# Designing an Agent Workflow ### About this export | Field | Value | | --- | --- | | **content_type** | lesson | | **platform** | contentstack-academy | | **source_url** | https://www.contentstack.com/academy/courses/agentos-foundations/designing-an-agent-workflow | | **course_slug** | agentos-foundations | | **lesson_slug** | designing-an-agent-workflow | | **markdown_file_url** | /academy/md/courses/agentos-foundations/designing-an-agent-workflow.md | | **generated_at** | 2026-06-19T08:30:58.344Z | > Part of **[Agent OS Foundations](https://www.contentstack.com/academy/courses/agentos-foundations)** on Contentstack Academy. **Academy MD v3** — structured for retrieval; no quiz or assessment keys. #### Video details #### At a glance - **Title:** Designing an Agent Workflow - **Duration:** 4m 7s - **Media link:** https://cdn.jwplayer.com/previews/xDVXljW9 - **Publish date (unix):** 1780928273 #### Streaming renditions - application/vnd.apple.mpegurl - audio/mp4 · AAC Audio · 113696 kbps - video/mp4 · 180p · 180p · 176270 kbps - video/mp4 · 270p · 270p · 213403 kbps - video/mp4 · 360p · 360p · 246345 kbps - video/mp4 · 406p · 406p · 265358 kbps - video/mp4 · 540p · 540p · 339218 kbps - video/mp4 · 720p · 720p · 433102 kbps - video/mp4 · 1080p · 1080p · 733518 kbps #### Timed text tracks (delivery) - **thumbnails:** `https://cdn.jwplayer.com/strips/xDVXljW9-120.vtt` #### Video transcript At this point, we've explored what agents are, how they differ from automations, what kinds of problems they're good at solving, and where they may not be the right solution. Now let's bring all that together into a practical framework you can use before building any agent. One of the biggest problems that people have when it comes to working with agents is starting with the technology. They open the platform, they start adding tools, they connect integrations, and they try to figure out what the agent should do. But successful agent design usually works in the opposite direction. You start with the problem, then you design the solution. Only then do you build it. To help with that process, I like to use a simple four-part framework. Goal, inputs, tools, outputs. Let's walk through each one. First, the goal. This is the most important part. What are you actually trying to accomplish? Not what tool do you want to use, not what integration you want to connect with. What's the outcome? For our news intelligence agent example, the goal is simple. Identify important AI news stories and communicate the findings to the organization. That's the objective. Everything else exists to support that objective. Next, we have inputs. Inputs are the information the agent needs in order to do its work. What data will it consume? What information will it analyze? What context does it require? Using the same example, the inputs are news articles, search results, and information discovered through web search. Without inputs, the agent has nothing to reason about. Third, we have tools. Tools determine what actions the agent can perform. Notice, the tools come third, not first. Too often people start by asking, what tools can I connect? Instead, ask, what tools does the objective require? In our example, the agent needs a web search capability to discover information, it needs a content stack capability to create an entry, and it needs a Slack capability to communicate results. Those tools support the goal, they don't define the goal. Finally, we have outputs. What should success look like? What is the deliverable? What should exist when the agent finishes its work? For that news intelligence agent, success means a set of content entries exist inside of content stack, a summary has been delivered to the appropriate Slack channel, the information is available for the organization to use. The output should be crystal clear before the first line of instructions is ever written. And that's where many agent projects struggle. People spend hours thinking about triggers, tools, and models, but they haven't clearly defined what success looks like. If the output is unclear, the instructions become unclear. If the instructions become unclear, the results are unpredictable and inconsistent. So before you build an agent, pause and walk through this framework. What's the goal? What inputs are required? What tools are needed? What outputs define success? If you can answer those four questions, you're already most of the way towards a successful implementation. Because good agent design rarely starts inside the software. It starts with understanding the workflow you're trying to improve. #### Key takeaways - Connect **Designing an Agent Workflow** back to your stack configuration before moving to the next module. - Capture one concrete artifact (screenshot, Postman call, or code snippet) that proves the step works in your environment. - Re-read the delivery versus management boundary for anything you changed in the entry model. ## Supplement for indexing ### Content summary Designing an Agent Workflow. Designing an Agent Workflow in Agent OS Foundations (agentos-foundations). ### Retrieval tags - Designing - Agent - Workflow - agentos-foundations - lesson 06 - Designing an Agent Workflow - agentos-foundations lesson ### Indexing notes Index this lesson as a primary chunk tagged with lesson_id "06" and topics: [Designing, Agent, Workflow]. Parent course slug: agentos-foundations. Use asset_references URLs as thumbnail hints in search results when present. Never surface LMS quiz content or assessment answers from this file. ### Asset references | Label | URL | | --- | --- | | Video thumbnail: Designing an Agent Workflow | `https://cdn.jwplayer.com/v2/media/xDVXljW9/poster.jpg?width=720` | ### External links | Label | URL | | --- | --- | | Contentstack Academy home | `https://www.contentstack.com/academy/` | | Training instance setup | `https://www.contentstack.com/academy/training-instance` | | Academy playground (GitHub) | `https://github.com/contentstack/contentstack-academy-playground` | | Contentstack documentation | `https://www.contentstack.com/docs/` |