Getting Started

Getting Started with Akurion

This guide walks through the fastest path from a new account to a useful cited answer.

1. Create or Join a Workspace

Sign in at Akurion. After email verification or SSO completion, Akurion opens the app workspace where you can create projects, manage sources, and use AI chat.

If your organization already uses Akurion, ask an admin to add you to the right project or organization.

2. Create a Project

A project is a scoped knowledge base. It contains:

• Data sources and uploaded files.
• Chat threads and saved messages.
• Members and roles.
• Metadata settings and project instructions.
• Retrieval settings such as topK, token limit, model, and web search toggle.
• Graph RAG settings and graph build state.
• Project API keys.

Create a project from the app dashboard, give it a clear name, and add a short description so teammates understand its purpose.

3. Configure Retrieval Settings

Open project settings and review:

Setting	Recommended starting point
Search topK	50 for balanced recall and latency
Token limit	100000 for broad answers without excessive context
Web search	Off for strictly internal knowledge, on only when external context is allowed
Project instructions	Add domain-specific guidance for answer tone, allowed sources, and formatting
Metadata keys	Add fields users will filter by, such as department, product, region, customer, owner, date, or document type
Graph RAG	Enable when entity and relationship discovery matters

4. Add Data

Use the Data section to upload files or add an enabled source type for your workspace. Connector availability is controlled in the app; disabled source types are intentionally omitted from public setup docs until they are enabled for customers.

5. Wait for Processing

Akurion processes content in stages:

1. Discover files or records from the source.
2. Download or access content.
3. Extract text and structured content.
4. Generate metadata when enabled.
5. Prepare the content for search and answer generation.
6. Update the project’s knowledge index and processing state.
7. Publish file status and credit usage updates.
8. Build relationship context when Graph RAG is enabled.

Check the source status and file status before relying on a project for answers.

6. Ask Your First Question

Open chat and ask a question that should be answerable from the indexed content.

Good first questions:

• “Summarize the main policies in this project with citations.”
• “Which documents mention data retention requirements?”
• “What changed in these release notes?”
• “List open risks from the uploaded vendor documents.”
• “Compare the uploaded incident reports related to this application.”

Review the answer, citations, source list, and any processing summary. If the answer misses context, adjust filters, increase retrieval depth, or connect additional sources.

7. Use Metadata Filters

If your project has metadata keys, use filters to narrow retrieval by fields such as:

• Source type, when available
• File type
• Date
• Department
• Region
• Customer
• Product
• Document category

Metadata filters can be used from the UI and from developer APIs.

8. Invite Teammates

Add members from the project management area. Use roles to control who can administer the project, manage data sources, or use the knowledge base.

For organization-level administration, configure workspace admins and SSO settings where available.

9. Create API Access

Create an API key when you want to use Akurion from code, agents, or automation.

API key types:

Key type	Use case
Project key	Access one project directly.
Subscription key	Access all or selected projects in a workspace. Requests must include a project ID when the tool needs project context.

Keep the plaintext key secure. Akurion stores key material hashed at rest and shows plaintext only at creation time.

10. Try the Developer API or MCP

Use the API explorer or call the endpoints directly:

curl -X POST "https://api.structhub.io/api/v1/queries/answer" \
  -H "Content-Type: application/json" \
  -H "API-KEY: YOUR_API_KEY" \
  -H "X-Project-ID: YOUR_PROJECT_ID" \
  -d '{
    "query": "What are the top risks mentioned in this project?",
    "search_mode": "focused",
    "max_depth": 5,
    "web_search": false
  }'

For MCP-compatible clients, use the JSON-RPC endpoint:

curl -X POST "https://api.structhub.io/mcp-server" \
  -H "Content-Type: application/json" \
  -H "API-KEY: YOUR_API_KEY" \
  -H "X-Project-ID: YOUR_PROJECT_ID" \
  -d '{
    "jsonrpc": "2.0",
    "method": "tools/call",
    "params": {
      "name": "answer_query",
      "arguments": {
        "query": "Summarize this project with citations",
        "search_mode": "focused"
      }
    },
    "id": 1
  }'

Recommended First Pilot

For an enterprise pilot, start with:

• One workspace.
• One or two high-value projects.
• A small, trusted data set.
• A short list of evaluation questions.
• A few metadata fields that match how users search.
• One developer API key for integration testing.

Measure whether users can get accurate, cited answers faster than their current process.