WEBVTT 00:00.120 --> 00:06.480 And so this brings us to the final lecture of the final day of the final week, the end of the journey, 00:06.480 --> 00:08.520 and what a journey it has been. 00:08.520 --> 00:12.400 What an adventure in the world of autonomous AI agents. 00:12.720 --> 00:17.480 It seems almost impossible to remember that at the beginning of this, six weeks ago, we were working 00:17.480 --> 00:18.280 on the foundations. 00:18.280 --> 00:26.760 We were making direct calls to OpenAI like ChatGPT completions and passing in messages, and we were 00:26.760 --> 00:32.240 building like the JSON stuff that describes a function all of that long JSON. 00:32.240 --> 00:36.480 And then we were taking back, we were having the handle tool calls that looks to see what the finish 00:36.480 --> 00:41.880 reason is, and having a silly while loop to reprocess what's coming back from the models, all that 00:41.880 --> 00:42.280 stuff. 00:42.680 --> 00:46.120 But it gave that foundation so that you really understood how this fits together. 00:46.120 --> 00:50.440 And we also went through the different genetic patterns, which we've seen time and time again throughout 00:50.440 --> 00:51.080 the course. 00:51.080 --> 00:55.960 And it ended with that cool project of the career assistant deployed to Huggingface spaces. 00:56.160 --> 00:59.760 And wow, now imagine what more you can do with that. 00:59.760 --> 01:02.630 Now you're armed with with all of this ability. 01:02.670 --> 01:07.270 Imagine the MCP, the MCP tools that you could give to your, uh, your career agent. 01:07.310 --> 01:07.990 Incredible. 01:08.350 --> 01:10.150 Uh, so all of that was week one. 01:10.590 --> 01:14.110 And of course, week two was my favorite OpenAI agents SDK. 01:14.270 --> 01:16.590 That's very simple, very clean. 01:16.990 --> 01:23.630 Uh, it takes away the hassle of stuff like the JSON, but it doesn't add too much in the way of learning 01:23.630 --> 01:26.550 curves or extra concepts to take on board. 01:26.750 --> 01:32.750 And we made like an SDR, like a sales assistant that could build sales emails. 01:32.750 --> 01:37.550 And then at the end of the week, we built deep research, something that is like a live production 01:37.550 --> 01:37.990 platform. 01:37.990 --> 01:40.870 We built it ourselves, our own version of deep research. 01:40.910 --> 01:42.910 Again, there's so much more you could do with this now too. 01:42.950 --> 01:48.870 There's an MCP tool that that anthropic have made that will guide through complex problem solving that 01:48.870 --> 01:49.470 you'll see. 01:49.510 --> 01:52.510 So there's lots that you could do to make this so much better. 01:53.150 --> 01:57.230 In week three we used crew a very different kind of platform. 01:57.430 --> 02:01.550 Uh, although in some ways I guess I say that in some ways it's quite similar to OpenAI agents SDK, 02:01.590 --> 02:03.310 but in some ways it was it was different. 02:03.350 --> 02:09.310 Obviously the separation the YAML files was was good, but it did add some some like cognitive load 02:09.310 --> 02:12.910 on understanding how it works and the tasks and so on. 02:13.070 --> 02:15.070 But it was we got a lot done. 02:15.070 --> 02:17.190 We made the stock picker app. 02:17.470 --> 02:24.150 Uh, and then we made the developer agent and then the whole engineering team in project five. 02:24.390 --> 02:26.270 That was that was amazing, I thought. 02:26.310 --> 02:30.150 And the way that it just worked and that user interface was really great. 02:30.310 --> 02:36.230 And it's cool that we used the code that it generated there in the current in this week's project as 02:36.230 --> 02:36.390 well. 02:36.390 --> 02:39.910 And the capstone project that's, uh, that that seemed seemed nice. 02:40.070 --> 02:42.990 And then the next week was the experience with Landgraf. 02:42.990 --> 02:47.070 It was a bit of a roller coaster for me from my, my working with, with Landgraf. 02:47.110 --> 02:51.510 In the past, I'd been left feeling like it was quite heavy weight and I thought it would go through 02:51.510 --> 02:54.630 the week quite quickly and that I'd be happy to have it behind us. 02:54.630 --> 02:58.230 But au contraire, I ended up really enjoying myself. 02:58.230 --> 03:03.460 I found my groove with Landgraf and I became very impressed with the way that you can be. 03:04.140 --> 03:09.780 You can use things like checkpointing so rigorously to be able to step through what's happening with 03:09.780 --> 03:11.220 your with your agent workflow. 03:11.380 --> 03:14.900 So I really enjoy myself and I hope you did too. 03:15.180 --> 03:18.500 And I thought that Sidekiq was a great project and I've kept using it. 03:18.500 --> 03:22.820 I've used it for some other purposes since, and I've been adding in some of my own personal, uh, 03:22.940 --> 03:24.780 things that I like to do that it can help me with. 03:24.820 --> 03:30.940 So I found it to be actually truly valuable commercially as well as a very interesting week. 03:30.980 --> 03:37.740 And for all of these weeks, we were repeating things like use of tools, use of structured outputs, 03:37.740 --> 03:43.620 use of memory, along with other genetic patterns that we were seeing recurring a lot with things like 03:43.660 --> 03:45.700 the evaluator pattern and so on. 03:45.860 --> 03:50.260 So it was each of these projects, while in some ways we were learning a new framework, we were also 03:50.260 --> 03:57.540 reinforcing the same kinds of design patterns and the same sorts of approaches and techniques that hopefully 03:57.540 --> 03:58.900 built on top of each other. 03:59.380 --> 04:01.220 Also, Jem was a bit of a of a wildcard. 04:01.220 --> 04:03.020 It was an interesting experience. 04:03.060 --> 04:03.740 Autogen. 04:03.780 --> 04:07.900 Agent chat was actually refreshingly simple and straightforward. 04:07.900 --> 04:14.900 The new version, the 0.4 version, and reminded me a lot of OpenAI agents SDK in its simplicity. 04:15.380 --> 04:18.060 Autogen core is different. 04:18.300 --> 04:19.620 It's a novel. 04:19.660 --> 04:20.780 It's exciting. 04:20.940 --> 04:21.900 Definitely exciting. 04:21.940 --> 04:26.100 It feels like that really could be a big, a big thing in the future. 04:26.100 --> 04:27.700 But it's still in its infancy. 04:27.700 --> 04:30.300 But it was it was great to get our hands on it. 04:30.300 --> 04:37.020 And I felt like the creator project was an interesting, innovative way to do something different that 04:37.020 --> 04:40.260 fit what Agent Autogen core is all about. 04:40.300 --> 04:43.420 And we could try out the agent chat at the same time. 04:43.420 --> 04:46.340 And somehow all of that got us to this week. 04:46.780 --> 04:54.220 MCP is all about MCP, MCP servers, clients, MCP hosts, uh, what it is that's that's not particularly 04:54.220 --> 04:55.220 special about MCP. 04:55.220 --> 05:00.060 And what is special about MCP, why it is that people are so excited about it. 05:00.100 --> 05:02.260 And we built a ton of stuff. 05:02.260 --> 05:07.370 And then we went into the capstone project that we were we were using some of the servers in advance 05:07.370 --> 05:10.770 of that, and it all came nicely together building. 05:10.770 --> 05:15.570 And it was perhaps the first of the the projects, which isn't a sort of general purpose. 05:15.610 --> 05:18.810 It's specifically targeted at financial services. 05:18.810 --> 05:24.970 But the great thing about it is that I can see a similar setup with that similar kind of, uh, of team 05:24.970 --> 05:30.090 of people working on something with autonomy, with a researcher as being a model that you could apply 05:30.130 --> 05:31.810 to many different professions. 05:31.810 --> 05:34.290 And I'm really excited to see people doing that. 05:34.570 --> 05:38.290 Uh, and I feel like it was a nice way to round up everything. 05:38.290 --> 05:42.170 We got to use open AI agents SDK a lot, of course. 05:42.370 --> 05:48.410 Uh, and we also got to play with different design patterns and we got to use MC. 05:48.810 --> 05:50.810 So it brought the syllabus together. 05:50.810 --> 05:56.170 And when you look now at this chart on the screen and everything in all of the boxes, hopefully in 05:56.170 --> 06:02.050 your mind you're mentally ticking every box in terms of a way that you have upskilled a way that you've 06:02.050 --> 06:03.570 built up your expertise. 06:03.730 --> 06:04.490 You are now. 06:04.760 --> 06:08.760 You've mastered all these different aspects of Agentic AI. 06:08.960 --> 06:14.760 And so I definitely would say at this point that you've reached a point where you have mastered a genetic 06:14.800 --> 06:16.360 AI engineering. 06:16.400 --> 06:22.240 And so of course, at this point, I have to take the serious step of showing you the celebratory picture, 06:22.520 --> 06:29.800 the, uh, the picture of our mascot for the course, uh, being super happy with, like, a little 06:29.840 --> 06:30.440 USB. 06:30.480 --> 06:35.440 I think it might even be a USB-C connector to to on one side. 06:35.440 --> 06:38.880 And it's lab stuff and some books from rag. 06:39.160 --> 06:43.840 Uh, and, uh, obviously our robot is super happy, as you should be. 06:43.880 --> 06:45.800 You should be extremely proud. 06:46.000 --> 06:48.000 Uh, you've you've accomplished huge things. 06:48.000 --> 06:49.800 You've learned such a lot. 06:49.920 --> 06:55.480 And you are now a professional and expert in many, many different fields of agentic AI. 06:55.600 --> 07:00.840 And I hope that you are brimming with ideas about ways you can take this forwards. 07:00.880 --> 07:05.800 And if you're not brimming with ideas, then luckily the creator project from last week will be generating 07:05.840 --> 07:08.160 tons of ideas, hopefully for you. 07:08.200 --> 07:13.440 Different ways that you can now be taking this forwards in your in your vertical, in your business 07:13.440 --> 07:15.000 and in your career. 07:15.000 --> 07:18.000 And this leads me to my final requests for you. 07:18.040 --> 07:23.080 Actually, before my final requests, I have a quick pre-final request, which is my editor would kill 07:23.120 --> 07:25.040 me if I didn't mention that. 07:25.040 --> 07:29.520 If it's possible for you to rate the course, it makes such a difference. 07:29.720 --> 07:35.680 The main way that the platform decides whether or not to to advertise this course to other students 07:35.680 --> 07:41.280 to be to to know about it, where, where it goes on its on its list of different courses in this world 07:41.320 --> 07:46.040 and and there are a lot uh is by by students like you rating it. 07:46.280 --> 07:52.600 So it makes a huge difference to me and to my editor and to makes and to all of the future students 07:52.600 --> 07:53.600 that might take this. 07:53.600 --> 07:57.400 So I'd be super, super grateful if you were able to do that. 07:57.400 --> 07:59.840 But then my proper three requests for you. 08:00.040 --> 08:03.400 First of all, help me build out this trading simulator. 08:03.400 --> 08:04.560 I have so much fun with it. 08:04.560 --> 08:07.990 Even if trading isn't really your thing, There's other things that could be done with it. 08:08.030 --> 08:15.270 Take a look at the MCP server marketplace and see other things that are out there that could be added 08:15.270 --> 08:15.590 in. 08:15.830 --> 08:17.750 I want to use more MCP servers. 08:17.750 --> 08:18.830 I want to have more agents. 08:18.830 --> 08:21.430 I want to have more autonomy for this thing to do what it wants. 08:21.550 --> 08:24.830 I would love it to move into crypto markets. 08:24.910 --> 08:31.830 So if you crypto fans out there, then then put that into it or take it in a completely different direction. 08:32.030 --> 08:35.910 And as part of that, look at the prior projects that we've worked on. 08:35.950 --> 08:42.870 You know, the the challenge for the end of this course is such an open one, just as Agentic AI is 08:42.870 --> 08:44.030 so very open. 08:44.030 --> 08:48.870 But look at look at what we've learned about MCP and see how you can apply that to anything, to the 08:48.910 --> 08:53.190 to the Autogen projects or the Landgraaf projects or back to the crew projects. 08:53.230 --> 08:58.350 See where you can put some of these kinds of tools and resources into practice. 08:58.470 --> 09:02.670 And then secondly, share on LinkedIn, post about your success. 09:02.670 --> 09:10.470 Post about completing this course and post about projects that you're doing with Agentic AI, and if 09:10.510 --> 09:16.430 you tag me as I say, I will weigh in and I will celebrate your success with you and it helps to build 09:16.430 --> 09:17.870 up the community. 09:18.150 --> 09:20.790 And on that note, you knew I was going to say it. 09:21.910 --> 09:26.150 If you don't mind, if you if you're if you're okay to to connect with me, if nothing else, just to 09:26.150 --> 09:26.590 shut me up. 09:26.590 --> 09:28.870 So I stopped pestering you about connecting with me. 09:28.870 --> 09:31.190 That should be reason enough for you to connect with me. 09:31.350 --> 09:32.350 Uh, but then. 09:32.350 --> 09:37.150 Then we get to build up a big old data science community, and that's a great thing. 09:37.230 --> 09:38.550 So I would love it. 09:39.310 --> 09:45.070 And with that, with that, let me finish by saying a enormous thank you. 09:45.150 --> 09:50.750 Thank you so, so much for making it all the way through to the bitter end, for putting up with all 09:50.750 --> 09:53.390 these hours of me and all these projects. 09:53.430 --> 09:58.630 Some of them fascinating, some of them perhaps silly, but hopefully all of them were fun. 09:58.670 --> 09:59.710 And thank you. 09:59.710 --> 10:01.550 Thank you for for getting through all of it. 10:01.550 --> 10:03.550 Thank you for getting to this point. 10:03.790 --> 10:07.950 Please keep building, keep sharing, and be in touch.