Anthropic's Claude Can Now Create Interactive Visuals Directly in Conversations

[Séamus Boyle]

That's a pretty poor way to determine the strength of AI models. The data models are trained on has a fixed cutoff date... they aren't continuously updated. It's like complaining that a book can't tell you what the current date is.
Every time you use it it's like you are asking questions to a person who fell asleep in 2025 and just woke up. But it can use the internet just like you can if it needs to look up more up to date information. But it forgets that information the next time you chat.

I just asked google Gemini what year it was and it gave me the date and said it was Friday the 13th and brought up the superstitions of the date.

 

[tomstone74]

I'm on a Claude Pro plan and things are moving very fast and it's continously 😱 how fast they progress with new stuff. Working with Claude (Opus 4.6 for fun - Sonnet might have been good enough as well, who knows) yesterday on some scenarios / calculations on my ETF portfolio. It not simply calculated the various scenario statically and then e.g. present stuff in a static output table or so, but made something interactively (without explicitely telling) available in the chat window via visual sliders to adjust expected annual yield, TER etc ... including an automatic adjusting line chart to reflect slider value changes. Incredibly useful.

 

[Jensend]

I just asked google Gemini what year it was and it gave me the date and said it was Friday the 13th and brought up the superstitions of the date.

I just asked Claude using the default model Sonnet 4.6, and it replied

It's 2026!

I asked their most simple model, Haiku 4.5, "what year is it?"

It's 2026. Specifically, today is Thursday, March 12, 2026.

(which is correct for my time zone)

But I don't know the complete context of the conversation of the poster I replied to.
But generally, asking about recent events will be less reliable:

Claude Sonnet 4.6 has a knowledge cutoff date of May 2025. This means the models’ knowledge base is most extensive and reliable on information and events up to May 2025.

 

[Jensend]

I'd argue that its a perfectly reasonable way to determine if any given AI is having trouble with a simple task from a simple man with a simple question, Claude struggled and was gaslighting me into an argument of what year I quoted vs what the current year actually is. I'm sure it's a great app and lots of people love it, I had a poor experience at the very first onset and another app was just better without having to waste more of my energy I simply went to the one which worked. It's reasonable I think. 🙂

And I just asked two different Claude models about the current year and both gave the correct answer.
But I don't care what you use.
If I want to know the date and I don't already know it, I look at the system clock/calendar on my computer/phone.

 

[Dredd67]

but can it do my laundry or take out the trash?

That's all what we'll be left with as a job... we poor humans will be doing the odd jobs that are too menial even for their lowest specs automatons.

 

[svish]

Will be useful and good to see that it is available for all users. Will try it out soon.

 

[Trioptre]

Wow, this approach mirrors how some people understand - visually and step by step. Maybe we are not doomed to become dumber.

I’m not doomed to become dumber. Most people will, for certain though!

 

[Trioptre]

Careful.This website is full of trackers and collects tons of your data. Way more than on FB. Get an ad blocker

More than FB? Are you sure? How do you know? Can you tell us?

 

[Trioptre]

I just asked google Gemini what year it was and it gave me the date and said it was Friday the 13th and brought up the superstitions of the date.

Sounds anecdotal. I’d need to see your prompt before I believe you. Can you provide your prompt?

 

[wanha]

Like I said before, this is sustitution. Is not a tool. Yes, it’s a tool today, but it’s clear where this is going.

say more

 

[EllZ89]

through work, ive long had claude integration through VS code with a sonnet backend via bedrock, which has always been able to generate ASCII diagrams, but this is a welcomed step up.

 

[yannxou]

Nice feature! But music from the video sounds so AI it hurts!

 

[yannxou]

Also, this reminds me of what Wolfram started a few years ago...

 

[Pruus]

Just try to make an electronic circuit with Claude. Very rudiment and a big fail. There has a long way to go…

 

[twistedpixel8]

These are input-output systems. When there is no input from you they are no thinking, they are not living, they are just nothing.

But what if the input was automated to mimic a person’s basic “existence”, constantly feeding the bot stimulus, as if it’s a person confined to a room with a computer.

 

[Oblivious.Robot]

Sigh, some people really need to chill out with their "Ai companions" and stop arguing with other humans on why they don't feel the same way about it. But then when we have https://www.reddit.com/r/MyBoyfriendIsAI/ so it's understandable, I guess.

 

[Markle]

That structural diagram would be correct if the pediment rested only on its corner points, but it doesn't. I am an architect not a structural engineer, but i doubt this is correct

 

[Orange Bat]

One of the things that’s become very clear to me is there are too many AI tools all claiming that they’re making leaps and bounds above their competitors. Just when you begin to use one AI, a new one comes along, or one makes claims that their new model is more advanced. As I tell everybody, be careful about the claims that are being made about AI. I generally find that the people talking about AI changing the world are the ones invested in you paying for the AI.

 

[humblecoder]

A bit confused on what's new about this. Clearly I'm missing something obvious? You could already ask it to create "animated [x]" or "interactive [x]" using "[insert language tool]". For example "Generate an interactive mermaid chart in HTML showing the water flow for a generic township."

Obviously a completely contrived example, but that sort of thing was already there. I guess showing it immediately on-screen without a "code middleman" is nice...? 🤷‍♂️

 

[Wx_Man]

Please remind Apple….why am I paying a subscription to Apple Creator Studio ?

In fact, without your own credible and working AI why should I or anyone pay you a subscription for anything ?

 

[Wx_Man]

The very fact that Apple didn’t acquire Anthropic tells us everything about how late the company is to the AI game. And I’m not convinced Apple would have led Anthropic to its current market‑leading position anyway, given how badly they’ve handled Siri since acquiring it back in 2010.

So in the end, consumers can probably be glad things turned out the way they did.

Apple refused to use Claude as the basis for Apple Intelligence after their manifest failure to produce their own in-house because Anthropic wanted too much money for that privilege. They were going to go with Claude because that is the AI Apple uses internally for business and development needs.

Apple then chose Google’s Gemini because they already get billions per year from Google for making Google Search the default search engine in Safari. The $1 billion. USD Apple pays to Google each year for Gemini just gets deducted from the amount Google pays Apple for search engine preference.

But make no mistake…Anthropic’s Claude AI was Apple’s first choice, not Gemini. And Apple continues to use Claude internally.

 

[CapitalIdea]

but can it do my laundry or take out the trash?

No, but you’ll have plenty of time to do that yourself if you’re laid off because of it.

 

[CapitalIdea]

Please remind Apple….why am I paying a subscription to Apple Creator Studio ?

In fact, without your own credible and working AI why should I or anyone pay you a subscription for anything ?

Listen to yourself. You act like nobody ever had subscribed to anything before AI LLMs came out a few years ago.

 

[Wx_Man]

Listen to yourself. You act like nobody ever had subscribed to anything before AI LLMs came out a few years ago.

Sigh…

Did I say anything in the text about believing that no one subscribed to anything before AI LLM’s? No.

What I WAS saying is that Apple has no reason to hobble non-paid and paid Apple apps and programs just to entice people to subscribe to suites such as Apple Creator Studio, particularly now that they bundled in the obvious inferior Pages and Numbers as compared to Word and Excel or even Google’s offerings.

That reason to force people to subscribe by making it seem that the non-paid and even paid apps are inferior unless they subscribe is being made more moot by the year by AI. Pretty soon the only technical or financial reason to subscribe to any technology platform is the company that makes your favorite and most capable AI.

Apple makes enough pure cash and has more cash on hand than just about any other company. For..a…reason. And they did that BEFORE becoming rent seekers in every thing they do. I doubt Apple, even by using another company’s LLM (Google) will continue to make a compelling case for their rentier status as the years go on at the pace Anthropic and OpenAI are going. Not to mention the Chinese.

 

[harriska2]

I’m sitting here going through our website page by page ensuring that it is WCAG 2.1AA compliant by 4/24/26. With Claude. I pay for it out of pocket but it was worth every penny ($200/yr). I can only dream of what it will do for curriculum. For 20 years I’ve been looking at either rewriting thousands of pages of Direct Instruction curriculum or purchasing rights to old versions from the 1990s for about $150M in order to making it freely available. Kids need to learn to read/write/do basic math/spell/have basic logic. And the DI stuff teaches ages 3 through middle school. I think Claude and other AI tools can totally make this possible.