Swirl Creative

We do digital product development and ecommerce for startups and scaleups.

[Cloudflare @Cloudflare]

We are launching Workers Cache, a regionally tiered cache that sits directly in front of your Worker entrypoints. Infinitely composable, configured via standard HTTP headers. https://t.co/eBbxHIgUBA

[🥞 Gavin @bintz_gavin]

https://t.co/TzqUOc6yR0

If it plays smoothly in the browser, Helios renders it smoothly to video

[🥞 Gavin @bintz_gavin]

what actually makes Agent One unique?

deployment

there are two main pieces to Agent One, Agents and Sites.

Sites can contain one or many agents. you choose either a subdomain on our site or link your own custom domain.

the most common use-case in existing customers is using Sites to create standalone support pages. they use the chat embed on their main site along with deploying it to chat.[theirdomain].com to have a direct support link to send out to customers as well

but you can deploy whatever random agent you want. i created cursor rule maker (cursorrules .agnt.one) in literally less than ten minutes by just uploading some good cursorrules examples and deploying it. it's been generating rules for over a hundred devs a day for months now

that is the real power of the platform, creating chatbots and agents that are discoverable on google and can plug your own products

[Andrej Karpathy @karpathy]

I don't have too too much to add on top of this earlier post on V3 and I think it applies to R1 too (which is the more recent, thinking equivalent).

I will say that Deep Learning has a legendary ravenous appetite for compute, like no other algorithm that has ever been developed in AI. You may not always be utilizing it fully but I would never bet against compute as the upper bound for achievable intelligence in the long run. Not just for an individual final training run, but also for the entire innovation / experimentation engine that silently underlies all the algorithmic innovations.

Data has historically been seen as a separate category from compute, but even data is downstream of compute to a large extent - you can spend compute to create data. Tons of it. You've heard this called synthetic data generation, but less obviously, there is a very deep connection (equivalence even) between "synthetic data generation" and "reinforcement learning". In the trial-and-error learning process in RL, the "trial" is model generating (synthetic) data, which it then learns from based on the "error" (/reward). Conversely, when you generate synthetic data and then rank or filter it in any way, your filter is straight up equivalent to a 0-1 advantage function - congrats you're doing crappy RL.

Last thought. Not sure if this is obvious. There are two major types of learning, in both children and in deep learning. There is 1) imitation learning (watch and repeat, i.e. pretraining, supervised finetuning), and 2) trial-and-error learning (reinforcement learning). My favorite simple example is AlphaGo - 1) is learning by imitating expert players, 2) is reinforcement learning to win the game. Almost every single shocking result of deep learning, and the source of all *magic* is always 2. 2 is significantly significantly more powerful. 2 is what surprises you. 2 is when the paddle learns to hit the ball behind the blocks in Breakout. 2 is when AlphaGo beats even Lee Sedol. And 2 is the "aha moment" when the DeepSeek (or o1 etc.) discovers that it works well to re-evaluate your assumptions, backtrack, try something else, etc. It's the solving strategies you see this model use in its chain of thought. It's how it goes back and forth thinking to itself. These thoughts are *emergent* (!!!) and this is actually seriously incredible, impressive and new (as in publicly available and documented etc.). The model could never learn this with 1 (by imitation), because the cognition of the model and the cognition of the human labeler is different. The human would never know to correctly annotate these kinds of solving strategies and what they should even look like. They have to be discovered during reinforcement learning as empirically and statistically useful towards a final outcome.

(Last last thought/reference this time for real is that RL is powerful but RLHF is not. RLHF is not RL. I have a separate rant on that in an earlier tweet
https://t.co/RMIpFPVpuM)

[🥞 Gavin @bintz_gavin]

I've learned it's much better to have tools that work for you, not the other way around!

My latest project, Agent One is born from my passion for solving exactly this—freeing real people from the constant pull of support tasks. Creating a system where routine issues are handled automatically, so you can focus on enjoying your journey without the weight of opening that laptop.

[Jen Simmons @jensimmons]

Today is the day. Web Push on iOS.

https://t.co/mZDtj112yN

[Swirl Creative @CreativeSwirl]

From finding the ideal price point to testing new features, A/B testing is a powerful tool that every business should consider using. Want to learn more?

Click the link to read the full blog post:
https://t.co/nkwnyQ6Zf3

#abtesting #conversion #website #seo #MarketingStrategy

[Swirl Creative @CreativeSwirl]

7/ The use of AI and machine learning will continue to increase, helping retailers optimize pricing, inventory management, marketing, and customer service.

[Swirl Creative @CreativeSwirl]

🧵: Ecommerce trends for 2023

#EcommerceTrends2023 #VirtualRealityShopping #PersonalizedShopping #ContactlessDelivery #SocialCommerce #SubscriptionBasedShopping #OmnichannelRetail #AIinRetail #CryptocurrencyPayments #FutureofShopping

[Swirl Creative @CreativeSwirl]

6/ Retailers will focus on building stronger, more seamless omnichannel experiences, allowing customers to shop seamlessly across online and offline channels.

[Gergely Orosz @GergelyOrosz]

An underrated way to attract great engineers to your company

(drumroll)

Have a great engineering blog.

This means ENGINEERS write the blog, not content writers. And they give the "real deal" about challenges, learnings, failures.

So few companies do this. A few that do:

[Swirl Creative @CreativeSwirl]

If anyone wants to reminisce some more about the olden days then check out our blog:

https://t.co/YnEEhN3Abf

[Swirl Creative @CreativeSwirl]

Once upon a time, in the dark ages of the internet, websites were hosted on dedicated servers running the LAMP stack (Linux, Apache, MySQL, and a bucket of PHP). It was a simpler time, when "500 Internal Server Error" was just a way of life.

🧵👇

[Swirl Creative @CreativeSwirl]

And so, we arrive at the present day. Websites and applications are hosted on platforms like Netlify and AWS Lambda, and the word "server" is all but forgotten. But just wait until tomorrow - who knows what crazy new backend architectures the future will bring!