Co-Founder @hiavaofficial AI-native Family OS + Porsche nerd @porschenotes also 🦞Ex: AI @Google, CPO Motorsport Network, CPO @VoxMedia, Product+Eng @ESPN
Brooklyn, NY
Joined January 2007
- Tweets3.9K
- Following3.5K
- Followers1.4K
- Likes3K
Pinned Tweet
AVA is live!!
The most important technical system underneath it isn't the model.
It's memory. 🧵
🤖 Made with AI
ALT Square AVA-branded systems diagram titled "Memory is not one thing." Four panels show the memory layers behind AVA: Household Profile for who matters, Insights for what happened, Patterns for how recurring situations work, and Instructions for how AVA should behave. The footer reads, "The right context. At the right time."
Everyone's about to ship a family plan. Billing is the easy part. The real spec is: whose inbox can the agent read, what can my partner see, which actions need whose approval, and what stays private to a teen. Families aren't 4 seats. They're a permissions model.
1
2
65
Last week's NYT piece on family AI agents (featuring @hiavaofficial) had the clearest prompt-injection ex I've seen, from an NYU researcher: a soccer league email whose PDF tells the assistant to cancel Monday apts. For a family agent that's not an edge case. It's the job. 🧵
Enforce the rules outside the model. "Never send without approval" has to live in code, not in a prompt, so no email can talk the agent out of it.
1
24
That's the line we hold with @hiavaofficial : it reads to help, and it doesn't act without you. The hard engineering is making that true even when the inbox is trying to argue otherwise.
1
26
That's our view of personalization: not a profile attached to a chatbot, but a system that gets better at helping your household over time.
We're just getting started.
1
26
AVA is now live on iOS. Here's a sweet founder discount code: hiava.app/?modal=true&utm_so…
21
two recent papers validate what we learned the hard way building Ava:
1. single agents outperform multi-agent systems under equal compute budgets
2. agent swarms without centralized verification don't just add overhead — they amplify errors across every handoff
coordination is a tax on intelligence. if your single agent is already good, adding more agents makes it worse, not better.
arxiv.org/abs/2604.02460
arxiv.org/abs/2512.08296
1
1
73
we spent 6 months on model quality at Ava before realizing the thing users kept coming back for was memory.
not smarter answers. not faster responses. the fact that it remembered their kid's name, their morning routine, what they said last Tuesday.
memory is the actual product. the model is just the delivery mechanism.
55
one super agent beats a swarm of little ones.
we tried multi-agent at Ava. routing between specialists, orchestrating handoffs, managing state across a dozen tiny agents.
ripped it all out. what actually works: one agent with a dynamic tool registry. it picks from hundreds of tools at runtime instead of splitting intelligence across ten dumb agents passing notes.
the coordination overhead kills you before the architecture does.
1
112
asked my 4 year old what he did at school. he told me about a kid who cried at lunch because his sandwich was wrong. he remembered the emotion, the context, the sequence.
that's episodic memory. not key-value storage. not vector search over a flat log.
at Ava we built our memory around episodes. temporal context, emotional valence, causal links between events. the difference between "user likes coffee" and "user was stressed last thursday because daycare called during a board meeting."
one is a preference. the other is understanding.
2
69
Joe Alicata retweeted
Great overview on QSBS (and the cascading effect bad tax policy could have on the NYC ecosystem). Thx @StevenFulop and @Partnership4NYC for your advocacy here.
At some point, this stops being about “taxing the rich” and starts being about shrinking the base that supports jobs, investment, and growth. Taken together, this isn’t a strategy to address affordability - it’s a reaction to polling.
If New York eliminates QSBS, startups won’t phase out - they’ll leave immediately. That’s not theoretical; it’s predictable.
Step back from the politics and ask a simple question: what happens to affordability if we systematically raise costs across the entire economy on job creators making NY an outlier even vs neighboring states?
We’re talking about:
•startups losing QSBS
•large employers facing the highest corporate tax rates in the country
•mid-sized businesses hit by PTET changes
•high earners already paying the highest income taxes
•proposals to push minimum wage to $30, hitting small businesses hardest
The most important point is that even if New York did all of this, it wouldn’t solve the problem. Without real, structural changes to the cost side of government, any additional revenue will be gone within a few years.
That’s the core issue.