Catch every move your market makes @satellyteai | Math & CS @ucla
steal competitors' leads →
Joined March 2025
- Tweets6.7K
- Following928
- Followers1.3K
- Likes10.7K
Auth page currently at ~200 daily visitor at 30% bounce rate.
Any suggestions to decrease bounce rate?
Does distillation not count as research?
There’s been a lot of speculation about where we stand on open-weights models. We’ve outlined our views in full here: anthropic.com/news/position-…
Is anyone surprised?
Beginning July 20, Claude Fable 5 will be included in all Max and Team Premium plans, at 50% of limits.
Pro and Team Standard users will continue to have access to Fable via usage credits, and will receive a one-time $100 credit.
Demand for Fable has been challenging to predict, which is why we rolled it out to subscription plans in stages, extending access several times as we secured additional capacity.
Open source frontier models might actually be the future of AI.
Introducing Kimi K3: Open Frontier Intelligence
🔹 2.8 Trillion Parameters, 1 Million Context, Native Multimodal
🔹 Kimi Delta Attention enables up to 6.3x faster decoding in million-token contexts
🔹 Attention Residuals deliver ~25% higher training efficiency at <2% additional cost
🔹 Built for long-horizon agentic coding and self-evolving workflows
Kimi K3 is now live on on Kimi.com, Kimi Work, Kimi Code, and the Kimi API.
Open Weights by July 27, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.com/blog/kimi-k3
YC startups live in the future, nearly 9 in 10 of the latest batch are AI-native. So I tracked 271 roles on their job board, and GTM looks nothing like 2024.
The sales rep who just closes is disappearing.
They're replaced by someone who builds systems to connect tools, run agents, and generate pipeline from the CLI.
Six GTM clusters came up again and again. Here's what YC startups are hiring for right now:
1. Founding GTM (14 postings)
The first revenue hire. Turns founder-led sales into a repeatable motion, finds the segment, sharpens the message, then builds the team.
2. Forward-Deployed Engineer (12 postings)
Sits between sales and product. Embeds with customers, builds the integration, makes the AI product actually work inside the account.
3. Growth Lead / Engineer (12 postings)
Instruments the funnel, runs experiments in code, automates onboarding and lifecycle, ships self-serve flows.
4. Performance / Growth Marketer (19 postings)
Runs paid and content experiments, automates campaigns end to end, owns the funnel from impression to pipeline, with AI doing the production.
5. GTM Systems / AI GTM Ops Lead (3 postings)
YC's version of the GTM Engineer. Runs AI agents for enrichment, routing, and follow-up. The tooling that lets a small team sell like a big one.
6. RevOps Lead (3 postings)
The revenue data spine. CRM architecture, pipeline accuracy, attribution, and the integrations every other motion depends on.
Every one of these rewards someone who can wire tools together, not just send emails.
The sales team that only dials and emails is already a generation behind.
It's coming home :')
One signal on its own is almost always meaningless.
Which is a strange thing to say when I sell signals for a living.
On its own, each one is just a guess.
Take a new VP of Sales. Is that a company gearing up to grow, or one that just lost the last one and is in freefall?
Same signal, two opposite stories. You can't tell from the signal alone.
Here's what actually works: stacking signals until only one story is left.
1. A company subscribes to a new outbound tool. They're investing in outbound. Could mean almost anything.
2. They're also hiring SDRs. Now it's not just a purchase, it's a team being built around it.
3. They closed a round last month. The budget is there and the mandate is to grow.
4. A new VP of Sales started six weeks ago. New leader, reshaping the stack.
5. News breaks that they're expanding into a new market overseas.
Now the picture is unmistakable: new money, new team, new leadership, and an aggressive growth mandate to back it all.
Each layer doesn't just add data.
It removes a story that isn't true.
One signal is a guess.
2+ signals are a target.
That's what Satellyte does.
It joins these signals at the company level, so you're not chasing one ambiguous data point.
You're watching the whole picture form.
