## Inviting Speakers:

- Reach out to potential and ask if they are interest in giving a talk at our events, add the company or person to our spreadsheet and mark "asked".
- Once they indicate interested, we then input their information into the spreadsheet and changing the status to "tentative".
- Once they have agreed to give a talk, set them to "Agreed to Talk" and Create a whatsapp group. Whatsapp chat groups are for more rapid communication closer to the event. Especially for last minute issues like speaker cannnot find the venue.
- Once the Whatsapp group has been created, set them to "Pending Talk Details" and sent them the "Admin Information to confirm speakr" form below.
- Once the speaker has replied with the completed form set them to "Talk Details Obtained"
- Add the talk details to the tracking google sheet, and the event page on Luma. Once the speaker has been added to luma, set them to "Added to Event Page". This is the final state the the speaker is properly registered for the event.

#### Speaker States
- Asked
- Tentative
- Agreed to talk/Pending WA group
- Pending Talk Details
- Talk Details Obtained
- Added to Event Page

### Scripts for common messages to send to speakers

#### Admin Information to confirm speaker

Could I trouble you to help me fill in the following information for our event page?
Speaker:
Affiliation:
Title of Talk:
URL (optional):

#### Whatsapp Group Creation
Can I have your phone number to create a whatsapp group for more rapid communication closer to the event?

#### For questions about the lightning talk
The lightning talk is 10 minutes per speaker. You can talk about anything your passionate about as long as it is related to AI. Demos are highly encouraged. However, our rule is that there should not be any product pitches. This event focuses on the tech, not the business.

#### Email template:
Hi <Speaker>,

Our next Singapore AI Showcase will be held on <date> (<link to luma event>), would you be available to present then?

If so, could we have a few details for the agenda on the event page:
1) Name of presenter
2) Organisation
3) Title of talk

We are also happy to have a link to your website on the event page.

## Promoting the Event:
- Youtube Livestream should be configured via hello.jan.ai
- Thumbnail for Livestream (To edit on Canva)
- Promotion should be done 1 week before then 1 day before (at the minimum)

## Presentation preparation:
- There is no hard rule, but a good way to do it is just to show a excalidraw panel in the background with the schedule for the talk. We should also include the QR code for feedback to NLB onto the excalidraw panel.
- We will have a time keeper and notice the presenter at 5 minutes and 1 minutes left mark

## On the day procedure:
- We will usually ask guest who come in early to introduce yourself to the rest of the room so that everyone get to know each other and connect afterward.
- When the event starts, introduce ourselves and our host NLB for around 1 minute at the start.
- Pass the time over to our NLB host for 5 minutes so they can introduce themselves and gather feedback.
- After the speakers are done, we thanks everyone for their participation and the host for letting us use the venue, ask them to fill out the feedback form for NLB and ours. And finally help the host to clean up if there are any.
- Showcase Slides should be standardized and prepared beforehand
- Final slide should have QR code for next showcase and menlo.ai site

## Showcase Technical Setup (Updated June 18th)
- Daniel's Logitech Brio WebCam
- DJI RX Mics
- USB C to C Cable
- USB C to HDMI Adapter

# Livestream

## Setting up for AI Showcase

We've started using OBS for recording / live streaming since the July 2025 AI Showcase, but livestream numbers have never been super promising (maybe 10 viewers at most.)
Regardless, it'll be nice to do _knowledge transfer_ of how livestreaming (and recording would work).
### Set up

#### Configuring YouTube
1. Ensure that you have the login credentials to `[e-mail]`. You would utilize this email for the setting up of the livestream via YouTube
2. After getting the login credentials, `studio.youtube.com` > Stream > Set up Live Stream
3. Off the top of my head, there are different configurations that you would have to go through (lost access to `[e-mail]` at the time of writing).
1. Configure the thumbnail picture through visiting this [Canva page](https://www.canva.com/design/DAGSPJv6AaY/lPTXLl-crifMp2S3Cg0o_Q/edit)
2. Add this thumbnail picture to the YouTube config
3. To enable live chat, ensure that the settings are set to "Not Suitable for Children :clown"
4. Set the time for when the livestream would start
#### Configuring OBS
1. When configuring OBS for the first time, you can follow [this video guide](https://www.youtube.com/watch?v=HMjtJBYAERE) on how to configure OBS with YouTube through connecting the proper account (should be the same `[e-mail]` account as well) on your device. Please ask Emre for the login details in case you get logged out!
2. After getting the correct configs on YouTube, I'd recommend starting a few livestream test runs (of course with the visibility set to `Private` / `Unlisted`) just to get a feel of how OBS works. Intuition definitely develops over time!
3. The showcase technical setup specs are as follows:
- Daniel's Logitech Brio WebCam
- DJI RX Mics
- USB C to C Cable
- USB C to HDMI Adapter
- Camera mount (in Asimov office, I think)
4. Unless SG AI Showcase becomes much fancier next time, 1 scene (Default: Scene 1) should suffice. Would add the following sources to capture video feed:
1. Under the `Sources` tab, click on the `+` button to add `Video Capture Device` and select `Logitech Brio` (or whichever camera source that you have added)
2. To add the DJI RX Mics, please follow [this guide](https://www.youtube.com/watch?v=F1yly2Kjh3Y) to set it up. At a high level: plug the type C (or lightning, whichever you're using) connector into the main MIC RX receiver, before plugging it into your computer. You can then take out the side mics (and turn them on) and there should be audio levels recorded on the main MIC RX receiver.
3. Under the `Sources` tab, click on the `+` button to add `Audio Input Capture` and select `DJI`. Great! Now you have everything configured!
> Sometimes, there might be a need to reconfigure some of the settings for the display on OBS to be displayed correctly (i.e. flipping the scenes, but that can be done through experimentation)

> Daniel C would usually record the entirety of the SG AI Showcase, unless some presenters explicitly state that they do not want to be recorded. Recordings can be found at your `Movies` folder (for Mac users)

## Setting up for Jan Community Calls
1. Under the `Sources` tab, click on the `+` button to add `Browser`
1. Paste the Google Meets link accordingly (and perhaps ensure that you are logged in)
2. Ensure that the `Width` and `Height` of the `Browser` screen matches that of your canvas (1920 x 1080).
2. Under `Browser`, ensure that `Control audio via OBS` is selected
3. Under Audio Mixer (for you to hear the output, click on Browser) > `Advanced Audio Properties`, there should be an option to turn `Audio Monitoring` to `Monitor and Output`
4. Start Recording! And then upload whatever recording you have onto Menlo's shared Google Drive after you're done :)

------

# April AI Showcase Learnings:
Rundown: 309 registered participants on luma, 10 speakers total

[Daniel's April AI Showcase Sequence of Events / Internal Checklist + Script](https://danielching.notion.site/April-AI-Showcase-1dd2ffbab35180409057f9f270042357?pvs=4)

## Event hosting overview

- Personally felt that event went smoothly barring any hiccups in live-streaming, there were okay transitions between speakers, and we ended slightly before 9pm which was the hard cut-off which was ok!
- Good turnout!

## Event hosting feedback

- Some feedback from friends who attended were to: 1. make transition to the next speaker smoother -- thank you xyz for that presentation, next up we have abc sharing about 123. <- was hyperfocused on making sure that we hit the time and the next speaker set up quickly so that the whole flow wouldn't be too disruptive for the audience, but would definitely bear this in mind for the next run
- On speaker selection: i think there could have been better filtering internally for folks who were presenting, maybe the slide decks could've been vetted so that it was more tech-demo focused rather than sales-y pitch decks, so a goal would be to get more eugene cheah like talks XD
- Focusing on more audience participation could've been better // what are some ways to foster building a stronger tight-knit community? Steve Gao from bytedance mentioned that he loved the sf-like energy from this event and i think that's a good sign

## Misc. Learnings

`Setup:`
- Camera: Sony ZV-E10 II + stand
- Audio Capture: DJI Mic (not sure which version) up until Dr Zhang's talk then Daniel's Macbook Pro + Sony ZV-E10 II for environment noise capture
- Livestreaming software: OBS -> YouTube (ask daniel c for config help!)
- Set the livestreaming to be at the highest quality and forgot how to change it back midstream (also because bandwidth was too little to handle both speaker setup and livestream troubleshooting concurrently), which caused overheating issues: **`Menu -> Globe Icon -> Streaming -> USB Streaming -> Select lower res / frame rate if overheating`**

`Problem:`
- Camera kept overheating and shutting off mid stream, had to manually turn it on and off for sporadic feeds.
- After changing Menu -> Image Quality / Rec -> File Format to XAVC S HD it might have improved, but not conclusive
- Connection was wonky both ways (after reboot, image froze, had to unplug + plug it back in for OBS to keep streaming)
- After Dr Zhang's talk, I forgot to switch on the DJI mic that was attached to speakers and so there was no audio coming from it into the livestream

`Solution:`
- Lower resolution (see setup) - need to try 1080p and 720p to check if the camera can survive the entire 1.5 - 2h of livestream.
- Get a backup livestreaming device (to investigate -- e.g. using iPhone) + dual concurrent streams (if possible) so that if one drops at least the other is still surviving (i will figure this out with whoever is hosting May showcase)

`Config notes:`

- All of the USB C to C wires except for the one in the green ziplock bag (Deluxe sth cable) are unable to handle outputs of >720p. So for livestreaming if we choose to use the camera we must use the one in the green ziplock bag in the hardware cabinet! can handle speeds of <= 10Gbps.
- Other configs are in this [notion page](https://www.notion.so/jan-ai/Event-Livestreaming-1dd5b91df6b180a392afeda0f58de78c?pvs=4)

Things to keep:
- I really like how the DJI mic had superb sound quality that was independent of NLB's sketchy microphone + audio control. Only hassle is taking it off each speaker and putting it onto the next speaker, but that's about ~30s of troubleshooting if the mic is on
- Backup audio systems such as direct input to the streaming device (laptop / camera) is necessary in case the DJI mic runs out of batt / turns off, so am happy that I could fix that on the fly
- The camera would probably have high quality feeds, so I think we can keep that

`Things to improve:`

- Get a backup livestream device (phone)
- Ensure that speakers are in frame for readjustment of camera feed
- Need to teach / ensure that whoever is manning livestream has the prerequisite knowledge _beforehand_ to man the livestream but thanks @rach :) for helping out i would have died if you werent there, more on this the next message
- Clarify speakers' level of comfort to be livestreamed beforehand, Dr Zhang's request to not be livestreamed put some online audience members off + reduced bandwidth for sudden request in terms of turning on / off livestream + audio feed etc

(edited)

### TLDR:

- I think that emceeing and live-streaming simultaneously might not be a good idea for one person to do. It'll be much better to have at least 2 people doing it.
- AI Showcase could have the following people in the future: 1 for livestream, 1 for emceeing, 1 timekeeper (emcee - timekeeper should work closely together!)
- good first start, but lots to improve + stress test before the next showcase if livestream is something that we want to continue doing
- take pics!

`Other feedback: `

Yuuki:
- If we are going to livestream, have to notify speaker before hand so they know to avoid Weston situation again, just in case
- We have discussion before with @yipjiaqi regarding to make it more tech focus or not, and he said it is ok as long as it is less VC pitching but more like selling to customer and users. We do prefer people building cool stuff more, but they are usually far and few in between.
- I still stand by my first initial assessment from the last time that 8 is the best with + 1 reserved, allow us for more flexibility, and I don't think it is generally a problem ending earlier, more of a problem ending later.
- This one is just a personal experience share from me to @Daniel C, I think no need to focus too much on time at the expense of experience. Cracking a few joke in between transition while allowing the speaker to set up their thing is sufficient enough. I also screw up a bunch with the order, but I feel like those feel padding joke, thank you, and welcoming between speakers make the room feel a bit less tense

Akarshan:
- Also adding to this, we need to add a cooler to the camera to prevent overheating (or buy a video cam?) The mic audio should be directly routed to YouTube(or any other streaming platform) and we need to hire a camera man for zoom in. Bonus: If we can share the screen that would be wonderful!