Runtype
Records

What are Records?

Records are the structured data layer of Runtype. They let you store, organize, and retrieve information that your Flows and Agents can use as context, whether that is a customer database, a Product catalog, a knowledge base, or anything else your AI needs to work with.

Why use Records?

Records give your AI the information it needs to deliver relevant, personalized responses. Common use cases include:

  • Knowledge bases — Store documentation, FAQs, or Product details so your Agents can answer questions accurately

  • Customer profiles — Keep customer preferences, history, and account details in one place for personalized interactions

  • Product catalogs — Maintain inventory, specs, and pricing that Flows can reference and update

  • Conversation history — Save chat logs and interaction summaries for continuity across sessions

  • Semantic search (RAG) — Enable AI-powered search that finds Records by meaning, not just exact keywords

How Records are structured

Every Record has four main parts:

  • Name — A display label to help you identify the Record, up to 500 characters

  • Type — A category that groups related Records together, such as customers, products, or docs

  • Metadata — A flexible JSON object where your actual data lives

  • ID — A unique identifier generated by Runtype

Here is an example Record:

{
  "id": "rec_01h2xabc123...",
  "name": "Jane Doe",
  "type": "customers",
  "metadata": {
    "email": "[email protected]",
    "tier": "premium",
    "totalOrders": 47,
    "lastPurchase": "2024-01-10"
  }
}

Flexible metadata

Metadata can hold any valid JSON, so you can structure it however your use case requires. That includes simple key-value pairs, nested objects, arrays, or a mix of all three.

{
  "productId": "PROD-123",
  "name": "Widget Pro",
  "price": 99.99,
  "specs": {
    "weight": "2.5kg",
    "dimensions": "10x5x3cm"
  },
  "tags": ["electronics", "popular"]
}

Runtype supports metadata labels, which are friendly display names for your metadata keys. For example, a key stored as company_name can display as Company Name in the dashboard. These labels also carry through to CSV exports, which makes your data easier to read and share.

Organizing with Record types

Types let you group Records into collections, similar to tables in a database. For example:

  • customers — Customer profiles and preferences

  • orders — Order and transaction data

  • docs — Documentation articles

  • faqs — Frequently asked questions

Each type can have its own metadata structure. Use the type filter on the Records page to quickly find what you need. To start populating your data store, see Creating and managing records.

Getting started with Records

Runtype gives you several ways to create Records:

  • Templates — Start with pre-built templates such as Customer, Lead, Product, or Order

  • CSV import — Upload a spreadsheet to create Records in bulk. Runtype normalizes your column headers and preserves the original names as metadata labels

  • Flow steps — Use the upsert-record step in your Flows to create or update Records automatically as part of a workflow

  • API — Create and manage Records programmatically through the Runtype API

Managing Records in the dashboard

The Records page gives you a clear view of your data and tools to stay organized:

  • Search and filter — Search by name or metadata, and filter by Record type

  • Metadata preview — See the first few metadata fields as badges in the table

  • Bulk actions — Select multiple Records to delete or edit them at once

  • Details panel — Open a slide-over with tabs for metadata, labels, execution results, and cost breakdowns

  • Run with a Flow — Execute a Flow directly from the table using a Record as input

  • Export — Download your Records as CSV for reporting or external use

Semantic search with embeddings

You can enable semantic search by adding generate-embedding and store-vector steps to a Flow. These steps create vector representations of your Record data and attach them to the Record. If you are new to building workflows, start with What are Flows? and Flow step types overview.

Once set up, a search like affordable laptop for students can surface Records about budget-friendly computers even if those exact words do not appear in the data.

Tips for working with Records

  • Use meaningful types — Consistent type names make filtering faster and keep your data organized

  • Start with templates — Pre-built templates give you a strong starting point, and you can customize fields later

  • Add metadata labels — Labels make the dashboard cleaner and CSV exports easier to read

  • Leverage Flows — Set up Flows that create or update Records from external data sources so your information stays current

Next steps

Was this helpful?