How to Automate Lead Research With an AI Assistant

Automate lead research by giving an AI assistant access to a lead-data API, not just asking it to describe prospects. For a task such as finding dentists in Chicago, have it propose the search filters, request a quote and wait for your budget approval before starting collection with optional email verification. It should then retrieve the results, preserve verification outcomes, check for duplicates and return a CSV for review. Keep filter choices, spending approval and the first outreach send under human control. An assistant without connected execution tools can draft this workflow, but cannot run the research or retrieve the file itself.

AI Lead Research: A Chat Assistant Versus an Assistant With API Tools

An AI assistant automates lead research only when it has tools that can query a data service and retrieve actual records. A chat assistant without those tools can draft search criteria or explain a workflow, but it cannot execute a collection job or return its results. For example, “Find dentists in Chicago and return a file with verified emails” requires access to a collection service, an email verification option and a way to retrieve the output. Require retrieved records rather than a table drafted in chat.

• Chat-only output: suggested values such as “dentists” for the Google Maps search term and “Chicago” for the place, plus instructions for someone else to run.
• Tool-enabled execution: an authenticated connection to the Evascrape API, with access to operations such as POST /jobs and GET /jobs/{id}/results.
• Evidence of execution: a job ID, a status returned by the service and retrieved records, rather than an unsupported statement that the research is complete.

Before assigning the task, ask which API tools the assistant can actually call in the current session. For Evascrape, that connection uses a bearer key created under API keys in the account. Require the assistant to distinguish actions it performed from actions it merely recommends. If no tool connection is available, request a research plan, not a collected lead file.

Google Maps Lead Research: Choose Dentists in Chicago and Approve the Filters

To research dentists in Chicago, have the assistant prepare a Google Maps search with “dentist” as the search term and “Chicago, Illinois” as the place, then ask you to approve the filters before starting collection. In Evascrape’s Google Maps job form, the inputs are a search term and a place, not a people-search URL. Tell the assistant whether you want general dental practices or a narrower audience, such as orthodontic practices, rather than letting it choose that distinction for you.

• Search term: use “dentist” for the proposed search. If your intended audience is a dental specialty, approve a replacement term before proceeding.
• Place: enter “Chicago, Illinois.” State whether nearby suburbs belong in the audience instead of letting the assistant silently broaden the location.
• Minimum rating: leave it unset unless your prospect criteria explicitly require a rating threshold.
• Categories and language: review these optional settings with the assistant. Leave them unset unless you have a specific reason to apply them.
• Permanently closed places: ask the assistant to select the option to skip them if you only want operating practices.

Require the assistant to return the proposed settings for approval without starting the job. For example: “Search for dentist in Chicago, Illinois; skip permanently closed places; leave language, minimum rating and categories unset.” Review that summary against the businesses you actually want to contact. Approve it or correct the search term, place and optional filters before the assistant proceeds.

Lead Research Budget Approval: Quote the Job and Add Optional Email Verification

Before collecting the Chicago dentist list, have the assistant request a free price quote with POST /quote and wait for your budget approval. The quote should reflect the approved Google Maps search, the requested record count and whether you want optional real-time email verification. For this example, request up to 500 business records with verification enabled, then ask the assistant to show the collection charge and verification add-on separately. Google Maps collects up to 500 places per search term. A larger total needs more search terms, and the Google Maps form states how many are needed before any money is spent. Require approval of those additional terms before quoting a larger job. Do not authorize POST /jobs until you have reviewed the quote.

• Check the requested count: jobs require at least 100 records, allow up to 50,000 and accept counts in multiples of 100. Keep this search at a ceiling of 500 unless you approve additional search terms.
• Confirm the collection rate: Google Maps business records cost 12 US dollars per 1,000 records.
• Approve verification explicitly: real-time email verification costs an additional 5 US dollars per 1,000. It is never included in the base collection rate.
• Check the billing exception: delivered records with no email address are billed at 6 US dollars per 1,000 instead of the full collection rate.
• Review available funds: the assistant can use GET /balance to retrieve your prepaid balance in US dollars. A confirmed account starts with 2 US dollars of credit.
• Set the approval boundary: tell the assistant to return for approval if it changes the requested count, approved filters or verification choice.

Your approval covers the amount reserved when the job starts, based on the requested record count. That count is a ceiling, not a delivery promise, and the undelivered portion is refunded automatically when the job finishes. Ask the assistant to present the quote as a reservation rather than a guaranteed final bill. Billing uses a single prepaid balance, with no subscription or per-seat fee, and credits do not expire.

Automate Lead Collection With the Evascrape API: Start, Check Status and Retrieve Results

An assistant with API access runs the approved lead research task by calling POST /jobs, checking progress with GET /jobs/{id}, and retrieving the completed records with GET /jobs/{id}/results. Use https://app.evascrape.com/api/v1 as the base URL and authenticate with a bearer key created under API keys in the account. For the dentists-in-Chicago task, carry forward the approved search term, place, record count and email verification selection without changing them. Use the request structure documented at https://evascrape.com/docs/openapi.json rather than guessing payload field names.

• Start the approved job with POST /jobs. Save its job identifier so subsequent status checks and result requests refer to that same job, rather than starting another collection.
• Check GET /jobs/{id}, replacing {id} with the saved identifier. The status values are queued, running, succeeded, failed and no_results. Keep queued and running jobs marked as pending, not complete.
• When the status is succeeded, call GET /jobs/{id}/results to retrieve JSON for further processing or CSV for the file handoff. Results are also available as a CSV file in the dashboard.
• If the status is failed or no_results, report that outcome instead of presenting an empty response as a completed lead list. Do not silently start a replacement job.

Keep status checks within the API limit of 120 requests per minute per key, and job starts within 20 per minute. Alternatively, supply an optional callback_url to receive the finished job, signed with HMAC-SHA256. Require the assistant to distinguish between a submitted job, a finished job and retrieved results in its updates. For example, say that the dentists-in-Chicago CSV has been retrieved only after the results request returns the file.

Email Verification and CSV Export: Return Good, Risky and Bad Email Files

With the optional real-time email verification add-on, Evascrape splits the export into good, risky and bad email files. For the dentists-in-Chicago task, ask the assistant to return those files separately and identify each by its verification label, rather than combine them into an undifferentiated contact list. If the add-on was not selected for this Google Maps job, require the assistant to state that explicitly.

• Good email file: Present this as the export labeled good by the verification process. Keep that label in the handoff without adding delivery claims.
• Risky email file: Return it separately so the user can distinguish these records from the good file. Do not silently move its addresses into the good file to produce a longer list.
• Bad email file: Keep it separate and clearly labeled. Do not merge it into a file presented as ready for outreach.
• CSV delivery: Direct the user to the CSV export in the dashboard. For an API-driven handoff, use GET /jobs/{id}/results, which returns records as JSON or CSV. Preserve the export’s verification labels when presenting the output.

Make the handoff explicit: “The dentists-in-Chicago export is separated into good, risky and bad email files.” Identify the corresponding files so the user does not have to infer their purpose from attachment order. If verification was not selected, state that plainly instead of inventing verification labels for the collected addresses.

Lead Research Automation Checks: Search Filters, Email Verification and Duplicates

For the dentists-in-Chicago task, ask the assistant to review the approved filters, verification selection and possible duplicate records before handing over the file. Do not authorize it to silently broaden the search or relabel email results. The requested record count is a ceiling, and fewer matching records means a smaller file and a smaller bill.

• Search filters: In the Google Maps job form, inspect the search term, place, minimum rating and categories against the approved request. Require approval before changing any setting or rerunning the job. Keep the limit of 500 places per search term in view when reviewing a smaller-than-requested file.
• Email verification: Check whether the optional verification add-on was selected. If it was not, state that in the handoff. When verification was selected, keep risky and bad results separate from good results rather than merging the files into an undifferentiated outreach list.
• Duplicate review: Before combining exports, inspect the available columns and compare records against earlier lists. Use business names, street addresses, domains and email addresses only where those fields are present. Flag possible matches for review instead of silently deleting records.

Before running another collection, return the existing job status, approved settings and retrieved record count for review. Preserve the original export and document any proposed filter changes or duplicate-removal decisions. Wait for approval before starting another paid job.

Lead List Review Checklist: Search Match, Email Status, Duplicates, File Contents and Final Cost

Before using the assistant’s lead list, review the search match, email status, duplicates, file contents and final cost. Ask the assistant to return a review summary alongside the CSV, with exceptions flagged for your decision rather than silently discarded. For the dentists-in-Chicago task, inspect the available fields against the approved audience rather than assuming which columns the export contains.

• Search match: Compare the job’s search term, place and optional filters with your approved request. Where the export provides location and business details, review them for matches outside the intended city or business type. Flag ambiguous records for review.
• Email status: If you paid for verification, confirm that the delivery includes good, risky and bad email files. Keep risky and bad results separate from the reviewed outreach list. If verification was not purchased, require the assistant to state that clearly.
• Duplicates: Use the available fields to check for repeated records within the file and against your existing lead list. Review proposed removals before applying them. Ask the assistant to explain which records it removed and why.
• File contents: Open the CSV from the dashboard and inspect its headers and rows. Check that values remain aligned with their columns. Where an email column is present, ask the assistant to flag blank values. Retain the original export separately from any cleaned version.
• Final cost: Compare the completed job’s billing details with the approved quote and requested record count. Google Maps records cost 12 US dollars per 1,000, with optional verification charged on top at 5 US dollars per 1,000. Delivered records without an email address are billed at 6 US dollars per 1,000 instead of the full base rate. Confirm that the undelivered portion was refunded and that credits_charged plus credits_refunded equals the amount reserved.

Human Approval Before Outreach: Review the Audience, Message and First Send

Require human approval before any outreach. Keep permission to research separate from permission to contact people: instruct the assistant that collection approval does not authorize sending. Evascrape has no native CRM integration or browser extension. Import its CSV into your CRM or outreach tool for review, then approve the recipient selection and draft message there.

• Review the recipient list after importing the CSV. Check any proposed recipient role against the available information. Do not let the assistant address someone as a practice owner without supporting information.
• Review the subject and message body before sending. Remove invented familiarity and unsupported claims about the recipient or practice.
• Confirm the sender name, reply-to address and opt-out wording. Ask the person responsible for outreach to review the intended use of the data before approving the campaign.
• Approve the first send explicitly. Record the approved recipient selection, message version and send timing. Instruct the assistant to pause again if any of those change.

Give the assistant a clear instruction: “Prepare the campaign as a draft. Do not send or schedule it until I approve the audience, message and timing.” If sending tools are connected, require a separate confirmation before using them. For a file-only handoff, assign the CSV import, campaign review and launch to a named human reviewer.

Evascrape collects only publicly available business data, never private messages or login-protected content. It never asks for your LinkedIn or Apollo login and never sends requests from your own IP address. For compliance information and removal requests, use https://evascrape.com/compliance.