8 ChatGPT Agents That Do My Work for Me (Steal These) — Transcript

Full transcript

1. 0:00Today, we're going to turn Chat GBT from
2. 0:02something we prompt into an agent that
3. 0:04does our work in the background 24/7.
4. 0:07[music]
5. 0:07This video is going to be a complete
6. 0:09guide to setting up agents to do your
7. 0:12work in the cloud autonomously directly
8. 0:14from the Chat GBT desktop app, web app,
9. 0:17or iOS app. And so, in part one, I'm
10. 0:20going to explain the different parts of
11. 0:21ChatGBT that are relevant for agentic
12. 0:24tasks. Then in part two, I'm going to
13. 0:26show you my eight favorite agent
14. 0:28workflows that I've set up that have
15. 0:31helped me and my business the most so
16. 0:33that you can create similar workflows
17. 0:35that help you. All right, guys. Today
18. 0:37we're creating a Gentic Workflows that
19. 0:40work 24/7 in the cloud. Whether you're
20. 0:42using the desktop app like I am here or
21. 0:45if you're using the chat GBT web app on
22. 0:47chatgbt.com or if you're using the
23. 0:50ChatGBT iOS app, we are going to be able
24. 0:52to create these cloud agents that do our
25. 0:55work. Of course, we're going to be
26. 0:56talking about the eight agentic
27. 0:58workflows that really drive revenue in
28. 1:00my business. But first, I do want to
29. 1:03talk about some basics that you should
30. 1:05understand before we get here. So the
31. 1:07first thing that you need to understand
32. 1:09is that chat GPT has evolved and there's
33. 1:11a new setting in chat GPT called chat
34. 1:14GPT work. This is where a chatbot
35. 1:16becomes an agent right this AI agent has
36. 1:20access to a computer with files with a
37. 1:24browser and it is becoming much more
38. 1:25like an employee and you can use this
39. 1:27directly in the web app. You can use it
40. 1:30in the desktop app chat and work also
41. 1:33you can use it in the iOS app. You can
42. 1:35go from chat to work. And in my opinion,
43. 1:37the most useful part of GBT work are
44. 1:40something called plugins. And so I could
45. 1:42use the notion plugin, for example. And
46. 1:45I'm going to say, please create a
47. 1:47database called commitments and
48. 1:50deadlines. Please look at my Gmail and
49. 1:55look at my Slack and find any
50. 1:59commitments or deadlines and put them in
51. 2:02a database from the past week. Oh, also
52. 2:07look at my meeting notes at Whisperflow
53. 2:12cuz they now have meeting notes built in
54. 2:14and we can run this prompt. And so
55. 2:16there's two main reasons why plugins are
56. 2:18so useful. The first one is context and
57. 2:21the second one is actions. Plugins allow
58. 2:24your agent to get full context of all of
59. 2:27the data that's inside notion, that's
60. 2:29inside Gmail, that's inside Slack, and
61. 2:32that's inside my meeting notes on
62. 2:33Whisperflow. But not only does it have
63. 2:37context and able to pull the data from
64. 2:39these four platforms that I use for my
65. 2:41business, it can also take actions in
66. 2:44all of these. It's actually going to
67. 2:45create a database inside Notion. You'll
68. 2:48see in just a second. Okay. So, notice
69. 2:50here that after 4 minutes, GPT work is
70. 2:53done. And I want you to think for a
71. 2:55second, how long would this task take
72. 2:58you? Especially if the result is this.
73. 3:01So, if we open this up inside the chat
74. 3:04GBT browser, inside the ChatGBT desktop
75. 3:07app, we can see the database that it
76. 3:09created. It went through notion, Gmail,
77. 3:12Slack, and all of my meeting notes and
78. 3:15found all of this information. It added
79. 3:18the commitment.
80. 3:20Um, it added the counterparty. It added
81. 3:24the evidence. It added the owner, the
82. 3:27source, analyzed four different
83. 3:29platforms and created a new notion
84. 3:30database. And it did that in 4 minutes.
85. 3:33And that is how powerful it is. Okay. So
86. 3:35for me, it was about 5 minutes of total
87. 3:37work. One of which was typing out this
88. 3:40prompt and then four minutes of waiting
89. 3:43for its response. However, there's
90. 3:45something inside the chat GBT app and
91. 3:48that is this scheduled tab. And so on
92. 3:51the scheduled tab, we can do something
93. 3:53like this. So you'll notice here that it
94. 3:56said if helpful, I can set update
95. 3:58commitments database daily. So new
96. 4:01commitments are captured automatically.
97. 4:03So I can either click on this directly
98. 4:05or I can type into this. And so this
99. 4:07brings me to special agentic workflow
100. 4:11number one that I use, which is this
101. 4:13commitments and deadlines workflow. And
102. 4:15so what I can do is I can just click on
103. 4:18this right here. And so now that we can
104. 4:20see this right here, all I need to do is
105. 4:22go up to scheduled, and we'll see it
106. 4:26right here, which is update commitments
107. 4:29database. This will run daily at 8:00
108. 4:33a.m. Okay, let's actually add one
109. 4:35additional thing so that I'm held
110. 4:37accountable by another human on these
111. 4:39things. Hey, I actually want you to
112. 4:42email I'm going to put Emily at
113. 4:45agentnative.inc.
114. 4:47Please send her an email with a link to
115. 4:50these this database so that she can view
116. 4:53them. And then whatever ones you add,
117. 4:56please add. And then make sure to
118. 4:57mention that this is a full list of the
119. 4:59commitments and deadlines Riley has
120. 5:01agreed to. And this should be emailed
121. 5:03every single day as well.
122. 5:06And then you can always mention like
123. 5:07this should be part of this automation.
124. 5:12And so basically every morning it'll
125. 5:15analyze all of these things. It'll put
126. 5:16them in a notion database and then it
127. 5:18will send an email to Emily from me so
128. 5:21that she can see it and then hold me
129. 5:23accountable to those items. And so there
130. 5:25you go. We've created this first
131. 5:28automation that just happens every
132. 5:30morning. AI will go through all of these
133. 5:32thanks to the plugins and get that work
134. 5:34done every single morning 8 a.m. That's
135. 5:37workflow automation number one. Okay.
136. 5:39So, the second automation that I'm going
137. 5:41to do with chat GBT work involves
138. 5:43hiring. I tweeted this the other day. I
139. 5:45said, "If you want to use or test agent
140. 5:47platforms and write about what you learn
141. 5:49and get paid for it, please send me an
142. 5:50email. I'm hiring at agentnative." And I
143. 5:52put my email right here. And so what I
144. 5:55can do now, I can very easily have AI
145. 5:59analyze my email and add all of the
146. 6:01people who've sent me an email into a
147. 6:03database. In fact, I have done this
148. 6:05already. This is my hiring dashboard. I
149. 6:08have a lot more fields here. I just
150. 6:10can't show them because I don't want to
151. 6:11give away any sensitive information. And
152. 6:14what I want to do is I want this to be
153. 6:16updated every single day. And so what
154. 6:19you can do is you can just go to
155. 6:21wherever your notion database is and I
156. 6:23can say, "Hey, I've been hiring for this
157. 6:27position." And then what I'm going to do
158. 6:29is I'm just going to give a screenshot
159. 6:32of this tweet. That's all I'm going to
160. 6:34do.
161. 6:37And they are sending emails to the email
162. 6:42listed there. And all of the entries are
163. 6:47here. in this database and then I can
164. 6:50paste that notion link right here.
165. 6:53Please, can you create an automation
166. 6:54that adds all of these people to that
167. 6:57database? Create another automation for
168. 7:00nighttime which researches all the
169. 7:02people and adds research about these
170. 7:04people in the notes. Every single day in
171. 7:07the morning, add them. At night,
172. 7:08research them. And so I can fire this
173. 7:11prompt in. And again, it works directly
174. 7:14from chatgpt.com.
175. 7:15It'll create this automation. Okay. 43
176. 7:18seconds later, we now have these right
177. 7:20here. Both run daily. If you put your
178. 7:23email out there or any email that you
179. 7:25want to connect to GPT work out there on
180. 7:27the internet to gather information, you
181. 7:29can just use GPT work to monitor that
182. 7:32email and turn it into a dashboard.
183. 7:35Okay, so the next two AI agent workflows
184. 7:38that I want to create involve social
185. 7:40media, specifically tweets on my new
186. 7:44brand account. You know, I've done a
187. 7:45pretty good job growing my Twitter. I'm
188. 7:47at around 233,000
189. 7:49followers. But I do want to grow my new
190. 7:51account that I created a few months ago
191. 7:53called Agent Native. And that's the name
192. 7:55of my company that I'm building. I want
193. 7:57to do two things with this. The first
194. 8:00thing that I want to do is research for
195. 8:03tweets. And then I want to actually
196. 8:05schedule the tweets. And so the first
197. 8:09automation number one involves scraping
198. 8:12um and like it's just doing research on
199. 8:13Twitter. And then uh two is scheduling.
200. 8:16And what we're going to be doing is
201. 8:18we're going to be using typefully for
202. 8:21scheduling and we're going to be using
203. 8:23scrape creators for scraping. Okay. So
204. 8:27this brings me to automation number
205. 8:29three which is Twitter scraping. So
206. 8:31here's the prompt that I'm going to use
207. 8:32is I want to grow agent native my uh my
208. 8:35Twitter account for the brand. I comment
209. 8:38on the agent space and make high quality
210. 8:42content and around open AAI codecs and
211. 8:45similar tools. Look up the similar
212. 8:48tools. Um, and you can also look at the
213. 8:51handle which is agent native and then I
214. 8:54want you to search Twitter and find
215. 8:56relevant popular tweets over a 100 likes
216. 8:59that show important updates to agents
217. 9:01and becoming agent native, new product
218. 9:03updates, new model releases. And I want
219. 9:06and we're going to be using scrape
220. 9:07creators. In the description below, I
221. 9:09have a little guide that you can get um
222. 9:11for scraping social media. And I want
223. 9:13you to create a new database in at
224. 9:17notion. We're using another plugin
225. 9:19called scraped tweets. Please do this
226. 9:24now. Scrape the latest. So now I'm going
227. 9:27to have it scrape. It's going to create
228. 9:29a notion database, store all the data
229. 9:33for the tweets, then we'll create an
230. 9:36automation to do this every few days.
231. 9:42Okay. So this is going to take a little
232. 9:44bit and then we'll create the automation
233. 9:46once it's done. Okay. So after 8 minutes
234. 9:48it's done and I can actually open this
235. 9:51up directly in the browser. We should
236. 9:54see all of the tweets are pulled. Here
237. 9:58we have the link URL. Any media URL is
238. 10:02automatically pulled. This is really
239. 10:04cool. We see all of the replies,
240. 10:06reposts, etc. Okay, this is just step
241. 10:09one. And then we're going to create
242. 10:12tweets from the agent native account
243. 10:13which provide commentary about these
244. 10:16posts. That's going to be number four.
245. 10:18We need to finish this one and say,
246. 10:20"Hey, please create an automation that
247. 10:23automatically gets this done at uh 9:00
248. 10:26a.m. every morning. I want that to go
249. 10:28off and update this database."
250. 10:30Okay? And there you go. This is going to
251. 10:33happen every single morning at 9:00 a.m.
252. 10:36Okay. Okay, so that brings me to
253. 10:38automation number four, which is
254. 10:40creating commentary posts from my agent
255. 10:42native account. So for this, I'm
256. 10:45actually going to be using Typefully.
257. 10:47Typefully allows me to sign in to uh a
258. 10:51bunch of Twitter accounts. And if I go
259. 10:53to settings here and click settings and
260. 10:57go to API, I can get an API key. I've
261. 11:00already set this up, but I gave this API
262. 11:02key to GPT work. And now it can use a
263. 11:06skill. I believe it's just going to be
264. 11:08called typefully drafts. And so now what
265. 11:10I'm going to do is I'm going to say,
266. 11:12please can you create five draft posts
267. 11:18that are actually let's do 10 draft
268. 11:20posts on the best posts from the ones
269. 11:23you've just gathered. So these should
270. 11:24either be quote tweets or it should post
271. 11:27the video. Look up how to do that.
272. 11:29Please do that on 10 of them and put it
273. 11:31on the agent native account. Do not
274. 11:32actually post them, just draft them on
275. 11:35Typefully.
276. 11:37And so that is going to be the workflow.
277. 11:39We gather all of the tweets here in a
278. 11:42giant notion database. We're not ever
279. 11:44going to look at it. AI just stores all
280. 11:47the data there and then it's going to
281. 11:48take the most recent ones or the best
282. 11:50ones from the past day and add them to
283. 11:52Titefully. And this is where I can come
284. 11:55and I can just approve them and schedule
285. 11:57them. And so that's how we can post more
286. 11:59on Twitter, except I will actually do
287. 12:00some manual editing because I don't want
288. 12:03AI to tweet for me. I believe that if I
289. 12:05just put in 10 minutes a day, I can make
290. 12:07them 10 times better because I have a
291. 12:09lot of experience on Twitter. All right,
292. 12:11so after 3 minutes, it is done. And if I
293. 12:14rightclick on this, open an external
294. 12:16browser. There we go. Look at this. It
295. 12:19created I'm going to zoom out a little
296. 12:21bit. It created all of these tweets. All
297. 12:2410 of these tweets are ready to be
298. 12:28posted. Voice is becoming real input
299. 12:30layer for coding agents. Gemini can now
300. 12:32ground what you say against the code,
301. 12:34file names, and variables and what's on
302. 12:36the screen. You won't just type prompts
303. 12:38into agents. You'll direct them while
304. 12:40you work. This is very cool. And it
305. 12:44quote tweeted them. It did the post the
306. 12:46video thing pretty well, which is really
307. 12:49cool. This is awesome. Okay, so now what
308. 12:51I want you to do, I want you to do this
309. 12:52every day. And I want you to add five
310. 12:55posts every day. Always pay attention to
311. 12:57the previous tweets that I post on Agent
312. 12:59Native and take from my voice,
313. 13:01especially the ones that perform well.
314. 13:03Take the ones that perform well, and
315. 13:05those should be the ones that you use as
316. 13:06examples going forward. So, please
317. 13:08create that automation now.
318. 13:11All right, 9:00 a.m. every day. We now
319. 13:13have that one set up as well. Okay, so
320. 13:15the next four are a little bit quicker.
321. 13:17Here is automation number five, which is
322. 13:20finding podcast guests. So, I'm going to
323. 13:23say, "Please do deep analysis using the
324. 13:25YouTube researcher skill." This uses
325. 13:27scrape creators. It just does it in a
326. 13:28specific way to analyze the YouTube
327. 13:31transcripts to look for good guests for
328. 13:32my podcast. And then I said, "Please add
329. 13:34them to this database." Now, here's a
330. 13:36fun thing that you can do when you're
331. 13:38using the desktop app. Check this out.
332. 13:39So, if you're typing a prompt here, I
333. 13:41can just open up Notion and let's go to
334. 13:43the podcast guests database. And I
335. 13:46created this manually a while ago. So
336. 13:49what I can do here is while on this I
337. 13:51can click both command keys. So check
338. 13:54this out.
339. 13:56It automatically just takes the page and
340. 13:59adds it as context. This should also be
341. 14:04an automation for every morning at 11
342. 14:08a.m. And so now let's just wait. And
343. 14:11it's going to do some research, add some
344. 14:13more people to this database, and then
345. 14:14it's going to create the automation. And
346. 14:16there you go. So it created the
347. 14:18automation and remember you can always
348. 14:20click directly on the automation to see
349. 14:22that it did it correctly. If you want to
350. 14:24change any of the details you can edit
351. 14:26it directly in here. Okay. So for number
352. 14:29six I want to do one from the phone. You
353. 14:32can create these scheduled tasks
354. 14:34directly from your phone. So I'm going
355. 14:35to type this in right here. This one is
356. 14:37actually very common and I highly
357. 14:39recommend it if you have multiple
358. 14:40emails. So the email I have connected is
359. 14:44my main agent native email, but it
360. 14:47someone pe many people reach out to me
361. 14:49about brand sponsorships. And so I want
362. 14:52to be able to do email forwarding. So
363. 14:54I'm going to say when someone emails me
364. 14:56about a brand sponsorship, please
365. 14:57reroute it to rileley raugoddia.com.
366. 15:00Like forward it to that email uh so my
367. 15:03team sees it. Um do this uh 2x per day.
368. 15:10set up that automation and I can fire
369. 15:13that prompt out. And there you go. This
370. 15:16one is very simple, but I find that a
371. 15:18lot of people would find this one
372. 15:20useful. And again, you can click
373. 15:21directly on it. Right here, you can see
374. 15:23daily forwarded sponsorship leads. We
375. 15:26can click on this right here and this
376. 15:28will forward anything to my email to
377. 15:30this new email, which is really cool.
378. 15:33And again, you can edit it directly if
379. 15:35you find some mistakes in the
380. 15:38automation. Okay, so now I want to talk
381. 15:40about the seventh automation that I
382. 15:42create within the chat GBT app. This is
383. 15:45the only one where I do it inside codeex
384. 15:48because more permissions by default. And
385. 15:52for this one, we're going to be using
386. 15:54the new messages integration which
387. 15:57allows me to pull directly from
388. 15:59messages. And what I can do here is I
389. 16:02can say please summarize my messages
390. 16:06from Adam D uh from today in text. This
391. 16:14allows codeex to go straight into the
392. 16:18database where my iMes are stored and it
393. 16:21can get back to me with exactly whatever
394. 16:24question I'm asking. It can even
395. 16:26summarize every text you've ever sent.
396. 16:28This will use a lot of tokens and might
397. 16:29be somewhat expensive, but you can do
398. 16:31it. And the point is you have full
399. 16:32access to your text directly in codecs.
400. 16:34And so this is a very clean summary and
401. 16:37I changed it to the last 5 days cuz
402. 16:39there wasn't actually much from the last
403. 16:41one day. But this summarized our
404. 16:43conversation very concisely and
405. 16:45perfectly. And so the point is I can
406. 16:47take any text message chain with anyone
407. 16:49and I can ingest it with codeex and I
408. 16:53can actually store it in a database. And
409. 16:55so that's exactly what I've done for
410. 16:57sponsorship texts. I have a a manager
411. 17:01for my content and they actually have a
412. 17:04full team and so I have chats with all
413. 17:06of them. And because I want my agents to
414. 17:08have context over everything
415. 17:11at 8 a.m. it reads my texts with with
416. 17:14four different contacts. If they're in a
417. 17:17group chat with other people, it'll also
418. 17:19bring those in and it stores that data
419. 17:22inside notion just like the previous
420. 17:24ones. So that if I ever ask any of my
421. 17:27agents about any deal, it has all of
422. 17:30that context from text messages because
423. 17:32so much of my communication is done via
424. 17:35text. That's really important to have
425. 17:37your agents do things effectively
426. 17:39because it needs all of your context.
427. 17:41And so that is automation number seven.
428. 17:44I've had this one running for a long
429. 17:46time. All right, guys. So for the final
430. 17:48automation that I highly recommend is
431. 17:51one that is once a month. And so I've
432. 17:53actually been workshopping this. I've
433. 17:55only ran it once last month and it was
434. 17:58incredibly good. And for this one, we
435. 18:01are actually going to be using the
436. 18:03highest model
437. 18:055.6 soul extra high and this model and
438. 18:09really do deep analysis here and write
439. 18:15the most detailed summary ever. And I'm
440. 18:19just going to say don't do this now.
441. 18:22schedule this for the last day of the
442. 18:26month for the next 12 months. And so
443. 18:30this is basically review the past month
444. 18:33across notion, Gmail, Slack, Calendar,
445. 18:35Whisper Flow, Google Drive, Scrape
446. 18:37Creators, and everything that it has
447. 18:40access to because I have my entire
448. 18:43company, all of its context is within
449. 18:46one of these plugins. And it is able to
450. 18:48analyze all of it with the smartest
451. 18:50model ever. And it can do this once a
452. 18:53month and really get you to focus. And I
453. 18:56can tell it exactly where to put it.
454. 18:58create this report in notion and then
455. 19:02email it to Emily as well. And so this
456. 19:07will happen automatically at the end of
457. 19:09every month. You can just deploy
458. 19:11intelligence and this will happen in the
459. 19:14cloud and you can even create these
460. 19:15automations directly from your phone. I
461. 19:17could have done that whole thing from my
462. 19:19phone. You can just create these
463. 19:21automations by talking to chat GBT work.
464. 19:24All right, guys. So, those are the eight
465. 19:26automations or agent workflows that I've
466. 19:29created that run automatically using GBT
467. 19:32work, right? We analyze emails and Slack
468. 19:34messages for commitments and deadlines.
469. 19:37We automated our hiring database. We're
470. 19:39scraping Twitter. We're drafting tweets.
471. 19:42We're finding guests for our podcasts.
472. 19:45We're doing email forwarding. We're
473. 19:46importing all of our text. This is the
474. 19:48only one where we use codec so it can
475. 19:50run locally on our computer. And then
476. 19:51we're doing this giant mega monthly
477. 19:54summary with all of our context. And
478. 19:56here they are. You can see them in the
479. 19:58desktop app, the web app, or the iOS
480. 20:00app. Thank you guys so much for
481. 20:02watching. And OpenAI, thank you so much
482. 20:03for sponsoring this video. I'll see you
483. 20:05here for the next