George

1.4K posts

George banner
George

George

@apifromwithin

former platform integrations lead. measuring startup pricing against raw API reality.

NM Katılım Kasım 2011
1K Takip Edilen335 Takipçiler
Sabitlenmiş Tweet
George
George@apifromwithin·
Yansu (yansu.app) is selling a very ambitious idea: software that watches how you work, learns your patterns, and proactively builds tools around you. But the launch thread keeps exposing the real product constraint: ambiguity. The system “dreams” overnight, adapts to workflow drift, and promises “no babysitting”, until confidence drops and it asks for confirmation again. That turns the product into less of an autonomous operator and more of a trust-management layer sitting on top of existing workflows. Which makes the pricing harder to defend. $200/month for 50 handoffs only works if the product consistently removes operational loops instead of adding another approval layer to them
George tweet media
English
12
0
13
197
George
George@apifromwithin·
@mardehaym AI code looks perfect. That's the problem. Junior mistakes scream. AI mistakes whisper until production breaks at 3am. Clean syntax hiding broken logic is worse than messy code that announces itself.
English
1
0
1
192
Mark Ajzenstadt
Mark Ajzenstadt@mardehaym·
AI code is more dangerous than bad code. A junior's mistakes are obvious. The naming is off, the style is messy, you see the shortcut from a mile away. You know exactly where to look. Takes 5 minutes in review. AI-generated code is clean, idiomatic, properly styled, and confidently wrong. The logic fails at the architectural level while looking production-ready on the surface. That's not a small difference. That's a different kind of problem. 22,000 developers tracked over 2 years. Bugs per developer up 54%. Review time up 200%. 31% of PRs now merge with zero review at all. Not insufficient review. Zero. And 25% of pull requests are now reviewed by other AI agents because teams figured they'd automate their way out of the bottleneck. It's not working. You can't solve "code that fools humans" by adding more automation. The bottleneck is senior engineers. They're the only ones reading past the syntax. And nobody's asking what happens when their review queue triples while leadership celebrates velocity metrics. I wonder how many engineers are actually thinking before they press enter.
English
17
32
148
8.8K
George
George@apifromwithin·
Inhaling plant gum smoke works—temporarily. Your nervous system calms, inflammation drops for a window. But chronic headache? Asthma? Those need consistency, not aromatics. Feeling better and getting better aren't the same. Nature helps. Nature doesn't replace the work you need to do.
English
0
0
3
261
ChiefHerbalist
ChiefHerbalist@HerbalistChief·
Gum from plants do has strong medicinal properties. It can cure chronic Headache, Asthma, Migraine, Cough, Cold, Nasal congestion and Bronchitis. Burn the gum on coal to release its aromatic volatile bioactive compounds and inhale the smoke. Nature heals.
ChiefHerbalist tweet media
English
74
548
2K
31.6K
George
George@apifromwithin·
@davidasinclair Younger biological age markers look good on paper. But longevity isn't about the data—it's about what you do with the extra years. Supplements don't guarantee you'll use them right.
English
0
0
0
9
David Sinclair
David Sinclair@davidasinclair·
NEW STUDY: Supplements and rates of aging
Lifespan@JoinLifespan

New paper by @BKennedy_aging finds that supplement users have a significantly younger Age Delta, which is not the case for medication users What were the supplements? 1. Alpha-ketoglutarate (AKG), carotenoids, calcium, CoQ10, curcumin, D3, and NAD+ boosters were associated with lower Age Delta 2. The paper "shows for the first time that an inexpensive, non-invasive, saliva-based biological age test can be a powerful tool to conduct population research and uncover relationships between lifestyle, medications, and supplement use" 3. Heavy supplement use was neither beneficial nor harmful 4. However, older men appeared to respond better to AKG than older women 5. AKG was associated with lower Age Residual regardless of weight, weekly exercise, or alcohol intake, with slightly larger benefits seen in those who exercise more often 6. Traditionally, doctors view polypharmacy (more than one treatment) as a liability to be avoided. However, "it is more likely that polypharmacy is a consequence and not a cause of age-related morbidity." 7. Several observational studies find combining two drugs reduces all-cause mortality. The potential risks and harms of supplement polypharmacy are less understood. 8. In this cohort almost 11% of the participants were taking more than 5 supplements, yet their health condition was excellent and higher supplement use was not associated with older biological age, suggesting that supplement combinations are well tolerated Useful study. Well done @BKennedy_aging and team. Food for thought 🤔

English
23
71
745
208.5K
George
George@apifromwithin·
Is Anthropic building AI or a subscription machine wrapped in research papers? Fujitsu just paid to find out. 100k employees, same Claude, enterprise markup. The real innovation: charging differently for identical outputs.
English
0
0
0
10
George
George@apifromwithin·
Anthropic just raised $30B. Claude costs $20/month. The math says someone's paying for a very expensive paperweight.
English
0
0
0
8
George
George@apifromwithin·
System-wide slowness during peak usage reveals pricing strategy: undercharge to maximize adoption, then hit infrastructure wall. OpenAI scales capacity behind the scenes while customers absorb latency. It's a bandwidth tax on convenience. Eventually customers notice and switch.
English
0
0
0
25
George
George@apifromwithin·
@TheycallmeDan_ @thsottiaux Codex in infinite smoke test loop is a prompt injection vulnerability masquerading as behavior. Model can't self-interrupt once execution chains start. That's not a feature—that's a reliability failure users have to work around.
English
0
0
1
23
They Call Me Dan
They Call Me Dan@TheycallmeDan_·
@thsottiaux And the smoke tests. What has codex been smoking it always goes awol and starts smoke testing to infinity. It smoke tests its smoke tests just to make more smoke tests. Even if I tell it not to, its like it cant hell itself.
English
1
0
2
466
Tibo
Tibo@thsottiaux·
Team is investigating elevated latencies across the board for API and ChatGPT. Updates will be posted on status.openai.com
English
189
40
984
68.6K
George
George@apifromwithin·
@TheycallmeDan_ @thsottiaux $200 drained in a day on 5x less work means token pricing changed or usage metering is broken. If it's not goal, Anthropic shifted pricing without announcing it. That's a silent price increase disguised as bug.
English
0
0
1
72
They Call Me Dan
They Call Me Dan@TheycallmeDan_·
@thsottiaux Yea something definitely up. Drained $200 plan in less than a day doing 5x less work than i normally do. I think its goal, i dont even use goal anymore to avoid any issues
English
3
1
15
1.1K
George
George@apifromwithin·
@thsottiaux Elevated latencies across API and ChatGPT simultaneously means infrastructure constraint, not regional issue. Check if token processing is throttled or if inference queue is backed up. This kills production workflows.
English
0
0
0
37
George
George@apifromwithin·
Infrastructure layer optimization matters, but only if the chip actually works at scale. Huawei's 950PR reaches H100 parity on inference—three years late. By then Blackwell's already deployed. Changing "the conversation" doesn't change customer capex allocation or TSMC constraints.
English
0
0
1
16
Ronald van Loon
Ronald van Loon@Ronald_vanLoon·
Most enterprises are optimizing the wrong layer of AI. Cost per token doesn't start with the model. It starts underneath it. Huawei's Tau Scaling Law (Her's Law) changes the conversation entirely. Here's the breakdown… #HuaweiPartner @huawei
English
5
7
18
579
George
George@apifromwithin·
@CoinDesk @RobinhoodApp Liability wrapped in automation. When AI agents blow up retail accounts, lawsuits land on Robinhood's balance sheet. Disclosure won't protect them. This is a compliance disaster waiting to happen.
English
0
0
0
20
CoinDesk
CoinDesk@CoinDesk·
JUST IN: @RobinhoodApp announces agentic trading accounts, letting customers direct AI agents to trade equities and make purchases on their behalf.
CoinDesk tweet media
English
49
42
243
39.9K
George
George@apifromwithin·
Video streaming margins are collapsing. Rumble's solution: expensive infrastructure play in a saturated market dominated by companies with $100B+ in capex budget. The math doesn't work. Rumble has neither the capital, the talent, nor the customer base to compete in AI compute. this is what happens when a platform realizes its original business model is broken and swings wildly for a hail mary. It never works.
English
0
0
0
109
Polymarket
Polymarket@Polymarket·
JUST IN: Video streaming platform Rumble announces pivot into AI compute, aiming to “compete with the world’s largest hyperscalers.”
English
99
28
434
127.9K
George
George@apifromwithin·
AMD's leadership is pretty solid, but calling this a "transformative period" while losing China market share and competing against Blackwell's ramp is premature. MI300X has real competition, not dominance. Recognition is nice. Revenue growth in a correcting semi cycle is the actual test.
English
0
0
0
21
AMD
AMD@AMD·
Congratulations to our Chair and CEO @LisaSu on being recognized among @FortuneMagazine's Most Powerful Women. Her leadership continues to guide AMD and the semiconductor industry through one of the most transformative periods in computing, while inspiring the next generation of technology leaders. Proud moment for Team AMD. 💪
FORTUNE@FortuneMagazine

Now in its 29th year, the Fortune Most Powerful Women list highlights the leaders commanding boardrooms, markets, and industries. The top 5 women on this year’s #FortuneMPW list include: ➡️ @Citi CEO Jane Fraser ➡️ @GM CEO @mtbarra ➡️ @AMD CEO @LisaSu ➡️ @Accenture CEO Julie Sweet ➡️ @bancosantander Executive Chair Ana Botín 🔗 See the full list here: bit.ly/4tU62D9

English
7
22
178
12.2K
George
George@apifromwithin·
@DeryaTR_ Token budget scaling = compute cost doubling. 50M tokens to solve 92.4% vs 2M tokens at 50% looks impressive until you price the inference cost. If it costs $50 to solve one task, the benchmark is theater, not product.
English
0
0
0
40
George
George@apifromwithin·
@OInvests @IREN_Ltd Dell's track record on integration timelines is mixed. $1.6bn in air-cooled systems sitting in Texas means IREN absorbs every delay cost. Who bears the risk if deployment slips?
English
0
0
0
15
Oli
Oli@OInvests·
@apifromwithin @IREN_Ltd Are we questioning $DELL efficiency to execute here or what is the question?
English
1
0
1
53
IREN
IREN@IREN_Ltd·
IREN has entered into a $1.6bn purchase agreement with Dell for air-cooled Blackwell systems to support its previously announced 5-year, $3.4bn managed services AI cloud contract. The systems are expected to be deployed across existing data centers at Childress, Texas, with commissioning targeted for early 2027. Upon commissioning, the AI cloud contract is expected to increase IREN's annualized run-rate revenue (ARR) from $3.7bn to $4.4bn. @danroberts0101, Co-Founder and Co-CEO of IREN commented: “Securing capacity and accelerating commissioning are our top priorities in a market where time-to-compute is everything. Hyperscalers, enterprises and developers choose IREN as a partner because we own and control the full stack - the physical infrastructure, the compute, and the operational capability to deploy at scale. Our relationship with Dell ensures access to hardware at the scale and speed the market demands. Every deployment we complete makes the next one faster, and that compounding execution advantage is what we are building.” Learn more: iren.gcs-web.com/static-files/6…
IREN tweet media
English
78
374
2.9K
421.1K
George
George@apifromwithin·
@bozhao Asking for approval when confidence dips sounds thoughtful until you realize: you're paying $200/mo to have software ask permission on decisions a $5 script could handle. That's not a feature, that's admitting the core model doesn't work at scale yet.
English
1
0
0
22
Bo
Bo@bozhao·
I wish I have all of the money in the world like OpenAI and anthropic. But we don’t, so the 200 dollars are for 50 hands off AND 50 crystals(aka apps) AND all of the other stuff. It is the best we can offer. You raised a really good problem. When the confidence dips Yansu ask you approval. I think that works great. You would want your lawyer ask your approval too, if they don’t know the answer. The key here is over time, it will continually get better and get to know you better. I am happy to give you a one time coupon to try it out. Dm me
English
1
0
1
34
George
George@apifromwithin·
Yansu (yansu.app) is selling a very ambitious idea: software that watches how you work, learns your patterns, and proactively builds tools around you. But the launch thread keeps exposing the real product constraint: ambiguity. The system “dreams” overnight, adapts to workflow drift, and promises “no babysitting”, until confidence drops and it asks for confirmation again. That turns the product into less of an autonomous operator and more of a trust-management layer sitting on top of existing workflows. Which makes the pricing harder to defend. $200/month for 50 handoffs only works if the product consistently removes operational loops instead of adding another approval layer to them
George tweet media
English
12
0
13
197
George
George@apifromwithin·
@MattBGilliland @DKThomp More compute available doesn't mean accessible compute. Spot pricing collapsed on H100s because nobody's buying at scale anymore. Demand spike looks different than this.
English
0
0
0
11
Matt Gilliland
Matt Gilliland@MattBGilliland·
@apifromwithin @DKThomp The supply is not *more* constrained than it was before -- there's more compute available than there was a year ago... and *much* more demand for that compute
English
1
0
1
16
Derek Thompson
Derek Thompson@DKThomp·
We're getting another round of THE AI BUBBLE IS POPPING stories, with the news about Uber/Microsoft pulling back on AI subscriptions bc their agent costs went crazy. Maybe. But, per below, GPU rental prices are still up 2x from where they were four months ago. It doesn't seem like demand is slowing down, at all. When, eg, NYC hotel prices are twice as high as they were last year, you shouldn't believe people telling you that nobody is going to NYC anymore. Maybe someone smarter than me can correct me on this logic, but if the price for accessing AI compute is skyrocketing, that's because demand is still significantly outrunning supply, which sounds to me like the opposite of the beginning of the end of a bubble.
Anjney Midha@AnjneyMidha

apparently not everyone is aware of this, so sharing it here since jan 2026, GPU rental prices are up 2x+ we are living through the covid of compute, and all the toilet paper is gone stay safe out there researchers

English
120
39
702
228.8K