> For the complete documentation index, see [llms.txt](https://docs.lleverage.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.lleverage.ai/run-operations/agent/examples.md). # Agent task examples See representative Agent tasks in operational terms: the outcome, evidence, human boundary and result. These examples are deliberately generic. They show the shape of good Agent work, not preconfigured templates that every organisation receives. The exact systems, rules, approvals and outputs depend on the Project and the capabilities available to the run. ## Supplier delivery exception **Outcome:** understand which open purchase orders need attention and move the follow-up forward. **Evidence and systems:** current purchase orders in the ERP, supplier email, expected delivery dates, supplier-specific context and any relevant Skill. **What the Agent does:** finds missing or changed commitments, reconciles them with the current order, drafts or sends follow-up where authorised, interprets the supplier reply and prepares the operational update. **Human boundary:** asks when the supplier response is ambiguous, when an update needs approval, or when the consequence requires a planner or buyer to choose between options. **Result:** an updated commitment where authorised, a clear exception where it is not, and evidence showing what changed and why. **What may become repeatable:** scheduled detection and routine chasing can move into a Workflow while the Agent remains useful for irregular supplier responses and exceptions. ## Sales order from an email or PDF **Outcome:** turn an incoming customer order into a correct ERP order without guessing unresolved commercial facts. **Evidence and systems:** customer email and attachments, ERP customer and product records, customer-specific mappings, pricing or commercial policy, and order-processing Skills. **What the Agent does:** extracts the requested products, quantities and dates, resolves them against current customer and product data, checks the applicable commercial context and prepares the order. **Human boundary:** stops when a customer, product mapping, price, term or requested date cannot be resolved confidently, or when the write itself requires approval. **Result:** a created or prepared ERP order plus a concise record of any unresolved exception and the source evidence used. **What may become repeatable:** common document shapes and stable validation rules can be captured in a Skill and, where the path becomes predictable enough, moved further into a background Workflow. ## Investigate production-plan risk **Outcome:** explain where the current production plan is at risk and give the planner the evidence needed to act. **Evidence and systems:** demand, open orders, material availability, supplier commitments, production capacity, planning spreadsheets or Tables and the current planning system. **What the Agent does:** gathers the relevant records, reconciles inconsistencies, identifies the constraints behind the risk and traces the affected orders or periods. **Human boundary:** planning trade-offs that change priorities, customer commitments or capacity allocation remain with the planner unless explicitly delegated. **Result:** an evidence-backed explanation of the risk, affected work and concrete options rather than a generic summary of the data. **What may become repeatable:** recurring risk indicators can feed Monitoring or a scheduled Workflow, while the Agent investigates the cases whose cause or remedy is not known in advance. ## Prepare a tender or RFP response **Outcome:** create a complete draft response grounded in maintained company facts and the customer's actual request. **Evidence and systems:** the tender documents, approved company information, prior answers, product/service material, templates and any customer or opportunity context. **What the Agent does:** identifies the questions and required evidence, retrieves the relevant maintained facts, drafts the response, tracks missing information and produces the requested document or spreadsheet output. **Human boundary:** pricing, legal commitments, unsupported claims and genuinely new business decisions stay with the appropriate person. **Result:** a reviewable response package with the remaining gaps made explicit instead of silently invented. **What may become repeatable:** the document method, approved sources, deterministic generation and regression cases can become a Skill so each new tender starts from a tested operating method. ## What these examples have in common The Agent is useful here because the route depends on evidence. It can combine structured records, documents, connected systems and reusable methods without requiring the whole process to be hard-coded in advance. People remain at the boundaries where authority or judgement matters. When enough of a path becomes stable, move that part into a Workflow rather than forcing the Agent to rediscover a deterministic sequence every time. For a wider set of business-process examples, see [Operational examples](/understand-lleverage/operational-examples.md). --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.lleverage.ai/run-operations/agent/examples.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.