The AI Parser can read incoming lead emails and fill in custom fields such as annual revenue, a preferred move-in date, or a lease term. You describe what to look for, test it with a sample email, and enable it for the sources you choose.
Before you start
Sign in as an account administrator to configure and test AI extraction.
Create the custom field under Settings → Custom Fields, in the Leads section, and make it available in the pipeline you want to use. Choose the field type that matches the information you need. See Custom Fields for setup instructions.
Use a source that receives new lead emails. This feature works during email lead import; it does not automatically fill existing leads, spreadsheet imports, or every email in an ongoing conversation.
1. Open the source you want to configure
Open the relevant pipeline’s settings, select Sources, and click a source. To create one, click Add Source, enter its name, and click Next.
Configure the source that will actually receive the lead email. Enabling a field on one source does not enable it on the other sources in that pipeline.
Open Sources in the pipeline settings, then select the source you want to configure.
2. Confirm automatic email import is enabled
In Configure Auto-Import, select Automatically import leads via email. Copy the source’s Forwarding Address and configure your website or lead provider to send lead emails to it.
Use the forwarding address from your own account, not the example shown in the screenshot. AI extraction cannot create a lead from an email if automatic import is turned off for that source.
Enable automatic import and send lead emails to this source’s forwarding address.
Click Next, review the Consent Documentation step, and continue to Configure AI Parser. Text-message consent is a separate setting; enabling AI extraction does not grant consent to text a lead.
3. Choose a field and write its instructions
Search for your custom field, or expand Disabled Fields to find it. Turn its switch on to open the instructions. This prepares the configuration; you will activate it after a successful test.
In AI Extraction Instructions, describe exactly which value to extract and how to format it. Instructions can contain up to 500 characters. For example, for an Annual Revenue currency field:
Extract the annual revenue explicitly stated in the email, in dollars. Return only the number without currency symbols or commas. If it is missing, return Not found.
Be specific about the information to extract, its format, and what to do when it is missing.
Use the same approach for other fields:
Move-in date: “Extract the preferred move-in date. Return only the date in MM/DD/YYYY format. If no date is stated, return Not found.”
Lease term: “Extract the requested lease term in months. Return only the number. If no term is stated, return Not found.”
Multiple choice: Ask for one of the field’s configured options, using its exact wording. AI extraction cannot select an option marked as restricted.
Shared field, shared instructions: The on/off setting is specific to each source, but the extraction instructions belong to the custom field. Editing those instructions also affects other sources where that same field is enabled. Use separate custom fields if different sources need different instructions.
4. Test with a sample email
Click Test beside the instructions. In the test window, paste a realistic email body into Sample Email Text, then click Test again. You can also use the quick examples to get started.
For the Annual Revenue example, try:
Hi, I’m Jordan Casey. I’d like to learn about your property management services.
Annual revenue: $42,500.
Monthly rent collected: $8,000.
Please contact me at jordan.casey@example.com.
The expected extracted value is 42500, rather than the monthly amount of 8000.
Check the extracted value before clicking Save.
A preview test does not create a lead or populate an existing lead’s custom field. It checks your draft instructions against the sample text.
You can run three tests per custom field, per user, in a one-hour window. Unsuccessful extraction attempts count toward this limit. Choose representative samples, including emails with competing amounts or missing information. If you reach the limit, wait for an earlier attempt to fall outside the one-hour window.
5. Save and confirm the field is enabled
When the test succeeds and the value is correct, click Save in the test window. Save stores the instructions and enables extraction for the selected source. The field appears under Enabled Fields. Click Finish to close setup.
Enabled Fields shows which custom fields will be extracted for this source.
Finish alone does not save untested instructions. If you change the instructions, test them again before saving. If a successful preview has expired, run a new test.
Repeat setup for each additional source you want to use. Newly created sources start with AI extraction off.
A pipeline can have up to 15 AI-enabled custom fields across its sources.
To stop extracting a field for one source, turn its switch off in that source’s AI Parser settings. This does not remove values already saved on leads.
6. Check your first imported lead
Send a new test lead email through the source’s actual forwarding address. Include the lead’s contact details and a clearly stated custom-field value. After normal email processing, open the new lead and check:
The lead is in the expected pipeline and source.
The custom field contains the expected value and uses the correct format.
Other amounts or dates in the email were not mistaken for the requested value.
Enabling extraction does not backfill older leads. If the AI cannot find a valid value, it does not write an AI value for that field; the rest of the lead can still import. Review the first few imports and refine the instructions if needed.
Troubleshooting
The field is not listed: Confirm it is a lead custom field and is available in this pipeline.
The Test button is disabled: Enter extraction instructions first, and enter sample email text in the test window.
No value is found: Confirm the email actually contains the requested information. Missing information is an expected result; do not ask the AI to guess.
The value is incorrect: Distinguish annual from monthly amounts, specify the date or number format, or name the exact label in your provider’s emails. Update the instructions and test again.
A test succeeds but no lead appears: Check automatic import and the forwarding address. A preview tests extraction only; it does not test email delivery or create a lead.
A lead imports without the expected AI value: Confirm the field is enabled on that exact source, check the original email body, and test that text against your instructions. Field type and multiple-choice restrictions can prevent a value from being saved.
Need help? Contact support@leadsimple.com with the pipeline, source, custom field, an example email, and the value you expected.