> ## Documentation Index
> Fetch the complete documentation index at: https://deepline.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Tasks
> How to enrich contacts and companies programmatically — Deepline tasks are single-step waterfall lookups that try multiple data providers in cost order.
A Deepline task finds, verifies, or enriches one data point per call. Each task is a waterfall lookup across 97+ integrations that tries the cheapest provider first and stops at the first valid result -- a work email, a phone number, a LinkedIn URL. You describe what you want in plain language from your IDE or terminal, and Deepline picks the right providers in the right order. No UI configuration, no template selection -- just natural language prompts that chain multiple providers automatically.
## Start with a pilot
Use a task when you need one field added to a list. Pilot a few rows first.
```text theme={null}
/deepline-gtm Find verified work emails for leads.csv.
Run rows 0-4 first, write the results to leads-with-emails.csv, and summarize any rows that need review.
```
## What is waterfall enrichment?
Waterfall enrichment tries multiple data providers in sequence, cheapest first, and stops at the first valid result. You only pay for providers that actually ran. Single-provider enrichment commonly leaves coverage gaps. Waterfall enrichment improves match rates by querying multiple providers in sequence. Specific lift varies by ICP and region. In Deepline, every task uses waterfall enrichment automatically: the system selects which providers to try, in what order, based on your input data and the provider's coverage for that region or data type.
> "The best enrichment workflow tries free patterns first, then falls through to paid providers only when needed." — Deepline documentation
B2B data decays at roughly 22.5% per year ([HubSpot via Cognism](https://www.cognism.com/blog/data-decay)), so nearly a quarter of your contacts switch jobs every year. Waterfall enrichment compensates by cross-referencing multiple providers — stale data from one provider gets caught by fresh data from another.
## How does Deepline handle provider routing automatically?
Your prompt should describe the goal and constraints, not the exact provider sequence. Deepline handles provider selection, cost optimization, and fallback logic automatically across 97+ integrations. It looks at what data you have and routes accordingly. Most lookups resolve on the first provider for just 0.3 credits, and the cost-aware waterfall can resolve for as little as 0.1 credits when a free email pattern validates.
Since Deepline runs inside your IDE or terminal, an AI agent can chain enrichment steps as part of a larger automated pipeline without switching to a separate UI.
## Which tasks are available in Deepline?
Deepline currently offers 7 tasks across 97+ integrations:
| Task | What it does | Providers |
| ------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------- |
| [Find Work Email](/docs/plays/task-find-work-email) | Verified work emails via cost-aware waterfall (free patterns first, then paid finders) | LeadMagic patterns, Dropleads, Hunter, LeadMagic, Deepline Native, PeopleDataLabs |
| [Find Personal Email](/docs/plays/task-find-personal-email) | Personal emails for ads and direct outreach | Dropleads, LeadMagic, PeopleDataLabs |
| [Find Mobile Phone](/docs/plays/task-find-mobile-phone) | Mobile numbers with automatic region routing | enrich\_phone\_finder, Dropleads, LeadMagic |
| [Find LinkedIn Profile](/docs/plays/task-find-linkedin-profile) | Resolve LinkedIn URLs from names, emails, or companies | Google, Exa, Crustdata, Apollo, PeopleDataLabs, Apify |
| [Find Decision Makers](/docs/plays/task-find-decision-makers) | Find contacts at companies by role and seniority | Dropleads, Apollo, Icypeas, Prospeo, Crustdata |
| [Enrich Company](/docs/plays/task-enrich-company) | Full company profile from a domain or name | Apollo, Crustdata, PeopleDataLabs |
| [Email Waterfall with Validation](/docs/plays/task-email-waterfall-with-validation) | Find + validate emails in one pass, free patterns first | Free patterns + paid fallback + ZeroBounce |
## How do Deepline tasks work step by step?
Every Deepline task follows the same pattern: analyze input, try providers in cost order, return the first valid hit. Most lookups resolve on the first or second provider at 0.3 credits.
Deepline inspects what columns or fields you provide (name, email, LinkedIn
URL, domain) and selects the right waterfall variant.
The cheapest provider runs first. If it returns a valid result, you're done.
If not, the next provider fires.
You only pay for providers that actually ran. Most lookups resolve on the
first or second provider.
Always pilot with 3-5 rows before running a full file. Just say: "Run this on
the first 3 rows of leads.csv."
***
**Related pages:** [Plays Overview](/docs/plays/plays-overview) | [Find Work Email](/docs/plays/task-find-work-email) | [Email Waterfall with Validation](/docs/plays/task-email-waterfall-with-validation)
## Frequently Asked Questions
### What is the difference between a Task and a Play in Deepline?
A task is a single-step waterfall lookup that finds one data point (e.g., a work email or phone number). A play is a multi-step workflow that chains multiple tasks together with AI reasoning — finding decision makers at a company, enriching each with emails, and generating personalized outreach. Research on 1.8M sales opportunities shows single-threaded deals close at roughly 5%, while engaging 5+ stakeholders reaches roughly 30% ([Gong](https://www.gong.io/blog/data-shows-top-reps-dont-just-sell-they-orchestrate-with-ai)). Plays help you build those multi-threaded pipelines. Use tasks when you need a specific data point; use plays when you need a complete GTM motion.
### How much does a typical Deepline task cost?
Most tasks resolve on the first provider for 0.3-0.4 credits (\$0.024-\$0.032). The cost-aware email waterfall can resolve for as little as 0.1 credits when a free email pattern validates. Worst case across a full provider chain is 7.2 credits, but this is rare.
### Can I run Deepline tasks in bulk from a CSV?
Yes. type `/deepline-gtm find work emails for leads.csv` and Deepline reads your column headers, picks the right waterfall variant, and enriches every row. You can also use the `deepline enrich` CLI command with `--input` and `--output` flags for programmatic bulk runs. Always pilot with 3-5 rows first.
### Why use Deepline for data enrichment?
Deepline runs inside your IDE or terminal, accepts natural language prompts, and chains multiple providers automatically -- no UI configuration or template selection. Waterfall routing analyzes your input data, selects the right providers, and tries them in cost order for better coverage than single-provider lookups because it queries multiple providers in sequence. Since everything runs in your coding environment, enrichment plugs directly into scripts, CI/CD pipelines, and agent workflows.
### Do I need to know which data providers to use?
No. Deepline selects providers automatically based on your input data. If you have a name and company domain, it picks the name-and-company waterfall. If you have a LinkedIn URL, it picks the LinkedIn-optimized chain. You describe the goal; Deepline handles provider selection.
### How is Deepline different from Clay for enrichment?
Clay is a spreadsheet UI where you pick templates, configure columns, and wire providers manually. Each enrichment type (work email, phone, LinkedIn) needs a separate template. Deepline works inside your IDE -- describe what you need in plain English and it picks the right providers, chains them in a waterfall, and handles validation automatically. Results feed directly into your code, scripts, and CI/CD pipelines without copy-pasting between a browser UI and your terminal.