> ## 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.

# Find your next Deepline workflow

> Use /deepline-start to turn an unclear goal or recurring manual task into a concrete Deepline plan.

[Agent guide: CLI and SDK instructions →](/docs/templates)

You do not need to know a tool name or arrive with a technical plan.
**`/deepline-start` helps you work out what is worth doing and how to do it.**
Describe a task that takes too much time or an outcome you want. The skill asks
one question at a time, recommends a next step, and helps you choose a workflow
that fits your inputs and constraints.

## 1. Have your context ready

Bring one recent example of the work. It can be an account list, a CRM record,
an unfinished research task, or a description of the steps you repeat by hand.

Good starting points include:

* Finding the right accounts or people to contact.
* Filling gaps in contact or company data.
* Working out which accounts deserve attention first.
* Researching accounts or preparing evidence-based outreach.
* Turning a recurring manual process into an automation.

You can start before the exact result is clear. The interview helps define it.

## 2. Ask your agent

Paste this into your agent and describe your situation when it asks:

```deepline-prompt theme={null}
/deepline-start Help me find a useful Deepline workflow. Start with the recurring task or outcome that matters most to me. Ask one concise question at a time and recommend a next step. Use a recent example to understand my current process, available inputs, and what a useful result would look like. Find the right skill or workflow, explain the proposed plan and workspace, and wait for my confirmation before starting it.
```

The conversation produces a plan, not just a list of possibilities. The skill
checks which installed skills and current workflows fit the job. Source research
comes first when the evidence is unclear. Scoring leads when the goal is to define
or evaluate a score. A known execution task can go directly to the GTM skill.

Once you confirm the plan and workspace, the skill hands the brief to the chosen
skill and starts the agreed work. You do not need to copy the brief into a new
conversation or explain the task again.

## 3. Review the result

Before confirming the plan, check that it names:

* The outcome and how you will judge whether it is useful.
* The available inputs, sources, and unresolved questions.
* The selected skill or workflow and why it fits.
* The intended workspace, pilot scope, and applicable usage.
* Any external changes or recurring work that need approval.

If the proposal does not fit, continue the conversation:

```deepline-prompt theme={null}
/deepline-start Refine the proposed plan using what we have already discussed. Explain the most important unresolved decision and recommend an answer. Ask me one question at a time. Keep the agreed outcome, constraints, and workspace unless I change them. Once I confirm the revised plan, continue with the chosen skill using the existing brief.
```

## Check your understanding

Do you need to choose a provider or Play before starting this conversation?

> **Show the answer**
>
> No. Bring the task or outcome you care about. The skill helps define the work,
> inspect fitting options, and choose a concrete plan with you.

## Already know what you want?

Go directly to a task guide when the outcome is clear:

* [Find work emails](/docs/human/plays/email-waterfall)
* [Find mobile numbers](/docs/human/plays/task-find-mobile-phone)
* [Research accounts](/docs/human/recipes/research-workflow)
* [Score accounts](/docs/human/guides/account-scoring-workflows)
* [Build a Play or Engine](/docs/human/sdk-v2/plays)
* [Build an ads audience](/docs/human/ads-audiences)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.