workflows and how do they differ from automation?","og_description":"Learn how agentic workflows differ from automation, their core architecture, governance requirements, and a phased rollout plan for enterprise AI.","og_url":"https:\/\/www.ringcentral.com\/us\/en\/blog\/agentic-workflows\/","og_site_name":"RingCentral Blog","article_publisher":"https:\/\/www.facebook.com\/ringcentral","article_published_time":"2026-05-20T01:14:42+00:00","og_image":[{"width":2048,"height":1152,"url":"https:\/\/www.ringcentral.com\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/contact-center-agent-using-ringcentral-air-pro-to-set-up-agentic-workflows-scaled.png","type":"image\/png"}],"author":"Wajih Kazmi","twitter_card":"summary_large_image","twitter_creator":"@ringcentral","twitter_site":"@ringcentral","twitter_misc":{"Written by":"Wajih Kazmi","Est. reading time":"17 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#article","isPartOf":{"@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/"},"author":{"name":"Wajih Kazmi","@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#\/schema\/person\/d9d0bdb423048421df12e805eec6061d"},"headline":"What are agentic workflows and how do they differ from automation?","datePublished":"2026-05-20T01:14:42+00:00","mainEntityOfPage":{"@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/"},"wordCount":3674,"publisher":{"@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#organization"},"image":{"@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#primaryimage"},"thumbnailUrl":"\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/contact-center-agent-using-ringcentral-air-pro-to-set-up-agentic-workflows-scaled.png","keywords":["artificial intelligence","Industry insights"],"articleSection":["AI &amp; Innovation"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/","url":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/","name":"Agentic Workflows: 2026 Enterprise Guide","isPartOf":{"@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#website"},"primaryImageOfPage":{"@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#primaryimage"},"image":{"@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#primaryimage"},"thumbnailUrl":"\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/contact-center-agent-using-ringcentral-air-pro-to-set-up-agentic-workflows-scaled.png","datePublished":"2026-05-20T01:14:42+00:00","description":"Learn how agentic workflows differ from automation, their core architecture, governance requirements, and a phased rollout plan for enterprise AI.","breadcrumb":{"@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#primaryimage","url":"\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/contact-center-agent-using-ringcentral-air-pro-to-set-up-agentic-workflows-scaled.png","contentUrl":"\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/contact-center-agent-using-ringcentral-air-pro-to-set-up-agentic-workflows-scaled.png","width":2048,"height":1152,"caption":"Contact center agent using RingCentral AIR Pro to set up agentic workflows"},{"@type":"BreadcrumbList","@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/agentic-workflows\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/"},{"@type":"ListItem","position":2,"name":"What are agentic workflows and how do they differ from automation?"}]},{"@type":"WebSite","@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#website","url":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/","name":"RingCentral Blog","description":"Intelligent Communications","publisher":{"@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#organization","name":"RingCentral","url":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#\/schema\/logo\/image\/","url":"\/us\/en\/blog\/wp-content\/uploads\/2025\/04\/ringcentral-logo.png","contentUrl":"\/us\/en\/blog\/wp-content\/uploads\/2025\/04\/ringcentral-logo.png","width":2048,"height":309,"caption":"RingCentral"},"image":{"@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/ringcentral","https:\/\/x.com\/ringcentral","https:\/\/www.linkedin.com\/company\/ringcentral\/","https:\/\/www.instagram.com\/ringcentral"]},{"@type":"Person","@id":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/#\/schema\/person\/d9d0bdb423048421df12e805eec6061d","name":"Wajih Kazmi","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/secure.gravatar.com\/avatar\/27f503e82446f4501466877ad804067ddebe18375abef4d17eff2f44fad4b6fb?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/27f503e82446f4501466877ad804067ddebe18375abef4d17eff2f44fad4b6fb?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/27f503e82446f4501466877ad804067ddebe18375abef4d17eff2f44fad4b6fb?s=96&d=mm&r=g","caption":"Wajih Kazmi"},"description":"Wajih Kazmi is a product marketing leader with a background in AI and a focus on building innovative, customer-centric solutions. He currently leads go-to-market efforts for RingCX at RingCentral, bringing deep expertise in customer experience and agentic AI. Wajih blends technical insight with strategic storytelling to drive adoption and impact in the CX space.","sameAs":["https:\/\/www.linkedin.com\/in\/wajihkazmi\/"],"url":"\/us\/en\/blog\/author\/wajih-kazmi\/"}]}},"rc_img_url":"\/us\/en\/blog\/wp-content\/uploads\/2026\/05\/contact-center-agent-using-ringcentral-air-pro-to-set-up-agentic-workflows-scaled.png","rcblog_by_author":"<a href=\"\/us\/en\/blog\/author\/wajih-kazmi\/amp\" data-dl-events-click=\"true\" data-dl-element=\"link\"><span class=\"image\"><img src=\"https:\/\/secure.gravatar.com\/avatar\/27f503e82446f4501466877ad804067ddebe18375abef4d17eff2f44fad4b6fb?s=96&d=mm&r=g\" alt=\"\" width=\"30\" height=\"30\" layout=\"fixed\"><\/img><\/span><span class=\"by-author-name\">Wajih Kazmi<\/span><\/a>","rc_author_full_name":"Wajih Kazmi","rc_author_avatar":"\/us\/en\/blog\/wp-content\/uploads\/2025\/05\/A1EC553E-0F67-4D08-8E96-24DFCEE102A9.jpg","rc_author_link":"\/us\/en\/blog\/author\/wajih-kazmi\/amp","rc_post_categories":"<a href=\"\/us\/en\/blog\/category\/trending\/ai-innovation\/amp\">AI &amp; Innovation<\/a>","amp_link":"\/us\/en\/blog\/agentic-workflows\/amp","excerpt_title":"What are agentic workflows and how do they differ from au...","_links":{"self":[{"href":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/wp-json\/wp\/v2\/posts\/61538","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/wp-json\/wp\/v2\/users\/1252"}],"replies":[{"embeddable":true,"href":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/wp-json\/wp\/v2\/comments?post=61538"}],"version-history":[{"count":0,"href":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/wp-json\/wp\/v2\/posts\/61538\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/wp-json\/wp\/v2\/media\/61539"}],"wp:attachment":[{"href":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/wp-json\/wp\/v2\/media?parent=61538"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/wp-json\/wp\/v2\/categories?post=61538"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/newrcblog.wpengine.com\/us\/en\/blog\/wp-json\/wp\/v2\/tags?post=61538"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}