Build your first Agent
Describe the job in plain language and Praxivara's AI builder plans, configures, and builds a custom Agent that works on its own.
Last updated July 14, 2026
Where the Assistant helps you in the moment, an Agent is a custom AI worker you set up once and let run on its own — on a schedule or whenever something happens. Building one is a conversation, not a configuration project.
Here's how to create your first Agent.
Build by describing the job
Start a new Agent and describe the job in your own words — for example, "Every morning, summarize new leads from my inbox and post them to Slack."
The builder asks clarifying questions to get the details right — which inbox, which channel, what to include.
It shows a visual Blueprint — a plan of what your Agent will do. Check it over, ask for changes, then let it build.
Run a test to see the Agent in action, then turn it on.
What you can configure
Your Agent is made of a few building blocks. The builder sets sensible defaults, and you can adjust any of them:
- Tools. The connected apps and abilities your Agent can use.
- Skills. Reusable playbooks for how to handle specific jobs.
- Files & Knowledge. Reference material the Agent can draw on.
- Memory. What it should remember between runs.
- Secrets. Credentials it needs, stored securely.
- Channels. Where it can reach you.
Put it on autopilot
Under Automations, choose how your Agent runs. Use a Schedule for regular timing, or a Trigger to react to events like a new email, a webhook, or activity in a connected app.
See what it does
Once your Agent is running, you can follow its work:
- Deliveries — the outputs your Agent produces.
- Activity — a history of every run.
- Errors — anything that failed, so you can fix it fast.