Runtype
Records

Creating and managing records

Records are the structured data layer in Runtype. They let you store customer profiles, product catalogs, support tickets, or any information your Flows and Agents need to do their work. Think of them as your AI-ready database: organized by type, enriched with metadata, and accessible from any Flow. If you need a conceptual overview first, see What are Records?.

Create a Record

From the Records page, click Create Record to open a dropdown with three options:

  • Quick create — Create a Record from scratch

  • Use template — Start with a pre-built structure with the name, type, and metadata pre-filled

  • Import CSV file — Bulk upload Records from a spreadsheet

Quick create

  1. Click Records in the sidebar.

  2. Click Create Record and choose Quick create.

  3. Enter a Name and Type. Both are required. Optionally click Add Field to add metadata key-value pairs.

  4. Click Create Record to save, or Create & Edit to save and open the Record details.

The Type field autocompletes with types already in your workspace. You can also type a new one, and it will be available for future Records automatically.

Metadata fields

Metadata is stored as JSON key-value pairs. In the Quick create form, you enter values as text. For more complex structures such as objects or arrays, use the API or a Flow with an Upsert Record step.

Example metadata:

customerName: Jane Doe
orderCount: 47
isPremium: true

Use a template

Templates give you a head start with pre-defined metadata fields tailored to common use cases. Runtype includes templates across four categories:

  • Business — Customer, Lead, Employee

  • E-commerce — Product, Order

  • Marketing — Marketing Campaign, Contact

  • General — Project, Task, Generic Record

Select a template to open the Quick create form with the name, type, and metadata fields pre-filled. You can change any field before saving.

Bulk upload via CSV

Import many Records at once from a spreadsheet:

  1. Click Create Record, then Import CSV file. From the empty state, you can also click Import records.

  2. Upload your CSV file. The first row must contain column headers.

  3. Map columns by choosing the Type column and Name column. Both are required. All other columns are imported as metadata.

  4. Click Upload to create the Records.

Example CSV:

customerId,name,email,tier
12345,Jane Doe,[email protected],premium
67890,John Smith,[email protected],standard

Runtype auto-detects columns named type and name, so you often will not need to map them manually.

Search, filter, and organize Records

The Records table gives you several ways to find what you need:

  • Search — Find Records by name or metadata

  • Filter by type — Narrow the list to a specific Record type

  • Column visibility — Choose which columns are visible from the column settings

  • Export CSV — Download your filtered Records as a CSV file

The table uses cursor-based pagination with page sizes of 25, 50, 100, or 1,000 rows.

Record types

Every Record has a type, which is a label you define, such as Customer, Support Ticket, or Product Listing. Types help you filter the Records list, let Flows and Agents find the right Records, and keep metadata consistent across similar Records.

Types are created on the fly. Type a new one when you create a Record, and it becomes available everywhere in your workspace. For a broader overview of how Records fit into Runtype, see What are Records?.

Edit a Record

  1. Click a Record row or the View Details action to open the details sheet.

  2. Click Edit, then update the name, type, or metadata fields.

  3. Click Save to apply your changes.

You can also copy a Record ID from the row action menu when you need to reference a Record in Flows or the API.

Delete Records

Delete a single Record: Open the Record action menu and choose Delete, then confirm.

Delete multiple Records: Select Records with the checkboxes, then click Delete Selected in the selection bar and confirm.

Deletion is permanent. If a Flow references a deleted Record, it will return null when it tries to retrieve it.

View results and run Flows

Each Record tracks its execution history. From the Record details sheet, click the Results tab to see which Flows have run on the Record, each run's status, and prompt-level details such as output, token usage, duration, and cost. You can also filter results by status, Flow, or sort order.

From the Records table, use the Run action to execute a Flow directly on a selected Record. This is a fast way to test or process individual Records without leaving the page. If you are new to Flow building, see Creating and Editing Flows.

Using Records with Flows

Records and Flows work together closely:

  • Upsert Record steps can create or update Records automatically

  • Retrieve Record steps pull Record data into a Flow for processing

  • Generate Embedding and Store Vector steps let you index Record content for semantic search

This lets you build Flows that enrich, transform, and act on your Records, from populating metadata to running AI analysis across your dataset. For a broader view of available Flow building blocks, see Flow step types overview.

Best practices

  • Keep metadata consistent — Use the same field names within a Record type so your Flows can reliably access the data they need

  • Choose descriptive types — Clear names such as Customer Profile or Support Ticket make filtering and Flow configuration easier

  • Keep metadata focused — Store the data your Flows and Agents will actually use

  • Use templates for consistency — Start from a template when you create many similar Records

Next steps

Was this helpful?