WEBVTT

00:00:00.000 --> 00:00:02.500
PandaDoc already has a powerful API.

00:00:03.000 --> 00:00:06.100
What gets interesting is when an AI worker

00:00:06.100 --> 00:00:09.380
can actually use it inside the right client workflow.

00:00:11.600 --> 00:00:16.400
In Outloop, that work starts inside one client workspace,

00:00:16.840 --> 00:00:21.280
with the context and approved tools that belong to that client.

00:00:22.020 --> 00:00:25.300
A human adds the PandaDoc credential once.

00:00:25.300 --> 00:00:29.200
It stays in Outloop's host-side flow while the worker gets

00:00:29.200 --> 00:00:33.080
the approved operation it needs, not the raw key.

00:00:34.160 --> 00:00:40.660
Before the worker relies on it, Outloop verifies the provider operation against PandaDoc.

00:00:40.960 --> 00:00:46.180
Then, each call is recorded, including the successful response.

00:00:47.040 --> 00:00:52.600
In this sandbox workflow, the worker prepared a draft,

00:00:52.600 --> 00:00:58.520
checked its state and brought back the PDF for review, using only the operations we declared.

00:00:59.820 --> 00:01:02.380
PandaDoc provides the document API.

00:01:03.360 --> 00:01:08.200
Outloop gives the AI worker the client workspace, approved access,

00:01:08.200 --> 00:01:12.460
context and the runtime to use it within those boundaries.

00:01:13.260 --> 00:01:18.320
If you work with agencies using PandaDoc, I'd love to show them this workflow.

00:01:18.600 --> 00:01:22.860
Send them this video or connect us and we'll walk through it together.