SparkAgent is an AI agent that you can chat with in the SparkReceipt web app and mobile app.
SparkAgent can do the following tasks for you:
Find income and expense documents, or other documents you have uploaded
Edit document data e.g. merchant name, date
Add, remove and update document tags
Find anomalies, duplicates or other errors in your uploaded documents
SparkAgent asks for approval if it wants to update your data.
SparkAgent can also help you with using SparkReceipt by searching information from the knowledge base.
As we are actively developing SparkAgent, its responses might not always make sense. Please contact us at support@sparkreceipt.com or https://sparkreceipt.featurebase.app/.
How to use SparkAgent productively
Everything starts with a good prompt. A prompt is roughly the same as a chat message that you write to SparkAgent.
A good prompt covers one or more of the following details:
A well-defined date range that you want to explore
Details for filtering documents such as merchant name, line item names, tags or similar identifying information
Action you want to perform on the found documents
In the following sections we'll look into how these requirements translate to usage examples.
💬 Tips for better prompts
The quality of your results depends on how clearly you describe what you want SparkAgent to do. Here are a few tips for writing effective prompts:
Be specific — Include time ranges, keywords, or merchant names to help SparkAgent narrow down the right documents.
Example: Find restaurant receipts from June 2025
Combine filters — You can use multiple details in the same prompt.
Example: Find travel expenses from Q1 2025 with tag “business”
Include the action — Tell SparkAgent what to do once it finds the right documents.
Example: Find all June receipts and add the tag “reviewed”
Use natural language — You don’t need to remember special commands. SparkAgent understands everyday language.
Keep it focused — Try to stick to one type of task per prompt (for example, “find and tag” or “find and update”), so SparkAgent can complete it accurately.
Examples & use cases
Example 1: Find documents with SparkAgent
You can use any detail available on documents as search criteria. The agent understands both simple and complex queries. For example, you can use one date range or even combine several date ranges.
Find documents from August 2024 and 2025.
Find income documents from week 1 of 2025.
Find expenses from Q1 2024.
Example 2: Manage document tags with SparkAgent
SparkAgent excels in handling large amounts of documents in one go. For example, you might have tens of documents spread across multiple months that you want to tag with one or more tags.
In this case you would write to SparkAgent:
Find documents with travel related purchases from the last 30 days. Add the tag "sales trip" to the found documents.
If the search produced results, SparkAgent asks you for approval to add the tags.
Example 3: Find duplicates
You can ask SparkAgent to perform all kinds of work for you. SparkAgent can find data and use its own intelligence to fulfill your request.
To find duplicate documents, you can ask SparkAgent to look for a specific date range for duplicates:
Find duplicates from March 2025.
SparkAgent most likely gives you a summary of the duplicates and asks what you want to do with them. You can then prompt SparkAgent to e.g. tag the duplicates so that you can manually delete them from e.g. the Search view:
Add tag "duplicate" to the found duplicates.
Example 4: Auto accept all expenses with totals less than 20 USD
You can use SparkAgent to edit large batches of documents at once. In this example, we'd like to set the status to Reviewed of all expenses with totals less than 20 USD.
Find all expenses from June 2025 that are less than 20 USD and set status to Reviewed.
Features in detail
🏷 Managing document tags
You can also use SparkAgent to organize your documents by adding or removing tags in bulk.
This is especially helpful when you want to tag many documents at once across different months or categories.
Examples:
Find all Q2 2025 documents and add the tag Q2
Find all Q2 2025 documents and add the tags Q2 and Summer
Remove the tag Q2 from all documents
Remove the tag Q2 from all June documents
SparkAgent will always show you a preview and ask for your approval before making tag changes.
🧾 Updating document details
SparkAgent can also update certain fields in your existing documents.
You can use this to fix or complete missing data such as descriptions, dates, or kinds.
Examples:
Find all restaurant expenses from June without attendee information in details. Add to description “Joel and Frank”. Tag these documents with “Franks Business Ltd”.
Find all June receipts and add a brief description of the merchant to the details field.
Find all utility receipts from June and change their kind to invoice.
Make sure all June receipts have “receipt” as the document kind.
Fields that SparkAgent can update
Name – Updates the document name (merchant or payer).
Description – Updates the document’s description or details.
Document date – Updates the date on the document (format: YYYY-MM-DD).
Due date – Updates the due date on invoice-type documents.
Billing kind – Changes between receipt and invoice.
Review status – Can be set to pending, rejected, or reviewed.
Reference number – Updates reference or invoice number.
Payment method – Updates the payment method linked to the document.
💡 Ask SparkAgent for help
You can also ask SparkAgent general support questions. It can find and summarize information from the SparkReceipt knowledge base to help you right inside the chat.
Examples:
How to add a document?
How does the bank statement feature work?
SparkAgent's known limitations – October 2025
SparkAgent is in active development. We are constantly making it better based on your feedback.
Here is a list of limitations that restrict what you can do with SparkAgent:
Document categories cannot be updated with SparkAgent
Line items cannot be updated with SparkAgent
Line item categories cannot be updated with SparkAgent
Search is limited to 1000 documents at once – if you have more than 1000 documents, ask to perform search in smaller batches e.g. month by month
Bank statement extractor cannot be triggered by SparkAgent
Documents cannot be opened yet from the SparkAgent chat
Documents cannot be uploaded from the SparkAgent chat
We are actively working on these to make SparkAgent better.