1
00:00:01,000 --> 00:00:04,970
An AI agent can choose next steps, but its authority needs

2
00:00:05,000 --> 00:00:06,490
clear boundaries first.

3
00:00:07,160 --> 00:00:10,650
A fictional company wants two software packages connected.

4
00:00:10,840 --> 00:00:14,890
A project assistant may look up customer details, identify missing

5
00:00:14,920 --> 00:00:17,730
information and prepare a follow-up question.

6
00:00:17,960 --> 00:00:21,690
It may not commit to a quotation or send a customer message.

7
00:00:21,960 --> 00:00:25,850
Describe the information the agent receives, the tools it may choose

8
00:00:26,120 --> 00:00:27,930
and when its work is complete.

9
00:00:28,200 --> 00:00:31,850
Here, a finished draft is the endpoint, rather than a project

10
00:00:31,880 --> 00:00:34,135
sold without human involvement.

11
00:00:34,160 --> 00:00:37,655
The application checks every requested system action.

12
00:00:37,680 --> 00:00:41,770
Telling a prompt not to send messages is not access control.

13
00:00:41,800 --> 00:00:46,410
Give the integration only the permissions required for this particular task.

14
00:00:46,440 --> 00:00:49,975
If the sequence is always fixed, an ordinary workflow may be

15
00:00:50,000 --> 00:00:51,050
simpler.

16
00:00:51,080 --> 00:00:54,890
An agent becomes worth investigating when the next step genuinely

17
00:00:55,080 --> 00:00:58,170
depends on information found during the task.

18
00:00:58,200 --> 00:01:01,530
If a source is missing or an instruction requests a forbidden

19
00:01:01,560 --> 00:01:05,050
action, the agent stops or asks for review.

20
00:01:05,320 --> 00:01:09,930
Bound time, cost and action counts too, so an unclear assignment

21
00:01:10,040 --> 00:01:11,450
cannot run indefinitely.

22
00:01:12,040 --> 00:01:16,170
Try normal requests, missing information and messages that attempt to

23
00:01:16,200 --> 00:01:17,490
change the task.

24
00:01:17,720 --> 00:01:21,290
Check that forbidden actions are blocked and that proposals reach the

25
00:01:21,320 --> 00:01:23,930
designated reviewer for an explicit decision.

26
00:01:24,520 --> 00:01:28,810
Choose one task, representative examples and a responsible staff

27
00:01:28,840 --> 00:01:29,290
member.

28
00:01:29,800 --> 00:01:35,530
Assess the output and actions taken before adding tools or granting wider authority.

29
00:01:35,880 --> 00:01:41,130
Read OXIAE’s guide to building AI agents with explicit permissions,

30
00:01:41,160 --> 00:01:43,290
tests and human oversight.