​

Plays

Deepline plays are multi-step GTM workflows that chain enrichment, AI research, and outreach into complete go-to-market motions. Tasks find a single data point. Plays chain multiple enrichment steps with AI reasoning and structured output into full workflows. Each play can be triggered with a single natural language prompt through Claude Code or Codex.

​

What is a GTM workflow?

A GTM (go-to-market) workflow is an automated sequence of data enrichment, AI research, and outreach steps that replaces manual sales and marketing work. In Deepline, a “play” is a GTM workflow you run by describing what you want in plain language. No template configuration required. For example, “build a prospect list of VP Engineering contacts at Series B fintech companies with verified emails” chains company search, decision maker lookup, email waterfall, and validation into one run. According to industry benchmarks, waterfall enrichment across multiple providers can improve email coverage by 20-40% compared to any single provider (Instantly). Deepline’s waterfall across 30+ providers keeps credits low by stopping at the first valid result.

​

How does the 2-pass AI research pattern work?

Many Deepline plays use a 2-pass AI research approach: gather data first, synthesize second. Separating the two passes means each step can be validated independently, and you can inspect raw search results before synthesis runs.

The GTM engineering principle: “describe the goal and constraints, not the exact provider sequence.” Deepline selects the optimal provider path automatically.

1. Pass 1 — Search: Web search (Claude Code’s native web agent, Exa, or Parallel AI depending on the use case) gathers raw data from the web
2. Pass 2 — Synthesize: call_ai processes the raw data into structured output

​

Which plays are available in Deepline?

Deepline currently offers 9 plays, each automating a complete GTM workflow. Each play is triggered with a natural-language prompt that handles step sequencing, provider selection, and data flow automatically — no UI configuration or template setup required.

​

What is the difference between a Play and a Task?

B2B buyers are 57% through the purchase decision before engaging a sales rep (CEB/Gartner), so plays help you engage earlier with richer context. Here’s how tasks and plays differ:

• Tasks are atomic: one input, one output, one waterfall. Use them when you need a specific data point (e.g., a work email, a phone number).
• Plays are workflows: multiple steps, AI reasoning, structured output. Use them when you need a complete GTM motion (e.g., build a prospect list, generate personalized outreach).

You can always break a play into its component tasks if you want more control over individual steps.

---

Related pages: Tasks Overview | Build Prospect List | Personalize Outreach

​

Frequently Asked Questions

​

How do I run a Deepline play?

Tell Claude Code or Codex what you want in plain language. For example: “Build a prospect list of VP Engineering contacts at Series B fintech companies with verified emails.” Deepline chains the necessary tasks — company search, decision maker lookup, email waterfall — into a single automated workflow. You can also use the Deepline CLI for programmatic access.

​

What is the difference between a Task and a Play in Deepline?

A task finds a single data point (e.g., a work email) using a waterfall of data providers. A play chains multiple tasks with AI reasoning into a complete workflow. The “Build Prospect List” play, for example, chains company enrichment, decision maker search, email waterfall, and validation into one run. Tasks are building blocks; plays are complete workflows.

​

Can I customize which steps a play runs?

Yes. You can break any play into its component tasks and run them individually with custom parameters. You can also modify play behavior by describing constraints in your prompt — for example, “build a prospect list but skip phone number lookup” or “only search for contacts in the US.”

​

Why run GTM workflows inside the IDE?

Deepline plays run inside Claude Code or Codex — the same environment where you write code, manage data, and build automations. Describe the workflow in plain language and Deepline handles step sequencing, provider selection, and data flow. No context-switching to a separate UI, no template configuration, no column wiring. Your ICP criteria, scoring logic, and output format all live in the prompt, so they version-control naturally alongside your code.

​

How much does a Deepline play cost?

Play costs depend on the component tasks and the number of rows processed. Each task within a play follows waterfall pricing (cheapest provider first, stop at first valid result). A typical “Build Prospect List” play might cost 2-5 credits per company (company enrichment + decision maker search + email waterfall for each contact). Always pilot with 3-5 rows to estimate total cost before running a full list.

​

How do Deepline plays compare to Clay tables?

Clay tables are spreadsheet-based workflows where you manually add columns, pick providers, and wire up enrichment steps. Each step is a separate template. Deepline plays chain multiple providers and AI passes into a single natural-language prompt — describe the outcome you want and the play runs the full pipeline. Deepline plays use a 2-pass approach for research — search first, synthesize second — so the AI only works with verified data. Plays run inside your IDE or terminal, so the results feed directly into your code, scripts, or CI/CD pipelines without switching between a browser UI and your codebase.