OpenAI Dot vs Grok Bot: qual assistente funciona melhor por voz?

Author: Cortes do Rafael Quintanilha · Watch on YouTube

Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.

Original language: English

Original video: OpenAI Dot vs Grok Bot: qual assistente funciona melhor por voz? · Channel: Cortes do Rafael Quintanilha · 09:09 · Watch on YouTube

Topics: AI agents, Enterprise AI applications, Task and workflow automation, Multimodal models

Summary

The presenter tests OpenAI’s voice assistant on analytics questions, coding tasks, and a YouTube ranking check, then compares it with Grok Bot. OpenAI completes the requests, including retrieving program metrics and checking a video’s position in YouTube Studio, but the presenter finds its long pauses and slow responses make voice conversations awkward. Grok Bot responds more continuously and makes repository checks and closes a pull request, though it initially gives inconsistent information about whether another pull request is open. The comparison highlights similar assistant capabilities but different trade-offs in naturalness, speed, and conversational flow.

Key points

The OpenAI voice assistant initially failed to respond during earlier tests, but worked in this demonstration.
For the launch period the presenter clarified, OpenAI reported 8,076 pageviews and 170 unique visitors.
OpenAI reported 130 lesson completions by 55 people on Friday and 99 completions by 35 people on Saturday.
The assistant handled a request involving a yellow UI change, a screenshot, and a pull request, although sending the image encountered an authorization error.
OpenAI checked YouTube Studio and reported the video ranked fourth out of ten by views, with 11.7k views in its first 7:29.
The presenter praised OpenAI’s natural-sounding voice but criticized its slow responses and extended silences.
Grok Bot gave Friday and Saturday class activity figures and explained that the current day’s data was still partial.
Grok Bot first said there were no other open pull requests, then found PR 57 and closed it at the presenter’s request.

Chapters

Start	Chapter	Summary
00:00	Testing OpenAI voice mode with program analytics	The presenter tests OpenAI’s voice assistant and asks it to retrieve launch-period visits, unique users, and lesson completions.
02:15	A more complex OpenAI workflow	The assistant is asked to handle a screenshot and pull request for a color change while also checking a video’s YouTube ranking.
04:22	OpenAI results and voice-mode drawbacks	The tasks eventually succeed, but the presenter emphasizes that long pauses and slow responses make voice use impractical for routine work.
06:29	Comparing Grok Bot in audio mode	The presenter tests Grok Bot with class activity and repository questions, observes its more continuous responses, and asks it to close a pull request.
09:09	Closing comparison	The presenter concludes that the two personal assistants have very similar proposals and has shown them working side by side.

Quotes

having your personal assistant and having it — 02:15

Naturalness without agility hinders the conversation. — 04:53

it is very, you know, slow for you to talk via voice. — 05:56

My bad, the rush got in the way. — 08:39

Concepts and entities

OpenAI Dot, Grok Bot, OpenAI, YouTube Studio, pull request

Full transcript (original language)

00:00 I want to show one of the main features that OpenAI insisted on highlighting at DevDay, which is the ability to operate through voice mode. So you can see here on your dot there’s a little button for you to connect. And actually, it even simulates a phone call here. Some people liked it, others didn’t. I tested this several times, guys, and at first it didn’t work. She wouldn’t answer me, but then she started to, okay? So let’s do this test together here. Let’s see if OpenAI answers me this time.

00:39 >> Hi, why are you speaking to me in English? >> Oh, my bad. Let’s go in Portuguese, yes . What do you need? >> Look, I wanted to know how many people accessed the program in total? How many unique users accessed the program during the launch we held last week? So , we held a launch on Tuesday, Wednesday, and Thursday. We had three live streams. I want to know what the total number of visits and unique users was during that period. >> Got it. I’m pulling the data for the

01:10 launch period from September 28th to October 4th and checking the total visits and unique users. >> No, it’s from Tuesday to Thursday. Tuesday, Wednesday, and Thursday. Oh, I see. Just Tuesday, Wednesday, and Thursday, adjusting for that timeframe now. >> Well, guys, here I’m showing you a use case that I think is really cool, okay? Which is having your personal assistant and having it—Thursday, September 29th to October 1st, there were 8,076 pageviews and 170 unique visitors to the EAI program. Each person is counted only once in total, even if they

01:42 accessed it on more than one day. >> Right? And can you tell me then how many lessons were completed on Friday and Saturday? >> Good question. Let me check that Friday and Saturday breakdown for you. On Friday, there were 130 lesson completions by 55 people. On Saturday, 99 completions by 35 people. Adding both days, that’s 229 completions. >> Is that right? Thank you very much. So guys, I showed you an example here that I think is really cool to do with a personal assistant, which is this idea

02:15 of brainstorming, analyzing your platform, your product, or your company , talking and asking your assistant to pull data. Here in this example, it was much faster than it was before, okay? If you perhaps tested it post-launch, maybe you didn’t like it as I didn’t like it much. Now in this demo with you , it really has improved. I think they are fixing their compute problem, okay? Actually, guys, it worked so well here that I even want to do another test. Let’s even try to do something a little more complicated here. Go ahead. I’ll

02:45 call her again here. Let’s see. >> Good evening. Back real quick, huh? >> Yeah. Yeah, I saw here that I just got the message saying that the task to change the color to yellow was done. Right. >> Exactly. The final result was implemented with yellow#FAC15 and the tests passed. >> Okay. Can you send me a screenshot of how this page looks and create a pull request for me? >> Sure. I’m taking care of it. I’ll request the screenshot and the pull request now.

03:16 >> Right. Uh, while you do that, I published a video this Sunday. Can you confirm to me what position it’s in, like 1st of 10, 2nd of 10, etc., on YouTube. Is this video in right now as I’m speaking to you? >> Yes, I will check the current position for you. >> Look, this here, guys, is a test I want to do, which is the following: her ability to work in parallel, and including this second check I asked for from YouTube, she will possibly need to have access to the studio that she will only be able to do locally here on my computer, okay? So I have some doubts

03:48 about how well she will be able to do this and the speed, okay? Because even though she impressed me here on the first task regarding speed, many of the criticisms we saw regarding the models were precisely the fact that she wasn’t that fast, okay? And when you are on a call, >> the finish was quick, the files were generated, but sending the images failed due to an authorization error. I’m looking for a supported way to deliver this to you. >> I think for an authorization error you can’t just take, ask the agent on the local machine to take a screenshot and send it in the chat itself. >> I actually can, I’ll ask for that now.

04:22 Or you could ask it to put the screenshot as an attachment in the GitHub pull request. >> That works too. I just need to ensure the screenshot shows the yellow version of the task in the cloud, because the published site hasn’t changed yet. >> Well, so getting back to it, guys, the big issue is that when you’re working and the model goes silent for a long time like she did, especially on a voice message, it gets a bit weird, right? So, >> perfect. I’ll include the option to keep the screenshot attached to the PR. Meanwhile, I’m still waiting for the attachment confirmation, the PR link, and the video’s ranking in the studio. >> Yeah, it’s much better than I

04:53 imagined. I still think it’s slow for a conversation that would be, you know, something you’d do in your daily routine, right? But it is better. It’s a breakthrough, and the voice is very natural, isn’t it? You’ll see, we’re going to compare it with the Grok Bot too, and you’ll see how OpenAI’s sounds much more natural, although it is slower; it’s useless to be natural if it doesn’t work, right ? So, see? The amount of time it takes to get back to me, it just stays silent , you know? And for a voice conversation, that’s really bad. Let’s wait here to see when it

05:24 replies. >> Fair enough. Makes sense. I’m still trying to show the screenshots in the task conversation and verify the video’s position. I still don’t have the confirmed result, but I agree with you. Naturalness without agility hinders the conversation. >> Look at this, guys, how cool. It appeared here, look, in the agent’s output, a home desktop. If I click here , look, it’s exactly the yellow color , you see? So it’s possibly generating there in that cloud conversation, right, the screenshot with the yellow buttons to put in the pull request. So it works, right? You see that it works well? The big problem

05:56 is that it is very, you know, slow for you to talk via voice. So it’s not really worth it for you to do this via voice. It won’t help you. >> Checking the studio now at 9:14 PM. Your video GPT 6.1, “Um sol e sonete 5.5,” is ranked fourth out of ten by views with 11.7 k in the first 7:29. >> Yeah, it got it right, it managed to go to the studio and get the information. But it took quite a while, it took, I don’t know, 4 minutes to be able to go open the studio, check, and get back to me, right? So, it really is quite slow,

06:29 okay? It seems it will work, so it’s just a matter of speed. I’ll even hang up here so we don’t waste too much time, but you saw more or less what the experience is like, right? If it were faster, it would be better. I believe that the moment we manage to bring a more interesting speed to this, then we’ll have a much better workflow. Back here then to Grockbot, let’s try to do something similar in their audio mode, okay? Let’s go. Hi, how are you? Are you there? >> I’m here. Great. >> Great. Can you tell me how many classes were watched in total on the program on

07:02 Friday and Saturday? On Friday, the 2nd , there were 84 distinct people with 313 plays and 55 completing at least one class. On Saturday, the 3rd, 50 unique people watched at least one with 187 total class starts and 35 completions. I want the ranking by class of >> no. Just answer what I asked you. How many people watched classes in total today? Searching for today’s people. >> Yeah, we have here in the Grock interface, you can see here below, right?

07:34 >> This is just the start of the work. As soon as the numbers come in, I’ll let you know. >> See, connecting to the post we can see that it’s fetching information from the >> Today’s data is still partial. On that side it’s still Sunday, October 4th. So , there are no numbers for Monday yet. So far, 35 distinct people started at least one class, 117 total plays and 19 completed, totaling 55 completions. Since the day isn’t over, these values can still go up, >> okay? Can you check the open request in the program repository for me? There

08:08 >> I’ll pull the open PR, it’s in progress . Do you want me to detail any specific part when it arrives? No, I just want you to tell me which PR is open. You see that she talks a lot more, she doesn’t leave much silence, right? And even when it’s silent, we can see better here what Grockbot is doing, right? Now it seems like it disappeared . >> The open PR was 56 Fit Green Primary Buttons, but actually it’s closed, not merged. It closed today around 3:13 PM Brasília time, still as a draft.

08:39 >> But there are no other open PRs? >> Checking if there’s any more. Nothing new around here yet. Let me know if you want me to force a new look. Actually, yes. The 57 Fix change website primary color to. Opened tonight around 9:02 PM Brasília time. >> And how did you say there wasn’t one and a second later you said there was? >> My bad, the rush got in the way. The result had just arrived when I spoke. >> Right. Do the following for me. Can you close this pull request too? >> Closing 57.

09:09 >> And there you go, folks, let’s see here . Grock it up. It’s open. >> Just the beginning. When it confirms, I’ll tell you. >> That’s what we created in that Cloud task, right? And it closed, I manage >> PR 57 closed without a merge around 9:19 PM Brasília time. The repository is left with no open PRs. >> Wonderful, everyone. So, what did I want to show you here? I wanted to show these two personal assistants working side by side, because their proposal is truly very similar.