Getting started
CLI workflows
Choose an AEO capability: measure AI visibility, analyze brand perception and competitors, or improve content. New users can start with CLI installation.
Give the agent an objective
You do not need to name a command. Specify the subject, the question, and the result you need.
Compare this brand’s last two weeks using the same platforms and prompts. Read answers for the prompts that changed, then choose three pages to investigate using quotes and citation URLs as evidence.
The agent selects the relevant workflow and proceeds in this order:
- Resolve brand and topic names to IDs from CLI lists. Fix the workspace, platforms, dates, and sample size.
- Fetch the current evidence needed for the decision. Do not reconstruct configuration, answer text, citations, or observed queries from memory.
- Read the result before choosing the next operation. If aggregates cannot answer the question, inspect complete answers. Leave conclusions unresolved when evidence is missing.
- Compare, prioritize, or write with the agent, then return answer IDs, URLs, conditions, and unknowns. Apply only approved mutations and read the results back.
A skill supplies the trigger, decision, and evidence requirements. Use the CLI reference and --help for current arguments. Installing a skill does not grant data access or make an unavailable command executable.
Each page identifies what the CLI fetches or changes and what the agent interprets or writes. Start with CLI installation. Before working with business data, confirm the workspace with orc workspace current --json.
Installation and setup
Start with CLI installation. If you do not have an account, sign up on the web first.
| Workflow | Outcome |
|---|---|
| Install the CLI and read your first data | Installation through the first read |
| Link a directory to a workspace | Select an existing target and verify a read |
| Run the first observation and read its results | Generate and confirm candidates, then retrieve the initial batch results |
Management and operations
Create workspaces and manage ongoing operation for clients and pitches.
| Workflow | Outcome |
|---|---|
| Create a workspace | Prepare an empty workspace with the same account, then start its first observation. |
| Start monitoring a client brand | Create a client or pitch workspace from one account, then review its brand, prompts, and first observations. |
| Move a pitch into ongoing monitoring | Preserve pitch observations while reviewing client settings and capacity before starting ongoing monitoring. |
| Check usage limits for each client | Read organization workspace quotas and each client's entitlement summary, then handle shortages or spare capacity within a workspace. |
| Pause and resume client monitoring | Stop collection while retaining client history, then verify capacity and settings before resuming. |
Agent connections and automation
| Workflow | Outcome |
|---|---|
| Connect an agent and verify a read | Verify connection, authentication, and retrieval |
| Send your first request with an API key | Issue a key, read data, and rotate it |
| Give CI only the permissions it needs | Run automation with limited permissions |
| Distribute Skills to a project or team | Verify installation targets and scope |
AI visibility
| Workflow | Outcome |
|---|---|
| Recording a visibility baseline | Check the brand and measurement scope, then save visibility, citations, and sentiment for the next comparison. |
| Investigating a visibility change | Align comparison conditions, inspect the relevant answers, and separate observed changes from hypotheses to investigate. |
Brand perception
| Workflow | Outcome |
|---|---|
| Audit the actual AI answers | Agree on an answer sample, then report wording, competitors, and claims with quotes and counts. |
| Checking how a brand is described | Read sentiment and answer text to identify pricing or feature claims that need verification. |
| Choose a brand perception gap | Inspect attribute mentions and competitor rankings, then choose one attribute and read its evidence. |
| Build an approved brand fact base | Turn approved product, pricing, and specification information into atomic facts with sources. |
| Check consistency across owned surfaces | Quote and compare your name, category, offering, and audience across owned pages. |
Competitors and citations
| Workflow | Outcome |
|---|---|
| Analyze a competitor’s strengths | Find a competitor’s strong topics and changing visibility, then inspect answers and citations to select an opportunity. |
| Investigating competitor citation gaps | Find domains and URLs that cite competitors, then select opportunities with supporting evidence. |
| Look up one source | Answer a question about one URL or domain with its observed retrievals, citations, and scope. |
| Benchmark pages within their type | Compare your position within observed homepages, product pages, comparison pages, and other page types. |
Content optimization
| Workflow | Outcome |
|---|---|
| Audit site-scoped ChatGPT queries | Find ChatGPT queries scoped to your domain and map them to existing or missing answers on your site. |
| Find missing topics on a page | Compare observed search queries with page text to identify missing topics and where to add them. |
| Draft content from evidence | Read target prompts, observed queries, and cited pages, then draft the amount of content the task requires. |
| Improve a page’s wording and structure | Inspect query intent, answer placement, and supporting evidence, then return an explained rewrite. |
| Comparing before and after an update | Record the changed URLs and publication date, then collect comparable answers and citations to identify changes and follow-up work. |
Shopping visibility
| Workflow | Outcome |
|---|---|
| Set up shopping prompts from products and personas | Build product-linked prompts from categories, comparisons, personas, and consideration stage. |
AI crawlers and site diagnostics
| Workflow | Outcome |
|---|---|
| Investigating bot access | Fetch access data for a connected domain and compare it with citation data to choose pages to investigate. |
Measurement management
| Workflow | Outcome |
|---|---|
| Clean up brands and competitors | Inspect missing brands, duplicates, domains, and aliases, then verify approved changes. |
| Set up prompts across the buyer journey | Create a balanced set of awareness, comparison, decision, and brand-evaluation prompts. |
| Reviewing prompt coverage | Inspect tracked questions, topics, and collection settings, then propose questions to add or revise. |
| Manage measurement settings and saved filters | Update workspace location, language, and models, then read back revisions and reusable filters. |
| Clean up topics and tags | Inspect prompt classifications and fix duplicates, thin topics, and tags that no longer distinguish the data. |
| Build personas from audience evidence | Use audience evidence to define personas, then connect approved profiles to prompt setup. |
Reporting and sharing
| Workflow | Outcome |
|---|---|
| Build a custom report | Turn requested rows, metrics, brands, and dates into a comparison table with explicit exclusions. |
| Handing an investigation to an agent | Save JSON and comparison conditions to produce an evidence-linked memo or weekly report. |
Updates and troubleshooting
| Workflow | Outcome |
|---|---|
| Update, switch, or remove setup | Verify updates and remove configuration |
| Find the failing setup stage | Identify the failed stage and next action |
| Diagnose missing data | Check scope, collection state, filters, and access conditions to distinguish missing data from failures. |
| Explain a setting or metric | Use the documentation and workspace configuration to explain current product behavior. |
| Report a failure and verify the fix | Submit a reproduction, keep the receipt, and verify the fixed result. |
Turn failures into improvements
If a failure remains, use the shared feedback workflow. Include the workflow, failed step, and expected and actual results. Submit the reviewed content with orc feedbacks create, retain the receipt, and verify the fix under the same conditions.
Forge
Sync, clone, and review with Forge: the target workflow connecting durable synchronization, local retrieval, and PR reads.