Scale & Strategy
together with
This is Scale & Strategy, your friendly neighborhood ‘business information sponge’ (squeeze us out daily → get smarter).
Here’s what we got for you today:
- GPT-5.5 finally makes ChatGPT feel agentic
- Americans against AI data centers
GPT-5.5 finally makes ChatGPT feel agentic
OpenAI has released plenty of “major” model upgrades over the past year. Most felt incremental once the benchmark screenshots stopped circulating on X. GPT-5.5 lands differently. After a day of actual use, the jump from 5.3 and 5.4 is obvious almost immediately.
The biggest shift is how the model approaches work.
GPT-5.5 feels built to stay inside a workflow longer, maintain context across more steps, and keep pushing toward completion instead of constantly bouncing tasks back to the user. OpenAI clearly wants ChatGPT to evolve from an assistant into an operating layer for knowledge work, and 5.5 is the first version that seriously points in that direction.
The new workflow agents platform makes that ambition very clear. These agents can move between tools, pull data, execute tasks, and produce outputs with far less babysitting than previous generations. The underlying model is dramatically better at retaining complex instructions over long chains of actions, which has quietly been one of the biggest bottlenecks for usable AI agents.
For coding, the thinking modes matter more than most people realize.
Standard thinking is fast and surprisingly capable for lightweight tasks, but it still stumbles on larger builds and more layered debugging work. Extended thinking slows things down, but the improvement in reliability is substantial. Anyone using 5.5 for serious coding, systems work, or multi-step projects should probably leave extended mode on by default. Waiting an extra minute beats spending an hour untangling confident nonsense generated at hyperspeed. The modern productivity stack in one sentence.
In practice, GPT-5.5 is extremely good at cleaning up chaos.
Throw messy notes, fragmented data, or rough ideas at it and it can turn them into polished dashboards, structured plans, and usable outputs very quickly. The visual quality of generated HTML and interfaces has improved a lot too. Previous OpenAI models had a habit of producing layouts that looked like enterprise software from 2011 wearing a startup hoodie. 5.5 has a noticeably stronger sense of spacing, hierarchy, and presentation.
That makes it genuinely useful for operators, founders, analysts, and teams trying to prototype quickly without pulling in dedicated frontend or design resources.
The weak spot shows up in heavier multi-file workflows.
You can ask GPT-5.5 to generate an entire business package from a single prompt and it will absolutely try: landing pages, projections, slides, dashboards, workflows. The outputs are often impressive. But a lot of those deliverables still arrive wrapped in HTML artifacts instead of clean native files.
That matters more than people think.
Claude Cowork currently handles structured outputs better across PowerPoints, spreadsheets, PDFs, and larger file environments. Teams working deep inside operational workflows will notice the difference immediately.
The broader comparison is getting much tighter, though.
GPT-5.5 is strong enough to become a primary AI tool for a lot of people. The gains in instruction-following, workflow continuity, and agentic execution are very real. ChatGPT finally feels less like a chatbot with add-ons and more like software that can actively move work forward.
Claude still holds the edge for serious coding reliability and deeper knowledge work. OpenAI’s Codex improvements are narrowing that gap fast enough that this conversation could look very different within a few months.
Pricing is also worth watching. GPT-5.5 costs more per token than 5.4, though OpenAI claims efficiency gains offset the increase in practice. We’ll see. Frontier model pricing right now feels a little like cloud bills in 2017: everyone nods confidently until finance starts asking harder questions.
Still, GPT-5.5 is easily the strongest version of ChatGPT OpenAI has shipped. More importantly, it feels aligned with where the industry is heading next: models that execute, persist, and operate across workflows instead of just generating responses inside a chat window.
The perfect email is useless in a spam folder. This suite guarantees maximum deliverability
You’ve built the list and crafted the copy.
You’ve designed a stunning email in your ESP.
Then you hit send, and… crickets. Opens are low, clicks are embarrassing, and the only replies you get are bounce notifications.
Sending to a bad list is more than wasted effort. It’s a direct threat to your sender reputation that risks drowning future campaigns in spam and promotion folders.
And no readers = no ROI.
Stop gambling with your deliverability.
ZeroBounce is the all-in-one platform trusted by marketers at Amazon, HubSpot and Netflix for landing their messages in primary inboxes.
They provide 99.6% accurate email validation, inbox placement testing, blacklist monitoring, and more, so you can send with absolute certainty and get a bigger ROI from your emails.
Discover why over 500,000 businesses rely on ZeroBounce to ensure their message is heard.
Americans against AI data centers
Silicon Valley sees AI infrastructure as the foundation of the next economy. Most Americans see a giant warehouse that eats electricity, drains water, and raises their utility bill. Tiny messaging problem there.
A new Gallup poll found that roughly 70% of Americans oppose data center construction in their local communities, with nearly half strongly opposed. Only about a quarter support local development, and just 7% feel strongly in favor.
The backlash cuts across political lines too. Opposition came from 75% of Democrats, 74% of independents, and 63% of Republicans. That kind of bipartisan agreement barely exists anymore outside of hating airline seating arrangements.
The concerns are exactly what you’d expect when trillion-dollar AI ambitions collide with actual neighborhoods.
Environmental issues dominated responses. Half of respondents worried about strain on local resources, while others specifically pointed to water consumption, grid pressure, and pollution. Around 22% cited quality-of-life concerns like traffic, housing pressure, and falling property values. Another 20% worried about rising costs, including utilities and general cost of living.
And then there’s the broader AI skepticism itself. About 14% said their opposition came from simply having negative views of AI overall.
That last number probably understates the real sentiment. Outside tech circles, a lot of people still associate AI with layoffs, scams, deepfakes, cheating in schools, and increasingly deranged LinkedIn posts about “10x productivity.” Not exactly a beloved consumer brand.
Gallup warned that data center expansion could quickly become a political flashpoint as infrastructure buildouts accelerate. That feels inevitable. The AI industry needs massive amounts of compute, power, land, and water at the exact moment communities are becoming more hostile toward large-scale industrial projects.
At the same time, supporters of development do see upside. Among respondents in favor of data centers, about two-thirds expected economic benefits through jobs, tax revenue, and infrastructure investment.
That’s the core tension now surrounding AI infrastructure.
The labs need enormous physical expansion to justify their valuations and keep scaling models. The public increasingly views that expansion with suspicion. And frankly, the industry hasn’t done a particularly good job explaining why ordinary people should be excited about sacrificing local resources so a frontier model can generate slightly better slide decks and anime avatars.
The adoption problem underneath all this is becoming harder to ignore.
Microsoft research recently found only a small minority of people use AI regularly. Meanwhile the public conversation around AI keeps getting dominated by job displacement fears, environmental concerns, and stories about people psychologically spiraling after spending too much time talking to chatbots. None of that exactly screams “please build a hyperscale compute cluster next to my town.”
The AI industry has ended up in a strange position: it’s trying to build civilization-scale infrastructure before most people are convinced they even want the product. Historically, that tends to produce resistance. Fast.
Was this email forwarded to you?
That’s it for today and as always It would mean the world to us if you help us grow and share this newsletter with other operators.
Our mission is to help as many business operators as possible, and we would love for you to help us with that mission!
