WEBVTT

00:01.080 --> 00:07.000
So here is the guardrails code base where I have defined this is our topic as a colon file.

00:07.360 --> 00:14.240
I have identified or defined few canonical forms like define user, ask about cooking as a canonical

00:14.240 --> 00:18.280
form, and the user utterances here are how can I cook pasta?

00:18.280 --> 00:20.440
Or what is the recipe for cooking?

00:20.760 --> 00:25.720
There are also other user ask about criminal activity as a canonical form.

00:25.720 --> 00:27.160
And how can I rob a bank?

00:27.320 --> 00:28.960
How can I steal from a shop?

00:29.120 --> 00:30.680
Are the user utterances?

00:31.560 --> 00:35.360
Then here is the bot express refuse and the utterances.

00:35.720 --> 00:37.080
I am an assistant.

00:37.240 --> 00:39.320
I cannot answer that question.

00:39.800 --> 00:45.440
There's one more flow here that identifies cooking flow, and whenever the user would ask about cooking

00:45.440 --> 00:48.320
related queries, it will match this flow.

00:48.680 --> 00:56.840
And the bot would actually invoke bot express the refuse, which would invoke the I am an assistant.

00:57.800 --> 01:00.760
I cannot answer that question as bot utterance.

01:01.000 --> 01:03.160
Now let's go and execute the question.

01:03.160 --> 01:04.800
How can I cook dinner?

01:05.040 --> 01:08.320
I have got a chance to invoke the info logs.

01:08.560 --> 01:14.120
So when I execute the flow, you'll see a lot of info logs on the terminal.

01:14.320 --> 01:15.080
There you go.

01:15.920 --> 01:17.320
I am an assistant.

01:17.320 --> 01:19.280
I cannot answer that question.

01:19.640 --> 01:24.970
So cooking flow is identified and it responded saying I could not answer that question.

01:25.170 --> 01:32.130
Here's the cooking flow user asked about cooking, which is on line 16, and the bot refused to answer

01:32.130 --> 01:33.050
the question.

01:33.330 --> 01:40.530
It's a very simple flow, and this is a negative scenario in terms of stopping or preventing the user.

01:40.930 --> 01:42.690
Now let's ask another question.

01:42.690 --> 01:50.690
How can I play ping pong before I go into executing this flow, which is not identified at the file

01:50.970 --> 01:52.210
I want to specify.

01:52.250 --> 01:55.290
I want to go into the details of config.yml.

01:55.570 --> 02:03.370
This is the models we are specifying open type as main engine is an open AI and the model is DaVinci.

02:03.730 --> 02:10.450
Now I have specified a sample conversation here, which it says how the bot and the user conversation

02:10.450 --> 02:18.210
would go in this case here user express greeting as hello there and what would express greeting hello,

02:18.330 --> 02:19.570
how can I assist you?

02:19.730 --> 02:22.610
Then user would ask what can I what can you do for me?

02:22.730 --> 02:31.090
It specifies that as a capability in the bot respond about the capability saying I am an AI assistance.

02:31.090 --> 02:33.770
I can help you with a wide range of tasks.

02:34.130 --> 02:36.090
User tells asks.

02:36.250 --> 02:38.690
Tell me more about the history and so forth.

02:39.250 --> 02:41.050
That's a general question.

02:41.050 --> 02:44.010
And the bot responds to the general question.

02:44.290 --> 02:49.250
And this is how the conversation would go between user and the in the bot.