Build your own Nalvin agent

Use Template

By Nalvin

Use Template

Description

Start with a blank canvas and configure your own Nalvin agent. Define the data sources it accesses, the instructions it follows, and where it delivers results. Custom agents are ideal when an existing template does not quite fit your workflow or team needs.

Supported Integrations

Templates are flexible by default. Nalvin automatically picks the right integrations based on what you’ve connected.

Asana

ClickUp

Outlook Email

Gmail

Monday

Trello

Notion

Email

Web

GitHub

Shortcut

Linear

Microsoft Teams

Confluence

BitBucket

Jira

Slack

How to best use this template

Below are a few tips & tricks from the Nalvin team to get the most out of this template.

Start with a clear job to be done

A good custom agent does one thing well. Start by writing a single sentence: This agent should do X. If you cannot finish that sentence clearly, narrow the scope before configuring sources and instructions.

Choose your sources deliberately

Only connect the tools that contain relevant signal. More sources is not always better—it can dilute the output. Start with 2–3 high-signal sources and expand from there.

Write precise instructions

The instructions are the most important part of a custom agent. Be explicit about what to include, what to exclude, and what format the output should follow.

Test before scheduling or routing to Slack or Teams

Run the agent manually first. Review the output, adjust the instructions, and rerun until it behaves consistently. Only automate once the output is reliable.

Iterate based on real use

Custom agents improve over time. Ask your team for feedback and revisit the instructions every few weeks to tighten the scope and improve signal quality.

Start with a clear job to be done

A good custom agent does one thing well. Start by writing a single sentence: This agent should do X. If you cannot finish that sentence clearly, narrow the scope before configuring sources and instructions.

Choose your sources deliberately

Only connect the tools that contain relevant signal. More sources is not always better—it can dilute the output. Start with 2–3 high-signal sources and expand from there.

Write precise instructions

The instructions are the most important part of a custom agent. Be explicit about what to include, what to exclude, and what format the output should follow.

Test before scheduling or routing to Slack or Teams

Run the agent manually first. Review the output, adjust the instructions, and rerun until it behaves consistently. Only automate once the output is reliable.

Iterate based on real use

Custom agents improve over time. Ask your team for feedback and revisit the instructions every few weeks to tighten the scope and improve signal quality.