<style> .chartjs-size-monitor { display: none !important; height: 0 !important; overflow: hidden !important; }p { margin: 0; }span.fr-emoticon.fr-emoticon-img { background-repeat: no-repeat !important; font-size: inherit; height: 1em; width: 1em; min-height: 20px; min-width: 20px; display: inline-block; margin: -0.1em 0.1em 0.1em; line-height: 1; vertical-align: middle; } span.fr-emoticon { font-weight: normal; font-family: "Apple Color Emoji", "Segoe UI Emoji", "NotoColorEmoji", "Segoe UI Symbol", "Android Emoji", "EmojiSymbols"; display: inline; line-height: 0; } blockquote { border-left: solid 2px #5e35b1; color: #5e35b1; margin-left:0; padding-left:5px;}blockquote blockquote{ border-color: #00bcd4; color: #00bcd4;}blockquote blockquote blockquote{ border-color: #43a047; color: #43a047;} table.grid{ border-collapse: collapse;} table.grid td, table.grid th { border: 1px solid #ddd;} .fr-fic.fr-dib{ display: block; margin: 5px auto;}.fr-fic.fr-dib.fr-fir{ text-align: right; margin: 5px 0 5px auto;}.fr-fic.fr-dib.fr-fil{ text-align: left; margin: 5px auto 5px 0;}.fr-fic.fr-dii{ float: none; margin: 5px auto;}.fr-fic.fr-dii.fr-fil{ float: left; margin: 5px auto;}.fr-fic.fr-dii.fr-fir{ float: right; margin: 5px auto;}img.fr-dib.fr-fir { margin-right: 0; text-align: right;}img.fr-dib.fr-fil { margin-left: 0; text-align: left;}img.fr-dib { margin: 5px auto; display: block; float: none;}img.fr-bordered { box-sizing: content-box; border: solid 5px #CCC;}img.fr-shadow { box-shadow: 10px 10px 5px 0px #cccccc;}img.fr-rounded { border-radius: 10px; -moz-border-radius: 10px; -webkit-border-radius: 10px; -moz-background-clip: padding; -webkit-background-clip: padding-box; background-clip: padding-box;}</style><style>
p {
margin: 0;
}
span.fr-emoticon.fr-emoticon-img {
background-repeat: no-repeat !important; font-size: inherit; height: 1em; width: 1em; min-height: 20px; min-width: 20px; display: inline-block; margin: -0.1em 0.1em 0.1em; line-height: 1; vertical-align: middle;
}
span.fr-emoticon {
font-weight: normal; font-family: "Apple Color Emoji", "Segoe UI Emoji", "NotoColorEmoji", "Segoe UI Symbol", "Android Emoji", "EmojiSymbols"; display: inline; line-height: 0;
}
blockquote {
border-left: solid 2px #5e35b1; color: #5e35b1; margin-left: 0; padding-left: 5px;
}
blockquote blockquote {
border-color: #00bcd4; color: #00bcd4;
}
blockquote blockquote blockquote {
border-color: #43a047; color: #43a047;
}
table.grid {
border-collapse: collapse;
}
table.grid td,
table.grid th {
border: 1px solid #ddd;
}
.fr-fic.fr-dib {
display: block; margin: 5px auto;
}
.fr-fic.fr-dib.fr-fir {
text-align: right; margin: 5px 0 5px auto;
}
.fr-fic.fr-dib.fr-fil {
text-align: left; margin: 5px auto 5px 0;
}
.fr-fic.fr-dii {
float: none; margin: 5px auto;
}
.fr-fic.fr-dii.fr-fil {
float: left; margin: 5px auto;
}
.fr-fic.fr-dii.fr-fir {
float: right; margin: 5px auto;
}
img.fr-dib.fr-fir {
margin-right: 0; text-align: right;
}
img.fr-dib.fr-fil {
margin-left: 0; text-align: left;
}
img.fr-dib {
margin: 5px auto; display: block; float: none;
}
img.fr-bordered {
box-sizing: content-box; border: solid 5px #CCC;
}
img.fr-shadow {
box-shadow: 10px 10px 5px 0px #cccccc;
}
img.fr-rounded {
border-radius: 10px; -moz-border-radius: 10px; -webkit-border-radius: 10px; -moz-background-clip: padding; -webkit-background-clip: padding-box; background-clip: padding-box;
}
</style><style>
p {
margin: 0;
}
span.fr-emoticon.fr-emoticon-img {
background-repeat: no-repeat !important; font-size: inherit; height: 1em; width: 1em; min-height: 20px; min-width: 20px; display: inline-block; margin: -0.1em 0.1em 0.1em; line-height: 1; vertical-align: middle;
}
span.fr-emoticon {
font-weight: normal; font-family: "Apple Color Emoji", "Segoe UI Emoji", "NotoColorEmoji", "Segoe UI Symbol", "Android Emoji", "EmojiSymbols"; display: inline; line-height: 0;
}
blockquote {
border-left: solid 2px #5e35b1; color: #5e35b1; margin-left: 0; padding-left: 5px;
}
blockquote blockquote {
border-color: #00bcd4; color: #00bcd4;
}
blockquote blockquote blockquote {
border-color: #43a047; color: #43a047;
}
table.grid {
border-collapse: collapse;
}
table.grid td,
table.grid th {
border: 1px solid #ddd;
}
.fr-fic.fr-dib {
display: block; margin: 5px auto;
}
.fr-fic.fr-dib.fr-fir {
text-align: right; margin: 5px 0 5px auto;
}
.fr-fic.fr-dib.fr-fil {
text-align: left; margin: 5px auto 5px 0;
}
.fr-fic.fr-dii {
float: none; margin: 5px auto;
}
.fr-fic.fr-dii.fr-fil {
float: left; margin: 5px auto;
}
.fr-fic.fr-dii.fr-fir {
float: right; margin: 5px auto;
}
img.fr-dib.fr-fir {
margin-right: 0; text-align: right;
}
img.fr-dib.fr-fil {
margin-left: 0; text-align: left;
}
img.fr-dib {
margin: 5px auto; display: block; float: none;
}
img.fr-bordered {
box-sizing: content-box; border: solid 5px #CCC;
}
img.fr-shadow {
box-shadow: 10px 10px 5px 0px #cccccc;
}
img.fr-rounded {
border-radius: 10px; -moz-border-radius: 10px; -webkit-border-radius: 10px; -moz-background-clip: padding; -webkit-background-clip: padding-box; background-clip: padding-box;
}
</style><div style="margin: 2rem 0px; padding: 0px; box-sizing: border-box; color: rgb(0, 0, 0); font-family: Arial, Helvetica, sans-serif; font-size: 14.6667px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: normal; text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;" data-pasted="true"><p data-pasted="true"><strong><span style="font-size: 11pt;">In this guide we will cover:</span></strong></p><p><span style="font-size: 11pt;"><strong>- What is Agentic AI?</strong></span></p><p><span style="font-size: 11pt;"><strong>- Halo's Agentic Building Blocks</strong></span></p><p><span style="font-size: 11pt;"><strong>- How Halo Facilitates Agentic AI</strong></span></p><p><span style="font-size: 11pt;"><strong>- Agentic Architecture Patterns</strong></span></p><p><span style="font-size: 11pt;"><strong>- Design Considerations</strong></span></p><p><br></p><p><strong><br></strong></p><p><strong>Related Guides:</strong></p><ul><li><a href="https://usehalo.com/haloitsm/guides/2336/" target="_blank" rel="noopener noreferrer"><strong>Conversational Agents</strong></a></li><li><a href="https://usehalo.com/haloitsm/guides/1630/" target="_blank" rel="noopener noreferrer"><strong>Custom Integrations/Runbooks</strong></a></li><li><a href="https://usehalo.com/haloitsm/guides/2597" target="_blank" rel="noopener noreferrer"><strong>Halo MCP Server Connection</strong></a></li><li><a href="https://usehalo.com/haloitsm/guides/2385/" target="_blank" rel="noopener noreferrer"><strong>Connecting AI to Halo</strong></a></li><li><a href="https://usehalo.com/haloitsm/guides/2633/" target="_blank" rel="noopener noreferrer"><strong>Halo AI – Security Implications</strong></a></li></ul><p><br></p><p><br></p><p><span style="font-size: 14pt;"><strong>What is Agentic AI?</strong></span></p><p>When we talk about Agentic AI in Halo, we're referring to systems that can make decisions and take action in a structured, iterative way. At a high level, Agentic systems are able to:</p><p><br></p><ul><li>Plan the steps needed to achieve a goal</li><li>Use tools or APIs to carry out those steps</li><li>Observe the results of each action</li><li>Choose what to do next based on those results</li></ul><p>In practical terms, this means the AI can:</p><p><br></p><ul><li>Understand user input or system events</li><li>Select relevant Halo tools (Ticketing, Runbooks, knowledge search, etc.)</li><li>Execute those tools</li><li>Interpret the output</li><li>Continue until the goal is reached</li></ul><p>Agentic behaviour emerges from the interaction between conversational reasoning, procedural automation, and tool invocation.</p><p><br></p><p data-pasted="true">This guide focuses on how the components fit together to support Agentic workflows. Each component—Virtual Agents, Runbooks, and MCP—has its own dedicated configuration guide.</p><p><br></p><p><br></p><p><strong><span style="font-size: 14pt;">Halo's Agentic Building Blocks</span></strong></p><p>Halo provides four core components that can be combined to create systems with agency.</p><p><br></p><p><strong><span style="font-size: 12pt;">Virtual Agents — Conversational reasoning</span></strong></p><p>Virtual Agents interpret natural language, identify user intent, and decide which tools to call. They are central to chat-based workflows.</p><p><br></p><p><strong><span style="font-size: 12pt;">Runbooks — Procedural reasoning and data-centric actions</span></strong></p><p>Runbooks offer structured, multi-step logic: conditions, loops, SQL queries, report access, custom API calls, and LLM-based evaluation. They are well-suited for data-driven or multi-stage workflows.</p><p><br></p><p><strong><span style="font-size: 12pt;">MCP Server — Tool layer for any agent</span></strong></p><p>The Model Context Protocol (MCP) exposes Halo tools to external or internal agents. External LLM systems or internal Runbooks can dynamically call the same toolset.</p><p><br></p><p><strong><span style="font-size: 12pt;">Halo Runbooks → MCP — Internal autonomous agent</span></strong></p><p>Runbooks can call Halo's MCP endpoint directly. This supports the following closed-loop autonomy:</p><p><br></p><ul><li>A Runbook gathers data</li><li>An LLM evaluates that data</li><li>The Runbook calls MCP tools to take action</li><li>The loop continues until complete</li></ul><p><img src="https://halo.haloservicedesk.com/api/attachment/image?token=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpZCI6ImI1NzVmMGM0LTE2MTctNDA1ZC1iODM2LTJhODRkZTUxNjE1NiJ9.8Uh5GTD6zQKVcfmJKwSMad9LxXhrR7637f5qn5ntuqM" class="fr-fic fr-fil fr-dib" width="1183" style="width: 1185px; height: 658.833px;" height="659"></p><p><strong><span style="font-size: 10pt;">Fig 1. Agentic Workflow</span></strong></p><p><br></p><p><strong><span style="font-size: 12pt;">Full circle — VA → Runbook → MCP → Runbook</span></strong></p><p>Working together, these components support end-to-end agentic workflows:</p><p><br></p><ul><li>The Virtual Agent handles conversation and intent</li><li>The Runbook handles structured reasoning</li><li>MCP provides the tool execution layer</li><li>Runbooks can continue autonomously after initial input</li></ul><p><br></p><p><br></p><p><strong><span style="font-size: 14pt;">How Halo Facilitates Agentic AI</span></strong></p><p>Agentic behaviour in Halo arises from the interaction of inputs, prompts, tools, and adaptive feedback.</p><p><br></p><p>Agentic workflows can begin through several entry points:</p><p><br></p><ul><li>Conversations handled by Virtual Agents</li><li>Runbook triggers such as events, workflow actions, or schedules</li><li>External systems calling tools via MCP</li><li>Internal loops where a Runbook calls MCP and continues execution</li></ul><p><br></p><p>Prompts define the reasoning context at each stage:</p><p><br></p><ul><li>Virtual Agent instructions describe the agent's behaviour and goals</li><li>Chat step instructions allow stage-based reasoning (e.g., classify, decide, route)</li><li>Runbook prompts help the LLM evaluate structured data and decide on actions</li><li>MCP agents use prompts outside Halo to plan and sequence tool calls</li><li>Runbook-to-MCP prompts enable internal decision-making loops</li></ul><p><br></p><p>Tools represent the capabilities the agent can choose from:</p><p><br></p><ul><li>System tools such as Ticketing or knowledge search</li><li>Virtual Agent custom functions</li><li>Runbook Halo API actions</li><li>Runbook external API integrations</li><li>MCP tools exposed to internal or external agents</li></ul><p>Prompted reasoning determines which tool to use, and the results shape what happens next.</p><p><br></p><p><br></p><p>Feedback is central to Agentic behaviour:</p><p><br></p><ul><li>Virtual Agents interpret tool outputs and determine next steps</li><li>Runbooks branch or loop based on results, or evaluate LLM output</li><li>MCP agents read structured responses and adjust their plan</li></ul><p>These feedback loops allow Halo to support iterative, adaptive workflows.</p><p><br></p><p><br></p><p><strong><span style="font-size: 14pt;">Agentic Architecture Patterns</span></strong></p><p><br></p><p><strong><span style="font-size: 12pt;">Chat First Virtual Agent</span></strong></p><p>The Virtual Agent interprets the request, chooses a tool, and carries out the action. This is useful for simple triage, Ticket logging, and information retrieval.</p><p><br></p><p>Example: A user reports a laptop issue. The VA searches knowledge, doesn't find a match, logs a ticket, asks for an attachment, and confirms the Ticket ID.</p><p><br></p><p><strong><span style="font-size: 12pt;">Virtual Agent with Runbook automation</span></strong></p><p>The Virtual Agent handles the conversation; while the Runbook performs the actual work. This is useful for multi-step workflows, integrations, and structured automation.</p><p><br></p><p>Example: An agent asks for the health of their assigned Tickets. The VA triggers a Runbook that uses an LLM to score each Ticket and escalates those at risk. The VA then summarises the result.</p><p><br></p><p><strong><span style="font-size: 12pt;">Autonomous Runbook</span></strong></p><p>Runbooks act as standalone agents running on schedules or events. This is useful for scheduled reviews, evaluations, back-end processing</p><p><br></p><p>Example: A nightly SLA review Runbook evaluates upcoming breaches using an LLM and escalates where needed.</p><p><br></p><p><strong><span style="font-size: 12pt;">External agent via MCP</span></strong></p><p>External agents plan the workflow and use MCP to execute Halo actions. This is useful for enterprise AI assistants or cross-platform orchestration.</p><p><br></p><p>Example: An intranet assistant submits an equipment request. It uses MCP to search services, log the request, and confirm the action back to the user.</p><p><br></p><p><strong><span style="font-size: 12pt;">"LLM-as-judge" chat flow</span></strong></p><p>Chat steps create a guided sequence of reasoning. This can be used for classification, routing, and approvals.</p><p><br></p><p>Example: The VA classifies a request, checks for risks, decides the appropriate team, and escalates to a human if needed.</p><p><br></p><p><strong><span style="font-size: 12pt;">Closed-loop Runbook → MCP → Runbook</span></strong></p><p>Runbooks analyse data, decide what to do, call MCP tools to take action, and repeat until finished. This is useful for fully autonomous workflows.</p><p><br></p><p>Example: A CMDB cleanup cycle uses an LLM to assess assets, updates them via MCP, and loops until all items are processed.</p><p><br></p><p><br></p><p><strong><span style="font-size: 14pt;">Design Considerations</span></strong></p><p>Where the reasoning should live:</p><ul><li>Virtual Agents for conversational understanding</li><li>Runbooks for structured, data-driven decisions</li><li>External MCP agents for enterprise-scale orchestration</li><li>Runbook→MCP for internal autonomous loops</li></ul><p>Synchronous and asynchronous behaviour:</p><ul><li>Chat workflows operate in real time</li><li>Background workflows run independently</li></ul><p><strong><span style="font-size: 12pt;">Prompt strategy</span></strong></p><p>Prompts should be consistent across layers. Use chat steps for staged reasoning and Runbook prompts for structured evaluation.</p><p><br></p><p><strong><span style="font-size: 12pt;">Tool exposure</span></strong></p><p>Provide only the tools each agent needs. Smaller toolsets lead to clearer, more predictable behaviour.</p></div>
