Ladies in the Building, Building Ai. How?
š¤·āāļø ESTIMATED READING TIME: FORTY MINUTES
Disclaimer:
Everything below is a mix of what I observed and heard during the event. The goal isnāt to pinpoint "who exactly said what," but to share (usually) an outsider's view and overall perspective on these industries. Iām not here to act as a definitive firsthand sourceāreaders should do their own research. I hope this inspires you to attend events, explore new industries, and hear what leaders are presenting. These notes combine my observations with thoughts on how things could run smoother and how ideas connect (IMO). Iām not an expert, you know? Just hanging out in the room with them. Enjoy!Topics: Entrepreneurship, Artificial Intelligence, Mindset, Building, Publishing, Problem Solving, Customer Satisfaction, Ai Solutions
This event is two truths and one lie, okay? I think that women entrepreneurs are super inspiringā¦ I think that artificial intelligence is super interesting and important and game-changingā¦ i think that the main sponsors of this event are full of integrity, honesty, kindness, care for humanity, and beyond. Okay? But there are so many reasons to attend this and I always want to be proven wrong about things. You know? I am ready for things to be changing and the world to start healing. And I think itās happening bit by bit, brick by brick, day by day. So, this event is worth my time, if you ask me. But, donāt be surprised if some crazy things are saidā¦ butā¦ letās also not expect it. I could imagine this event is just totally inspiring the whole time. and hope so. Letās seeeeeeeeeeeeeeeeeee!!
Why Attend: My goal is to help educate the future workforce - and kids/humans - in a way that is empowering, teaches longterm thinking, compounding value, and more. I just have such clear vision for how great hte world can be and tiny ways we can get there. And it is fun to talk with Ai about these dreams and realityies, cause itās super encouraging. Evne has some ideas I didnāt think of. And itās super useful as a sounding board to be brave and keep steppijng forward as I keep building out my dreams and visions for the future. So, hearing women who have literally created their own Ai businesses (I guess thatās what this is?) - of course that will be useful.
Event Reviews Overall: Venue (5/5), Food (2/5), Speaker Content (2.5/5), Networking (1/5), Likeliness to Return (5/5)
Event info soonā¦. until then - plan yaā calendar, right? Finding things last minute can be stressful, but having things ahead of time is great. What do they say? The best plans are cancelled plans - so you get good plans no matter what, with this strategy. Go learn from reality!! :)
Photo Collage & Commentary
Notes from the Event
I found a cozy seat - now theyāre all introducing themselves t the panel. But the first person wasnāt an entrepreneur, she worked for a big corporation, and then I realized everyone was the same. So thiw asnāt a bunch of women ai founders- just women who work in ai.
But the pizza sponsored by this company is loud to setup, as they unpack it. RIght as the speeches begin.
17 pizzas in total
and btw i dontā like these sponsors cause of their crimes in drug trafficking, funding questionable islands, financial manipulation, etc.
OMG its so hard to hear her - the mic isnāt working but now it is. squeeky. ouch.
And then you can see the Barbucks HQ building in the distance hahah and mountains. SO seatlte iconic.
Data science + data engineering at a big LLM. Sales engagement, product experiences, research, optimizing, forecasting. Also was workign at big corporations like huge companies weāve all heard of.
Her first ai project was at another big corporation.
All the cpus in your phones and gpus are here
OMG itās so funny cause some people arrive late and want pizza - and everyone else is sitting down and the speeches start. SO yeah now the late arrives get pizza. But I have a dinner meeting tonight, so idk if Iāll eat any pizza. Probably not - or maybe only one person.
People cheer in the crowd about workign for robot companies that you can talk to the answer you from little circles.
Next lady worked at a studio, a tool for companies who want their own copilot and expecting Microsoft copilot. Itās been crazy for everyone - a crazy last couple years.
We started workingā¦ Curious about China
Looking at the conversations going on overseas with WeChat. It evolved over hte year. Cool, not cool, then cool again.
yeah wechat is pretty awesome cause you chat and send money and literally do everything in it. its kinda amazing, in china. you can run your whole life off of wechat. but also they have your entire chat history right there. how do i know? :)
One of the first ai projects, workingfor a group and started some labs. Social software (this stuff sheās talking about I donāt understand much but she said she started building infrastructure for game consoles to predict your game and gameplay)
Thanks for all being here, you all coupld probably be up here too. Is that true? Is this a smart crowd? Hahah - invite me :). Jk. Iām so tired today. And hearing these ladies introduce themselves is a little slow. Iād rather just talk about this atmosphere. Some people are really dressed up, some more casual. Probably lots just came from work. Itās pretty crowded and all women here.
Iāll write more once they talk about more, but right now theyāre just telling niche stories about their times at work.
Also, Iām really realizing that sometimes panels can be tough cause its like so many people talking and all want their answers but not the best at sharing time - or taking enough time, idk.
This girl talking now talks very quickly. And Iām getting distracted thinking about my day yesterday. My daughterās old daycare review I updated yesterday and how crazy it was. Both what I wrote and what I experienced. This teacher was screaming bloody murder at the kid and I thoguth she was goign to hurt him. Just from what I was seeing and hearing and observing and then its just how it sounded. So, I updated my review and I was really getting liek back in my dhanghai days when I was āmother wolfā in a play. Two year. I felt protective like a mother wolf of those dear kids.
Oh and Iām happy there is a lady getting pizza. I think Iāll get myself a piece now too - then I wonāt go into the night super hungry haha. This is a good idea. Iāll go get pizza. Then eat, then keep taking notes.
The pizza is undercooked!! Canāt believe it. I wonder what company this is they got it from. SOmething local.
What are the criteria that had you lean in at a pitch and decide to fund it?
What are you solving? The urgency of the problem youāre solving.
Does the buyer care?
Is now the right time?
Whatās different about tech to let you innovate?
Why you? Why your team instead of others?
How do you build over time?
Figure out what matters for a C-stage investors.
Value and execution. Also, in the Ai fundsā¦ having a view on (idk what)
What applications are getting more funds? Which arenāt?
Okay. I got so distracted cause I was emailing this band I used to love in high school . They are back!!! Hahahahah they are playing again. And I loved them in college. And yeah, tehy said Iāve been in the band since 1999 and I can do a duet with them int he future. So, Iāll hold them to it and invest in them post-MOASS and help the whole world hear them. Theyāre so good and itās unreal they still exist hahaha. Okay - back to this event.
If youāre building product experience at an edge, youāre onto something
Do you have a framework when founders need to use models that are ready versus liek (idk what she said) - how shoudl tehy be thikning about that?
Fine tuning, how much do you care about the UX? Execution of the UX, the end product. Espeically for application level products. Do you care about being accurate, more transaction, can you just throw something at it? Or do you need more personality hold into waht youāre getting from the model? Thats the general rule of thumb for what to do when fine tuning instead of just kinda (idk what she said)
Look at the evaluation.
Sometimes the base model is sufficient for use cases. Using those emails about the quality of responses, actuality, safety and the tone. It can be a tool to use the base models sufficient for your needs if there is more beyond that.
Going from emails to experience (idk what sheās saying omg hahah this is kinda like so much tech talk)
Two questions: is there such a thing as āgood taste in codingā? How about in eval? And related to that, is there also maybe data science done wrong in some ways? In terms of not getting all the evals, all the data you need, and even peopel misinterpreting data as the field evolved quickly?
Yeah, I like to use emails + experimentation. Sometimes there is a lot of benchmark. You see examples of that. Making sure it continues to not
ā idk - they all talk so fast and the audio keeps squeaking super high pitched.
80% would be a good rate over liability match, LLMās can not get beyond that.
The question is for you, Iām a founder - first of all, I have all of the for me. Youāre all on the same panel, weāre all friends. What shoudl I use for my company?
The audio is so painful.
She says sheās biased
This audio is squeaking so much.
Usecase, all these tools are made with a specific user in mind. Are you looking to get in the first 0-1 MVP where you want to have super easy to get going, easy to do something. A person like you? Or do they already have a well defined product? - Now I realize I need to fix two things online, Iāll fix them and then come back to this - I KNOW haha but I realize now something so important isnāt right on my website and at any moment a feature will go up for sale on a 3rd party - and I dont want the initial first customers to get a bad product. I realize this now. Itāll take a minute to fix. Soā¦ lemme do that and then get back to this.
but then I got an email that the company removed 8 of my classes from their website due to small errors and thing I need to fix, geeze. So Iāill have to do that tomorrow.
-Okay back. And the microphone is still squeaking!
Cloud computing costs a lot of money. Machines cost a lot of moeny, so you need to decide what kind of performance you actually want to see. How much juice do you actually need? Test and iterate as much s posable to see how little you need.
BTW iām editing/typing this on the subway and the guy who was sitting next to me, after a lot of peopel got up, i asked him to move to another seat - cause there were so many around ,so i could take out my laptiop but he keeps looking at selfies of himself and videos, and now he keeps stretching his feet out towards me . Heās in a full fledge like gown/uniform/costume thing - for an event, heās rewatching pics of his time on stage. And his shoes are like slippers.
Iām like dude, idk if oyuāre doign all this on purpose - but are you trying to get my attention again with your feet??
Cause youāre stretching it from your chairs that are across the aisle, back to where you were sitting next to me?? like what hahahah - just stretch your feet into the aisle, not by your old chair by me? Or idk - are you trying to get my attention? You didntā really smile at me when you had me sit by you in this window seat, but then when everyone left you just stayed there - so its why i was like, can you scoot over? So i can work? hah. omg. people. who knows?
Thatās one thing I learned overseas.. sometimes yo'uāll just never know whats going on. And youāll never find out or get a resolution.
Anway, just expericing that right now. But my stop is in 2 stations, so probably Iāll have that resolved in time. Letās see if I can complete editing this blog int ime too.
Whatās the problem youāre tryin gto solve? What are you optimizing for? And whats the ROI?
Am I trying to make fetch happen? She asks.
I thin that buildin an Ai company is cheap these days. Especially int he application space. (I see a lot of peopel nod their heads). There are so many ai coding tools out there, lots of the shelf models work well.
By now like half hte audience isnāt paying attention. Or maybe tehyāre taking notes too. But defiantly some people arenāt and just browsing linkedin, but about these women. SO itās related.
One smart thing is not just saying, āIām starting one modelā - the smart founders have a really complicated system with a lot of models and with each usage theyāre trying to decide to feed it into a beefy or liteweight model. Thatās what I need.
-Understand how you model and product behave. When you move into teh agentic world, you know how to do it without ruining the customer experience.
As you scale, you can find larger data nad user base.
I see a girl arrive late and now she feels so awkward- the seating here is a little uninviting. And the āgoing to grab a seatā culture is so funny here in America. Ib et its different in different countries.
One girl arrived late and was a little rude to tlet this next girl who married late go by her to sit down - but this next girl doesntā know the first girl was late. BUT THEN the first girl who was lateā¦ guess what she has? The cutest shoes! Hahah.
Can you tell that I donāt now what theyāre talking about much in this panel.
How much did the models cost? What purpose do they serve? Tehy orchestrate workflow in that system model. But then a company is building ontop of the model, really complicated reinforcement learning for the usecase cause they have to be almost 100% accurate. Lots of compute costs and lots of expensive talent. It really depends on waht youāre serving, which industry youāre serving.
What can the models do you you and how much is needed to serve the usecase.
Across all the Ai lenses, divide your view between training and serving.
Some of the more sophisticated companies are using multiple models and routing based on the usecase, baed on which model is best. Routing based on the user base. Free users, paid, higher tech users . They may wan to serve the more expensive models depending on the different quality of service.
lol my neighbor is stretching out his foot towards me again. idk if heās trying to get my attention. but its working, but iām pretending it isnt.
Mobile and latency, Alexa, teh different services having different requirements.
Each individual is now thinking about every task: can I do this with ai? And how about managnig a team of Ai. Each person is now becoming more interdisciplinary. It helps you strattle in that way.
Look at the percentage of Tass people are usign Ai for and then map it to the department of labor.
-Then look at the task each profession requires, there is intersting info - developers are nearly 4% of the us population, but 37% of tasks are being automated or augmented with Ai. So, if you think developers will start to get more productivity faster, that may change teh ratios in your teams for example.
We all need to grow.
Old people basically use chat gpt to replace google search.
How do we get into the mind of younger peopel who grow up with this. Itās just so different.
literally exactly the question Iām asking, too.
At NYU they teach programming. I donāt need to teach progtramming anymore. Have them vie code. Just start to have an idea. I need to rewrite all of my classes. We need to challenge ourselves to keep rethhking what does it mean with so much different.
Some peopel really resist change. Probably yhte smarter and more established you are the harder it is to admit youāre dumb and donāt know.
If you know someone who thinks theyāre an expert, no noe was doing this more than three years ago. To be an expert in Ai is real third. To know customers, UI , all teh data science. No one knows everything. Just, you know. Just start making stuff and try and donāt stress out too much. Nobody knows anything. For certain personalities, you know, just partner with them. Just go try it out. Pick something that is a pain in your butt and try to solve it.
Okay - are they wrapping this up? Feels liek time is flying but also not at the same time.
Oh yeah now itās audience questions. I have no questions
Audience - thank you so much for this panel. Woman works at big searching company in safety and ethics. Her question is whether you can give perspective as an individual in the sense of your company on the importance of developing and eplyoing safe and trustworthy Ai from the stage of being a startup to a frontier
Lol they canāt answer this question too honestly, right? Thyāe like in a crowd. They have to say the right thing, sortally social-norm-ly
Safety is of course the core missino to help us transition in a safe way itās been a big part of the approach.
Broad awareness for the work being done to work through it ourselves and develop a product with high quality
Safety Iāve worked on my entire career. Looking at the safety of content, things shown there. Some of the safety and responsibly ai there.
Just thinking about the types of harms that we look for and itās also been compared to a particular focus to cyber security, bio harms, nation state threats.
There are councils across these labs figurin gout how to team together agaisnt these bad guys. Itās critical to talk about hte good but also teh enabling of bad actors.
Safety always comes up, itās such an imprint topic.
Humans are not always that creative and repeat things a lot (dude!! Donāt underestimate humans though - but there is some science and data to this)
Good work comes out of these things we create, usign these models nad understand waht its real and how are we creating from the hardware up thats baked into our systems.
We have another friend in our larger group tat has a startup focused on don this.
Itās an awesome time to be in Ai security. Things need to be rethought for sure. Itās not just bad actors, itās nation states. Why go to war when you can attack bots. You know? Whatever. So, I think itās really interseting when you think about automating more things with ai. The risk is much higher and itās a super intersting space. Nobody ever has a magic wand.
Now sheās in th audience and its really hard to hear waht hte question is and whats going on.
One lady says this is tragic, āsheās been rying to get me to talk the whole timeā idk - an audience member who is smart, I guess? As if she shoudl be on the panel too.
This woman works for an ai company, focused on thinking about how to bring business compliance to ai. So, you have a short-term goal and then some other point the longterm goal.
Short term these businesses are responsible to be compliant with the law. Sometimes there are finacneil services, healthcare, import/export, cybersecurity. There are almost draconian rules and laws that are different across different jurisdiction. Not even international. Can you rely on Ai solutoin to run your business within the confines of the law?
Most peopel donāt know.
You canāt test in teh traditional way having peopel show up and bang on your products. ITs gong to evolve over time. You have to ues a different set of tools and skills. Who are you solving? What are you solving for? We focus on smaller segments of that Its fascinating.
Now the girl who was let and got hte cold shoulder from another girl says sheās workign on making an Ai family wellness coach. One of the new ideas recently come about is the idea of managing agents. Having agents with frameworks, a number of startup are looking into this iea. I recently had a chance to discuss, features on a podcast and this ethics and - idk - omg, sheās talking a long time
- I was curious to hear your perspective on how do you evluattae, um, teh complexity of new agent frameworks, teh failure modesā¦ as you look into new investments and new products that youāre starting with.
She sys she doesntā know how to answer her question . Itās really intersting. Something just came out antis a different character than before. Itās interring, is the personality part of hte model or a layer ontop. Now it is. But (the mic squeak os loud) - the idk what sheās saying
Yeah idk this. Theyāre just talking. And now really half the audience isnāt listening. I think theyāre talking pretty niche. Lemma count how many peopel who dontā seem to be listening. 15 right now I see. Also probably cause this audio! for me its painful time to time.
Be flexible how you think youāre deigning your system to work with models. Itās going to be more complicated. You canāt jsut make the model. Things are changing. Just ake the pieces and be willing to change and stack it across.
This is very true and applies to what Iām doing. Omg the future IS UNKNOWN.
The audience has a lot of questions - SO MANY. But I may get going soon
One lady works at a big company. But yeah I might go early. And then walk to my next meeting.
Yeah this speech was good but I think not exactly my world - and - a bit late. And Iāve got more to do tonight. And panels Iām not as into as other things. But itās still good.
Now the emcee just is stilling everyone to network and hope ecryone can share and support each other.
Thereā a better path when you search of your own path in all this.
The mic is so squeeky so much this whole time. Itās very subtle, but my hearing is very sensitive, so I can hear it and itās just painful, Iāll keep going.
Women are focused on everyone on the planet, the customer. We have a EQ no Ai can replace. The human to tech interaction is a feel under invested in, but the future. Cause the way we can talk to computers today, itāll keep evolving. Thats my fuel.
Literally, hahah, this sentence above I coudl have said myself. but I wanted to get going -
Meeting After
I was given great advice just to sell my books directly online. So serious. But, I think I will.
Along with the website Iām using - which then last night kicked 8 of my classes off their platform. Iāll admit, I was giving them some sass behind the curtian - cause they kept tellign me to change one single class and Iām like, no way, iāve done this class in teh same format as all my classes. I pushed back so many times at them and said - GO looka t my classes and tell me how this single one is any different
Then guess waht!? They realized I was right, Iād done them all wrong, so now i have to actually go fix all my classes- so, its good cause it pushes me to add more to the class package. BUT I do roll my eyes a little cause they by mistake passed 7 out of 8 of my classes, to only then realize they werenāt being clear with me.
So, this idea last night to also branch out and put my stuff other placesā¦ okayā¦ā¦. i guess soā¦.
Also, lately iām tryingto be in a bit more of a āseason of observingā as a default.
Overall Event Reviews Elaborate:
Venue (5/5), this venue is fun to get to, itās a nice size, and itās just pretty ācoolā - but their audio was painful.
Food (2/5), pizza is so unhealthy and even a little contriversial these days. I was surprised the sponsors bought it.
Speaker Content (2.5/5), so much of it I couldnāt keep up with, panels tend to be tough to keep up with too, but it still had good points and useful takeaways. Lots of reassuring moments for my situation.
Networking (1/5), No, I didnāt talk to anyone. I think if Iād arrived a little early I could have, but there was no intentional discussion going on, teh seating wasnāt set up for much conversation, and then the panelists talked the entire time I was there. So, it was easy to talk to no one. I recognized the woman who checked me in though. Sheās nice.
Likeliness to Return (5/5) - yeah, Iām working closely in this field and this venue is very relevant and prominent in the city. I think its a good place to visit often. And itāll start to transform this city and keep adding characfter.
Until next time, I wish you the motivation and success to search for opportunities around your area. Search and explore: Who is out there giving talks? There are new things happening all of the time.
Find relatable or interesting topics you like and check them out! Maybe even something hosted at a cool venue, if thereās no other reason to go. Letās see what you can learn and discover not too far from home. š
