Skip to main content

Workflows and examples

Preview an audience for free​

Preview founders at US companies with 11–50 employees. Show the match count and whether it is exact or estimated. Do not spend credits.

The agent reads get_account and get_search_filters, then calls preview_leads and get_search_count with the chosen filters. Contact details stay masked; no credits are deducted.

Find leads, save a list, and export​

Find up to 25 US founders with verified emails. Save them to a new list called US founders, keep every returned field, and export the list as CSV.

  1. Check the connected account, permissions, and subscription-credit balance with get_account.
  2. Discover filters with get_search_filters; preview the audience for free.
  3. Call search_leads with a new request_id, the agreed filters, and max_results: 25.
  4. Keep the returned job ID and poll get_search_job until the job is no longer pending. Do not call search_leads again to check progress.
  5. Read results with get_search_results. If the job failed or stopped with partial results, report its status and actual returned rows.
  6. Use create_list, then save_search_results for each result page. This preserves all fields in lead_data without another paid search.
  7. Use export_list_file with format: "csv". Give the user its download_url or download the file with your client tools. Dievio combines all pages automatically.

The search uses up to 25 subscription credits at one credit per returned lead. Reading, saving, and exporting these results adds no credit charge.

Take the results from my latest completed search, create September outreach, and save every contact with all fields. Don't run a new search.

Use list_search_jobs to identify the job, check it with get_search_job, create the list, and paginate save_search_results using the existing job ID.

Work with your team​

Show the lists and available credits in my connected team account, then preview marketing directors in the UK for free.

Start with get_account, get_team, and list_lists. If the connected account is wrong, reconnect with the intended team. The agent cannot switch ownership by supplying another user's ID to a tool.

Enrich LinkedIn profiles​

Find work emails for these LinkedIn profile URLs. Return at most 20 matches, with no phone lookup.

Use lookup_linkedin with the supplied URLs, maxResults: 20, includeWorkEmails: true, includePersonalEmails: false, and includePhones: false. This is a paid operation for returned matches; previewing an audience is a separate free workflow.

Download a file without opening Dievio​

Export my US founders list as a downloadable CSV, with email, name, company, and company website as separate columns.

Find the list with list_lists, inspect its saved fields if needed, then call:

{
  "list_id": "<list UUID returned by Dievio>",
  "format": "csv",
  "columns": ["email", "full_name", "company_name", "lead_data.company_website"]
}

Use these arguments with export_list_file. A requested path missing from a contact produces an empty cell. Use actual field paths from the saved contact payload. Omit columns for all available data, or choose json / jsonl to preserve nested structures. The result contains a ready-to-download link; the agent does not need to paste contacts into the chat or assemble pages. Download within 15 minutes; ask Dievio for a new link if it expires. This is free and works in the selected personal or team account. The site is needed only for initial connection approval.

Pagination for inline results​

Result, saved-contact, and inline export_list tools use offset / limit. Start at offset 0, advance by the number of returned rows, and continue while has_more is true. save_search_results also returns next_offset; use it for the next page.

Each CSV page includes a header row. When combining pages, keep the header from the first page only. The lead_data column contains additional fields encoded as JSON. CSV cells are escaped to avoid executing spreadsheet formulas.

list_lists and list_search_jobs return arrays. Request successive offsets until a page contains fewer items than the requested limit. A single page is not necessarily the complete account history.

Retries without duplicate work​

For mutations, generate one request_id UUID per intended operation and keep it alongside the arguments. Retry with the same ID and identical arguments after a lost response. Do not use a new ID to bypass an uncertain or in-progress result.

For a search, reconcile against the returned job ID or recent jobs before starting anything new. For a list mutation, read the list again to see whether it was applied. If the outcome remains uncertain, resolve it before creating another operation. Never blindly repeat a paid lookup with a new request ID.

Only queued searches can be cancelled. Running searches cannot safely be cancelled while the worker is processing and charging for results.