Spark AI Autofill: Automatically Populate Record Fields with AI

Spark AI Autofill helps you accelerate form completion by automatically filling out editable fields in a record using AI-generated suggestions. You can provide an AI prompt, upload a supporting file, or use both to generate context-aware responses.

Autofill Demo Video

How to enable Spark AI Autofill

1. Log in to Risk Cloud as an admin user.
2. Go to Admin → Integrations → Spark AI Integration Card.

3. Click Configure, then enable Autofill Record (Beta). 

   By enabling Autofill in the Spark AI Integration Card, Autofill becomes available across all Applications and Workflows within the environment. Record assignees—both internal and external users—can then use Autofill to complete records efficiently.

How to autofill a record?

1. Assign the record to yourself. Only the record assignee can see and access the Autofill banner on the record.

2. Configure Autofill Sources: Click the AI Autofill button on the banner on a record to open the Configure Autofill Modal. Choose how Spark AI should generate suggestions. You can:

• Use an AI prompt

• Upload a file. Supported formats include: .PDF, .CSV, .JSON, .TXT, .PNG, .JPG, .JPEG, .GIF, .WEBP, .WAV, .MP3, .MPEG, .MPGA, .M4A, .OGG, .WEBM, .DOCX, .PPTX, and .XLSX.

• Combine both (for richer context)

Note: 

• The following limitations when using Autofill with a source prompt or file:
  • Maximum file size: 100 MB
  • Maximum prompt length: 1,047,576 tokens (~785,000 words)
  • Attachments: Only one attachment is supported per Autofill session (Beta version)

3. Review AI Recommendations: Spark AI scans all blank, editable, and visible fields across every page of your record in the current Workflow Step. Suggested values are automatically filled in and marked with a purple “Autofilled with Spark AI” label. Autofill may take up to 2–3 minutes to populate values depending on the number of fields on the record and input source complexity.

   Note:

   • All AI-generated values are autosaved and the Autofilled with Spark AI label is preserved on the Record across Steps. If you manually change a value, the AI label is removed.
   • All Autofill-generated values are also recorded in the Field Audit Log, preserving the history of what Spark AI previously populated.
   • Autofill will populate supported fields only when enough context is available from the selected sources. If the provided prompt or file lacks sufficient information, some supported fields may remain blank.

4. Filling Conditional Fields: When new fields appear due to conditional logic, click Autofill Empty Fields to automatically populate the newly revealed blank fields. 

   After each Autofill run in a session, Spark AI displays key statistics, including:

   • Number of fields autofilled in the latest run
   • Number of new empty fields revealed
   • Autofill run time in seconds
   • Source file used if applicable

   Note: Refreshing the page resets the Autofill session. When this happens, previous Autofill session stats will be cleared.

Supported Field Types

Autofill Best Practices

AI might make mistakes, so always review AI-generated values before submitting your record.

Autofill prioritizes parsing and searching the information provided in your prompt and/or uploaded file to generate recommendations. It also considers each field’s context—such as workflow name, field name, label, tooltip, and guidance text—to ensure relevance. 

To improve Autofill accuracy, provide detailed prompts or input files and include meaningful field context in your records.

Roadmap

The following capabilities are not yet available in Spark AI Autofill Beta. If you’re interested in these enhancements, upvote the related Feature Requests to help us prioritize them in future releases:

• Overwrite existing field values with Autofill
• Display confidence scores and citations per field
• Retain session stats and source history after page refresh
• Support password-protected file uploads
• Allow builders to enable or disable Autofill at the Application, Workflow, or Step level
• Use multiple input files or linked records as Autofill sources

FAQs

1. How do I enable the Autofill Beta version? Do I need to sign up?

No. Autofill Beta is available to all customers—no sign-up is required. Risk Cloud admins can enable it by going to Admin → Integrations → Spark AI Integration Card, selecting Configure, and toggling on Autofill Beta.

2. What file formats can Autofill process to populate field values?

• Documents: PDF, CSV, JSON, TXT, DOCX, PPTX, XLSX
• Media: PNG, JPG, JPEG, GIF, WEBP, WAV, MP3, MPEG, MPGA, M4A, OGG, WEBM
• Max file size: 100MB

3. What fields will Autofill populate?

Autofill fills blank, visible, and editable fields in the current Workflow Step. For newly revealed conditional fields, click Autofill Empty Fields to populate them.

• Fields Directly Supported: Text, Text Area, Radio, Checkbox, Select, Multi-Select, Number, Date Picker
• Fields Indirectly Supported (saved but not labeled by "Autofilled by Spark AI"): Standard & Date Calculations, Text Concatenations, Linked Record Counts 
• Fields Not Supported: User, Attachment, E-signature, Due Date

4. What happens if a user edits the field autofilled by Spark AI?

When a user edits a field previously autofilled by Spark AI, their new input replaces the AI-generated value, and the “Autofilled by Spark AI” label is removed.

5. Who can run Autofill? Can I limit who has access to use the Autofill?

Only the record assignee—whether an internal or external user—can run Autofill. Other users can view the results and field audit history generated by Spark AI Autofill but cannot rerun or modify it. 

In the Beta version, access control isn’t yet configurable, so Autofill is automatically available to any record assignee.

6. Is Autofill available for external users?

Yes - autofill is available for external users if the user is the record assignee. However, Autofill is not available via the public URL.

7. What prompt does Spark AI Autofill use?

Spark AI uses a built-in “base prompt” for every field it fills. This prompt gives the AI enough context about what the field means and where it belongs, so it can generate accurate and relevant suggestions.

In plain terms, the base prompt says something like:

“This is the [field name] field, labeled as [label name], in the [workflow name] workflow. Here’s the tooltip: [tooltip text]. Use the exact information from the file or text provided. Don’t make things up.”

Depending on the type of field, Spark AI also adds a few simple rules:

• Text / Text Area: write normal plain text — no formatting or symbols.
• Number: give a number (like 42 or 3.14). If it can’t find one, leave it blank.
• Date / Due Date: give a date in this format: YYYY-MM-DD HH:MM:SS. If unsure, leave it blank.

If the field includes Guidance Text, the AI also sees that and uses it to fine-tune its answer.
For example, if a guidance note says “Convert to percentage,” the AI will factor that in.

When both a prompt and a file are provided, Spark AI combines them — using the text prompt for direction and the file as supporting context — before generating answers.

8. Are prompts and files saved between sessions?
No - Prompt and/or file used in each Autofill session is not saved in the database. By refreshing the page, the session resets and the prompt and/or file will be cleared. 

9. Where is AI traceability shown?

• On Record: purple “Autofilled by Spark AI” label under each field
• Field History Log & Export: fields' executor is “Spark AI", showing prompt and file metadata

10. What models are used in Spark AI Autofill?
OpenAI Whisper-1 (for audio/video transcription) and GPT-4.1-mini (for text generation). For the latest models used by each Spark AI feature, please refer to the model version on the Spark AI Integration Card.

11. How is my data protected when it’s sent to AI for Autofill?
Your data is protected through end-to-end encryption. When you run Autofill, only the supported fields on the record and prompt/file in the session are transmitted to OpenAI’s API for processing. The data is never used to train AI models, and if you use LogicGate’s managed key, OpenAI retains it for 30 days only for security monitoring before deletion. If you prefer full control, you can connect your own OpenAI API key, which lets your organization manage data handling and retention directly. For a deeper look at Spark AI’s privacy and governance safeguards, see the Help Center article: Convince Your Boss: Why Spark AI Is Safe, Secure, and Ready for Governance Approval.

12. Can I turn on Spark AI Autofill for only a specific Application, Workflow, or Step?
Not yet. In the Beta version, enabling Autofill in the Spark AI Integration Card activates it for all Applications and Workflows. The ability for admins to limit Autofill availability to specific Applications, Workflows, or Steps is on the roadmap.

13. What happens if my record has Hidden or Inactive fields?

Autofill skips any fields that are hidden by conditional logic or located within inactive sections. Once these fields become visible or active, click Autofill Empty Fields to automatically populate them with Spark AI suggestions.

14. Can I review or edit the AI prompt before each Autofill run?
Yes – before each Autofill run, you can review and modify the AI prompt and/or import file directly in the Configure Autofill modal. This lets you tailor instructions or add specific context before generating field suggestions.