9:24

Plays here · Open on youtube ↗

UnknownThe exact OpenAI DevDay Community Spotlight visibly features Mindtrip, but the page exposes no paid-promotion label or contract evidence; event inclusion proves attribution, not payment: https://www.youtube.com/watch?v=Xjl2SqqrA4s9:24 · spoken hook

“So I am Garrick Toubassi. I am a co-founder of Mindtrip, and I think more importantly,”

The opening line · 0:00 · move: story

Uses a DevDay stage demo to show how Mindtrip converts blogs, images, and video into mapped, structured itineraries, then exposes the GPT-4o, OCR, transcript, and frame-sampling choices underneath.

1.7Kviews43likes↑0.1×vs their usual post · 15K views14creators ran this script

@OpenAI · 2M followers · 0 of 30 recent posts mention Mindtrip · creator page →

Playbook · Any inspiration becomes an itinerary →

The script, step by step10 steps

1Disarm the audience before a technical democreator-claim0:00–0:49

Disarm the audience before a technical demo at 0:00

0:00

Disarm the audience before a technical demo at 0:00

0:00

Disarm the audience before a technical demo at 0:01

0:01

Disarm the audience before a technical demo at 0:02

0:02

Disarm the audience before a technical demo at 0:03

0:03

Disarm the audience before a technical demo at 0:04

0:04

Disarm the audience before a technical demo at 0:05

0:05

0:00So I am Garrick Toubassi. I am a co-founder of Mindtrip, and I think more importantly, I'm standing between you and lunch. So I have just a tight 90 minutes, and if we all cooperate, we'll get out of here. Just kidding. I'm going to talk to you a little bit about what we're doing at Mindtrip with regards to multimodal inputs. Spoiler alert: I will not be talking much about the cool new Realtime API. We've been doing some prototyping with that, and I'll comment on that at the end. But mainly, I'm going to talk about all the great stuff you can do with the API you already know and love, Chat Completions. All right, so let's dive in.

The lunch joke and promise to use the familiar Chat Completions API make the founder's multimodal demonstration approachable to the DevDay audience.

2Establish the travel job beyond a chatbot0:49–1:26

Establish the travel job beyond a chatbot at 0:09

0:09

0:49First of all, what the heck is Mindtrip? Mindtrip is your AI-powered travel platform, and we have this ambitious goal to help you with the full travel lifecycle, whether it's inspiration and discovery, planning, collaborating with other travelers, booking, and then on the trip, et cetera. So this is a big, expansive vision, and this is my one marketing slide. My CMO will be happy that I included it. It also represents the peak production value of the whole deck, so it's all downhill from here.

The full travel lifecycle gives the application a job that extends from inspiration through planning, booking and being on the trip.

3Name inert text as the problem1:26–2:39

Name inert text as the problem at 1:53

1:53

1:26But what really are we talking about? When we started ideating around travel, we started, as many people did, with ChatGPT. And believe me, I love ChatGPT, use it all the time. But when you're doing travel planning, you're left with inert text. It kind of leaves you right at the altar. You can't really plan the trip. You can't really action it. And so I think with a lot of LLM-based applications, the challenge is: how do you take that inert text and make it actionable and alive? With Mindtrip, we're able to connect entities in the conversation, put it on a map, et cetera. We've got photos. We've got reviews. Bada boom, bada bing. By the way, does this look a little familiar? That Wanderlust demo. Just for the record, we were there first. They saw an early demo of this and then asked if we didn't mind if they'd do a little thing in DevDay a year ago, so that's pretty cool. A lot of what we're doing is taking the text, making it alive and actionable.

The founder contrasts a text-only answer with maps, photos and reviews, defining actionable travel planning as the reason for the product.

4Turn existing inspiration into the input2:39–3:25

Turn existing inspiration into the input at 3:46

3:46

1 lines · members

5Convert a Portugal article into a mapped plan3:25–4:16

Convert a Portugal article into a mapped plan at 5:39

5:39

1 lines · members

6Start from an image instead of a place name4:16–5:17

Start from an image instead of a place name at 7:33

7:33

1 lines · members

7Extract a London trip from a social video5:17–6:15

Extract a London trip from a social video at 8:29

8:29

1 lines · members

8Explain which part of the input carries meaning6:15–7:49

Explain which part of the input carries meaning at 0:08

0:08

2 lines · members

9Make the pipeline practical to operate7:49–8:17

Make the pipeline practical to operate at 0:07

0:07

1 lines · members

10Return the technical lesson to the blank-page problemcta · none8:17–9:24

Return the technical lesson to the blank-page problem at 0:06

0:06

1 lines · members

8 more linesmembers read and copy every script on the site

Where this post sitsthe repetition is the proof

Anchors the playbook

Any inspiration becomes an itinerary

Start with an unstructured travel input—a blog, photo, Reel, map list, preference, or question—then make Mindtrip's value visible by converting it into places, maps, and an editable day-by-day plan.

OpenAI, external AI and travel educators, a destination partner, and Mindtrip's owned account · 14 creators · 23 posts

Capture context, then hand off the work One source to multi-output inventory

Open the full playbook →

297@Germanaicreator

this post · other creators who ran it

Who made ittheir public relationship signal, and the receipt for it

@OpenAIyoutube · 2M followers · median 15K views · n=60best post1.7K0.1×views · for MindtripUnknown The exact OpenAI DevDay Community Spotlight visibly features Mindtrip, but the page exposes no paid-promotion label or contract evidence; event inclusion proves attribution, not payment: https://www.youtube.com/watch?v=Xjl2SqqrA4sOpen creator →

Views and likes are public counts at capture time. The multiple compares this post with the creator’s own recent median, not a claim about what caused its reach. Read the methodology →