Cameron Michael Malloy

Living life to the fullest
@trystudioai | YC S26
@ucberkeleymet

[Cameron Michael Malloy @CameronMalloy_]

What if you knew exactly how society would respond to a decision?

Today we’re launching @tryStudioai (YC S26)

Studio lets brands test a product, price, campaign, policy, or message against a simulated population before putting it into the real world.

Change the decision. Run it again. See what happens.
Keep going until you find the version most likely to create the response you want.

Our predictions have reached 95% match accuracy against live KPI outcomes with F500 brands.
Engineer response before you deploy it.

Email us at [email protected] or check out https://t.co/FZyh46jBW8

[Cameron Michael Malloy @CameronMalloy_]

@pinakpaliwal @speedrun @andrewchen @KatiaAmeri @justmazer Lets go PINAK!

[Cameron Michael Malloy @CameronMalloy_]

@malhalla06 @TheAntheaGuo @lumeriaskin congrats!

[Cameron Michael Malloy @CameronMalloy_]

Had some great conversations and an amazing time at @Ai4Conferences. Face to face conversations have way higher impact and conversion than any online ones

[Cameron Michael Malloy @CameronMalloy_]

@pragunseth Congrats!

[Cameron Michael Malloy @CameronMalloy_]

@pinakpaliwal Inspirational stuff Pinak

[Cameron Michael Malloy @CameronMalloy_]

First post, but wanted to share that @NikitaMullangi and I are building @tryStudioai, backed by 𝗬 𝗖𝗼𝗺𝗯𝗶𝗻𝗮𝘁𝗼𝗿!

We allow consumer brands to predict how new launches will perform by simulating KPIs like revenue or churn.

Excited to be creating a more informed future :)

[Stephen Xie @stephenx_]

interesting things I noticed:

1. model can optimize to produce gibberish (!)
2. GEPA can produce long chains of domain-specific few-shot examples
3. diverse data are needed to generalize ood
4. reward shaping incentivizes faster convergence at the cost of early plateau

(3/3) 🧵