Issue 39 — 2026-06-20
Amazon's chip business is taking a harder swing at Nvidia
Amazon is pushing to challenge Nvidia more directly by selling its own AI chips to outside customers, rather than keeping them locked inside AWS. The company has been quietly building custom silicon for years, but now it's treating its chip division as a real competitor to Nvidia's dominance in the data center market.
Separately, NASA picked Eric Schmidt's rocket company for a Mars mission, setting up a direct race with SpaceX. Schmidt, the former Google CEO, has been investing heavily in space infrastructure through his company's investments and ventures. The win puts him in competition with SpaceX's own Mars ambitions.
Why this matters for us: Amazon's chips and Schmidt's rockets are two sides of the same story — big money consolidating the tech and space we rely on, and the people who built that wealth betting they can do it better.
The genome's tangled DNA is tripping up AI
A new Quanta Magazine piece drills into why AI keeps stumbling on the human genome — and it's not because the models are dumb, but because they're looking at the wrong thing.
Most AI models treat DNA as a linear string of letters — the A, C, T, G sequence — the way they treat words in a sentence. But the genome isn't a string. It's a three-dimensional structure, folded and twisted inside the cell nucleus. Genes that sit far apart on the linear sequence can be right next to each other in physical space. That folding matters because it determines how genes interact, switch on, and respond to signals.
When AI ignores the 3D geometry, it misses the mechanics. A model might spot a pattern in the base-pair sequence and call it a finding — when the real action is happening in the folded structure, not the flat string. The fix is getting models to learn spatial relationships, not just sequence patterns. It's a small shift in how AI approaches the problem, but it's the difference between guessing and understanding.
Why this matters for us: The same pattern-recognition models powering everything from medical diagnostics to ancestry tools will be more reliable when they stop treating DNA like a sentence and start treating it like the tangled mess it actually is.
Embeddings: how AI remembers what matters
Computers don't understand words. They understand numbers. When you type a question, the machine turns your words into a list of numbers — an "embedding" — and matches that list against the numbers it already has stored.
Think of a family photo album. Each picture has a number that says how similar it is to every other picture. The photo of tía Rosa at the quinceañera gets a high similarity score with the photo of her dancing, but a lower score with the photo of the food table. The album doesn't know "quinceañera" or "dancing." It just knows the numbers.
Embeddings work the same way. They're a way for machines to say "these two things belong together" without knowing what the things actually are. A recipe for enchiladas and a recipe for chilaquiles get close together in the number space. A tweet about the migra and a news story about ICE get close. The AI doesn't need to be told this. The numbers tell it.
This is why your search engine can find what you mean even when you misspell things. This is why your phone can tell you're talking about "la migra" when you type "la migra app" and show you the right app. This is why you can search for "something to fix my tire" and get results about tire repair, not something about fixing your relationship.
The trick is that embeddings capture what things are for, not just what they're called. Two things can have different names but live near each other in the number space. Two things can share a name but live far apart.
Next time you search with your phone, notice what it finds. If it finds the right thing, the embeddings are doing their job. If it doesn't, you know the numbers didn't quite line up.
Why this matters for us: The way your phone and search engines understand your words determines what you see — and what you miss — in the stories that affect your family, your neighborhood, and your wallet.
When NASA picks a contractor, it's not just choosing a company—it's choosing who controls the next decade of space access.
— techcrunch.com
#nasa-picks-eric-schmidt-s-rocket-company-for-mars-mission-7c224dAmazon's AI chips are no longer just for AWS
Amazon is selling its own AI chips to other companies, not just using them inside AWS anymore. The company has been quietly building custom silicon for years — the Trainium chips for training models and the Graviton chips for inference — and now it's opening the doors wider. Nvidia, which has been the undisputed king of AI chips, is getting real competition for the first time.
Amazon's been in this game longer than most people realize. The company started developing its own chips in the early 2010s, initially for storage and later for general computing.
TradeWork: Jobs, crews, payments, and paperwork in one place
Painters, plumbers, electricians, and landscapers are still juggling estimates in notes apps, invoices in PDFs, crews over text, and payments by check.
The general-purpose CRMs are too heavy. The trade-specific apps are usually built for the office, not the truck.
TradeWork is a mobile-first work platform that puts jobs, crews, invoices, and payments in one place. It's built for trades crews who work out of a truck — not a desk.
It has bilingual surfaces so the office, the foreman, and the helper can all read the same job. No guessing what the other person meant.
Why this matters for us:
https://tradework.work
New media is getting stuck looking at itself
A new crop of media companies is wrestling with a quiet problem: they're becoming too good at reflecting their own tastes back to themselves. The result is content that feels smart but doesn't always land with the people outside the room.
The piece — written for TLDR Product — calls this "the trouble with mirrors." It's the same instinct that makes a creator's newsletter read like a diary and a podcast's audience sound like a single tribe. When media companies build around what they love rather than what their audience needs, they end up talking to each other. The audience becomes secondary to the mirror.
Monthly subscriptions — when to offer them
RevenueCat just published a guide on when companies should offer monthly billing. The piece walks through what drives the decision: how often customers use the product, how long the value sticks around, and whether people actually want to keep paying month to month.
The a16z article behind the piece looks at the broader shift — new media companies are testing subscription models as their main revenue engine, not just an add-on. A year in, they're seeing which formats and audiences actually stick to monthly payments and which ones bounce back to one-time purchases.
Why this matters for us: the companies building the tools and platforms we rely on are betting their revenue on our willingness to pay month after month, so how they design those subscriptions affects what we pay and what we keep.
Jenny Wanger Built an AI That Critiques Her After Every Call
Jenny Wanger built an AI that critiques her after every call. Not a polished dashboard, not a subscription tool. A disposable one that lives in her workflow and tells her what she did wrong — and what she should try next.
It's a small thing. But it points to something bigger about how software is shifting from permanent to disposable. Tools that used to stay in our pockets are becoming more like temporary helpers, useful for a moment and then set aside. Software is becoming less about owning a platform and more about borrowing what works right now.
This isn't Silicon Valley thinking.
Software is becoming disposable — and that's a problem for la gente
Startups are building software that burns out faster. Auren Hoffman's latest note argues that the old playbook of shipping once and compounding value over decades is cracking. Tools now launch, catch a wave, then dissolve — replaced by the next shiny thing.
The pattern is familiar to anyone who's watched their neighborhood businesses cycle through. The bakery on the corner closes, a café opens in its place, then a smoothie shop takes over before the espresso machine cools. Software is doing the same. The tools we depend on — the ones that run our payrolls, track our inventory, even manage our health — are increasingly built to be swapped out rather than inherited.
This matters because disposable software doesn't just mean more apps to juggle. It means the people who actually use it — the shop owners, the delivery workers, the small operators — get stuck holding the bag when a tool shuts down or pivots. The data they've accumulated? Gone or locked behind a new pricing wall. The workflows they've built? Unraveled. The cousin who figured out how to run their side business on a specific platform now has to learn the new one — or risk losing customers.
Why this matters for us: when the tools that run our daily lives keep changing without warning, it's usually la gente — not the founders or the VCs — who pays the price.
Meta's AI for Work Chief Walks
Nikhil Singh, Meta's head of AI for Work, is leaving. He's been with the company since 2016, and he's not the only one. The departure is part of Meta's broader shift to put AI front and center — from the product side to the business side.
Meta is reorganizing around AI. They're putting a new head of AI for Work in place, and they're consolidating AI teams across Facebook, Instagram, and WhatsApp into one group. The idea is that AI will be baked into every product, not bolted on as a feature.
Singh's departure is a signal that Meta is trying to figure out how to make money from AI in the enterprise world.
The new bottleneck isn't code — it's deciding what to build
Stack Overflow's latest piece points out something most folks in product miss: the bottleneck shifted years ago. It used to be whether we could build fast enough. Now it's whether we're building the right thing. The technical team is ready — the real friction is in the decisions.
This is the quiet truth nobody says out loud at product all-hands. We got so good at shipping that shipping became the easy part. The hard part is the 3 a.m. call: "Should we cut this feature, double down, or kill it entirely?" That's the bottleneck now. Not velocity. Clarity.
The piece lands because it names what the comunidad already feels: la gente who's been through the hustle — the abuelo who spent thirty years at the same plant, the cousin running a side business, the auntie on Facebook who knows every price change before anyone else — they all know this. It's never been about doing more. It's about knowing what to do.
Why this matters for us: the next generation of workers won't be measured by how fast they ship — they'll be measured by how clear they are about what's worth shipping.
Your DNA is tangled — and AI might miss it
AI models have gotten very good at reading DNA. They see it as a string of letters — A, C, G, T — and predict how genes behave from that sequence alone. But the genome isn't just a string. It's a physical object that folds, tangles, and loops inside the cell nucleus.
That physical shape matters. A gene can be switched on or off by where it sits in the fold, not just what letters it contains. Two people can share the same DNA letters but have different diseases because the genome folded differently in them. Current AI models mostly ignore this. They treat the genome like a spreadsheet row instead of a crumpled piece of paper.
Sahin.io launches a new homepage — a cleaner take on personal branding
Sahin.io has launched a fresh homepage. It's a personal site — the kind that lives at a single domain, usually built by a founder or developer who wants something that doesn't look like a template.
The page is straightforward: it's a landing spot for Sahin to put out content, share work, and connect with readers. No fancy dashboard, no subscription wall — just a clean layout with a clear purpose.
The personal site has been having a quiet comeback lately. A lot of us are tired of algorithmic feeds and the noise of LinkedIn and Twitter. A personal site is a place that belongs to you — no algorithm can bury it, no platform can shut it down.
Why this matters for us: when founders build their own spaces instead of renting them on social platforms, they keep control over how they show up to the world — and that's the same thing we want for our businesses.
Mobileye entra al robotaxi con servicio propio
Mobileye, el brazo de conducción autónoma de Intel, está entrando al mercado de robotaxis en Estados Unidos con un servicio propio — no como proveedor de sensores, sino como operador de flota.
La jugada es directa. Mobileye ya tiene la tecnología de detección y la experiencia en visión artificial; ahora se está yendo por su cuenta para correr sus propios taxis sin conductor. La compañía ya ha desplegado flotas en China y Japón, y ahora está llevando la misma apuesta a las calles norteamericanas.
Esto es distinto de los otros equipos de robotaxi que llevan años en el mercado.
TLDR Newsletter ships you what matters in tech
TLDR is back to its daily rhythm — Dan and the team are curating the links that actually matter for Brown folks in tech, not just the Silicon Valley noise. You get a quick roundup of what's moving in AI, startup funding, and the tools we use every day, with enough context to know whether it's worth your time.
The newsletter has been around long enough to build trust with a readership that doesn't want another tech bro newsletter telling us what to think. It's lean — no 2,000-word deep dives, no sponsored content buried in the middle — just the links that matter and why they matter. If you're a Brown dev, founder, or someone trying to figure out where the money is flowing in tech right now, it's worth keeping on your morning read.
Why this matters for us: When TLDR picks up the right stories — about funding going to underrepresented founders, new tools for immigrant entrepreneurs, or policy shifts that hit our communities — it quietly shapes what we pay attention to and what we ignore.
Running local models is actually good now
Vicky Boykis put together a long-form post about running AI models locally — and the TLDR crew picked it up as one of the week's best reads. The piece walks through what's changed: you no longer need a $5,000 GPU to get useful results, and the software stack has caught up.
What matters is the shift. Cloud models are great when you have a steady stream of traffic and someone to manage the bills. Local models are for the people who want to own their data, who don't want to pay per call, who want to run things on their own machines. The hardware is cheaper. The tools are easier.
The mirror test nobody tells product managers about
Dan Pereira's latest piece asks a quiet question that most PMs avoid: are you building things that actually land, or just polishing the same old mirrors?
The TLDR Product newsletter has always been one of the more honest reads in the tech space — not because it's loud about it, but because it stops to look at what's actually happening. Pereira's point is that product work can become a hall of mirrors: dashboards, roadmaps, stakeholder meetings, all reflecting back the same motion. The work feels real. But does it move the needle? Is the product actually getting better, or just getting more meetings about getting better?
There's something familiar here for anyone who's spent years in the grind — the cousin who runs a side business and has to juggle accounting, suppliers, and a dozen WhatsApp groups, all while pretending she knows what she's doing. The PM role has become that same kind of work: you're responsible for everything and accountable for something, and the pressure is to keep looking like you've got it together while the thing you're building slowly reveals whether it was worth it. The real test isn't the review deck. It's whether people actually use what you shipped.
Why this matters for us: the same mirror-chasing that traps product managers in Silicon Valley is what keeps Brown and Black founders working harder for less recognition, and it's on us to ask whether the work we're doing is moving things or just moving us in circles.
Para la comunidad
Tech affecting the Hispanic community
The stories below land different for our gente — immigration tech, language access, the unbanked, kids of color, gig-worker rights.
Los AI agents are getting tired
The Atlantic reports that AI agents — the ones that do the heavy lifting for you now — are starting to show signs of exhaustion. The work that used to be done by a human in an afternoon is now being offloaded to AI, and the agents are getting tired too.
It's not that they're breaking down. It's that they're getting tired. The work that used to be done by a human in an afternoon is now being offloaded to AI, and the agents are getting tired too.
What's happening is that AI agents are starting to show signs of exhaustion. The work that used to be done by a human in an afternoon is now being offloaded to AI, and the agents are getting tired too.
Why this matters for us: When the AI agents get tired, the work that was supposed to be free becomes expensive again, and la gente who depend on them to get things done end up paying for it.
The Mom-and-Pop SaaS Era Has Arrived
AI just broke the old rule: you used to need a team and a runway to build software. Now a solo founder with a laptop can ship a product that competes with the big guys. This isn't a buzzword — it's happening right now, and it's changing who gets to play.
The shift is practical. AI tools handle the heavy lifting — writing code, designing interfaces, even marketing — so the bottleneck is no longer technical skill but the hustle. A cousin running a side business can spin up a SaaS product. A small shop owner can build a tool that solves their own problem and sell it to others. The mom-and-pop shops of software are no longer just a metaphor.
Why this matters for us: when AI levels the playing field, it means our families and communities can stop waiting for permission to build — and start building.