Sign up at dexto.ai to try it out and get $5 in free credits
Dexto retweeted
we just built this app in dexto - tinder for your x bookmarks. what do you guys think?
Dexto retweeted
i've been looking for something like this for so long!
repurposed and deployed this into my own version of mage arena, with prep actions, hold to charge and more things coming
hoping to make it more avatar (the last airbender) inspired soon
I've built a new opensource Elemental Sandbox, VFX playground built with @threejs . it contains 6 fully custamizable abilities.
Everything you see is generated: no textures, no sprite sheets, no baked meshes.
hand-written GLSL, GPU-simulated instanced particles, World position resolved in the vertex shaders...
Source code :
github.com/achrefelouafi/Lin…
Dexto retweeted
I think one reason the usage of AI agents hasn’t fully broken through yet is that the platforms built around them still present a fairly narrow view of what they can be used for and more importantly don’t do enough to inspire curiosity among the general public about what they can do with it.
Most tools showcase their capabilities through packaged workflows like write this, research that, generate an image or complete this task.
Those might be useful, but they make agents feel like a collection of separate tools for separate jobs. You get an output and move on. The platforms and their demos dont do enough to help the average person see how these capabilities could be combined, adapted, or turned into something entirely different.
The platforms environment plays a massive role in this imo. If the environment through which agents are called and their work is consumed is mostly a chat box, a wall of text, or a one-off artifact, then even a highly capable system can feel narrow. You experience the output, but not the larger creative and operational space behind it.
The more powerful experience might be when a platform lets you move fluidly between different modes:
→ an idea into a visual
→ a visual into a prototype
→ a prototype into a deployed experience
→ a repeated action into a skill
→ a skill into an automation
→ an external tool into a larger workflow
That changes your relationship with AI. You move from thinking “what task can this tool complete?” but rather start feeling, “what can I make this environment do?”
I feel like @dexto_ai is one of the few products moving in this direction well.
The biggest unlock for me wasn’t discovering that the models had suddenly become more capable. It was using Dexto that made more of that capability available, connected, and easy to experiment with.
AI adoption won’t fully happen when models become capable enough. It will happen when platforms become good enough at making that capability feel visible, personal, and endlessly explorable.
Hot take… isn’t it kinda crazy that nobody is really using AI Agents? I don’t mean software engineers or AI early adopters. I mean “college friends talking about it in group chat,” the feeling you got when everyone started using Instagram or TikTok.
These frontier AI models are *insane* (as are the harnesses & tool calls & the like). And every large tech co has an AI agents platform, not to mention all the YC startups doing vertical agents. Yet all of your friends and family outside of tech — who spend all day staring at their iPhones and get paid to work in browser tabs — don’t really care or find themselves using any AI agents yet.
Yes ChatGPT, Claude, etc. are extremely popular… but if you look at the engagement data the vast majority of people are still using these aI chat tools like a glorified Google + Grammarly. That’s why the AGI labs are all pushing desktop apps for Codex, Cowork, etc. so hard to non-technical ppl. And yes exceptions for lawyers and customer service but even those have some asterisks and exceptions to rule.
Look I’m not saying the ChatGPT moment for AI Agents is not coming… it most definitely is! Remember we pivoted from Arc to Dia precisely because we believe computing is going to be radically reimagined around these AI primitives. No doubt. But that’s my point: it’s just so surprising it hasn’t happened yet because all of the tech you’d need is there.
Again if you stop for a second and think about it… for all the press and money and hype and models and crazy ARR numbers… this “AI Agent” moment does not *feel* like the other breakthrough tech moments we’ve lived through (e.g. think the shift to Stories via Snapchat & Instagram, or shift to on-demand via Uber/Airbnb/Doordash).
Which is a long way of saying: if you can figure out the answer to “why” most people don’t care about AI agents yet (and have no enduring interest in using them) — especially since the models and harnesses are here and ready — the answer to that question will allow you to capture a lot of marketshare and make a lot of money in 2027.
Theoretically, the tech is ready for AI Agents to totally transform how we work and live our lives… but alas the general public dgaf… that’s the generational puzzle to solve for the next 12 months for anyone not working on the models themselves.
Dexto retweeted
one of dexto's most beloved apps is the "Skills Library"
you can start with one of the available skills, remix 2 or more skills or even create your own and publish them for others in the community to use!
whether you're a developer, designer, content creator, artist or simply tinkering - dexto comes with a full catalogue of ready-to-use-skills showing you how to get the most out of all the tools, apps and ai capabilities that come with it
it's the best way to combine your personal context & creativity while ensuring your agents produces high-quality outputs
Dexto retweeted
interactive html is mind blowing
we recently added a new app into @dexto_ai called artifacts, which uses html to create rich interactive components for plans, reports, mockups and just whenever you need a good way to visualize things
i was working on setting up a loop to clean up stale documentation in our codebase every day, and dexto produced this really nice report in the artifacts app, without even me asking it to!
the artifacts app groups all your generated artifacts across chats in one place, and you can share them with other dexto users + add comments on the artifacts as well
Dexto retweeted
If you want to give your agents access to X data, @dexto_ai makes it surprisingly easy.
Two clicks and you’re connected to the X MCP.
No complicated setup that X typically requires you to do and no need to buy X API credits.
Just connect X and start asking Dexto to use it.
Dexto retweeted
one of the things i often see people get wrong about context in agents is trying to "connect more sources"
you can plug in Slack, Notion, GitHub, Drive, your database… and still make the agent worse by throwing everything into its context window
the real job is figuring out what it should know now, what it should be able to pull in when needed, and what can stay as a reference in the background - and design a system around this so that your context can evolve over time
in @dexto_ai context engine, we had to take extra care in designing this as an independent context layer that takes care of pulling the right data/sources while interacting with the agent runtime
this way, your agent is always knowledgeable as well as capable of expanding and querying anything else it needs
🤖 Made with AI
Dexto retweeted
i bookmark a ton of stuff on X
new tools, product ideas, technical insights, design inspiration.
yet i almost never revisit them
so i built an X knowledge base in @dexto_ai that turns my bookmarks into visual idea graphs + compounding context my agents can actually use every day
what app do you guys use to create context for ai? e.g. where do you share your links to articles you want to remember, thoughtful/fun images or memes, things you want to read, things you want to watch (youtube vids, tiktok’s, reels), things you want to listen to, interesting tweets, spotify tracks, even clever musings you write to yourself, etc.
this is all so fragmented right now. ideally you’d love a place where all of this magically lives & is highly queryable / organizable, etc. i.e. when you use claude/gpt, it should able to reference this database easily for context too.
there needs to be a first class citizen here but i think it’s sorta missing.
Dexto retweeted
so i tried self hosting observability myself on AWS
turns out codex + terraform + aws + cloudflare is insane
was able to set up clickhouse running on an m6g.large instance, connect it to EBS volume (for local disk) and s3 (cold storage), set up all the ALB/VPC etc. network stuff, and wire it up to my domain on cloudflare with very little effort - about 3 high level steps and about 30 mins of active effort:
1. step 1: asked codex to tell me all the api key/manual configuration i need to do. here i had to set up aws iam, and do some manual work on cloudflare to create apps/api keys/separated for dev/prod, and connect the domain name i wanted to the clickhouse endpoint i would use
after step 1, my agent was able to access my aws account with ssm after i signed in.
2. step 2: prompted a /goal to setup the entire infrasturcture stack using terraform + scripts, then deploy clickhouse + OTEL collector on the ec2 instance.
codex then worked for 5 hours, wrote the entire code to set up the infra in terraform, deployed it with terraform CLI + some custom scripts, verified/tested that clickhouse was deployed and working fine with the local ssm connectivity
after that, i asked codex to write some scripts for me to monitor my ec2 instance health, test scripts for verifying that logs/metrics/traces/spans etc. work properly and a few other scripts for making it easy to monitor/maintain the instance
then i wired it into @dexto_ai, which was pretty easy, i just had to swap out the collector config to use my new clickhouse endpoint.
it's working pretty insanely, and now i no longer need to deal with arbitrary caps on traces anymore. i can just scale my ec2 instance up when i need it, and use my startup AWS credits ($67 a month).
mostly i cared about traces/spans anyway, and less about fancy monitoring features, so this is an incredible unlock compared to paying over $100 + usage based pricing on sentry/datadog etc.
Dexto retweeted
just added message navigation using these hooks into @dexto_ai, feels so clean
Dexto retweeted
1-shotted this with the new sonnet 5 and a custom design harness i've been working in dexto
i have 0 design experience but it feels like i can bring the infinite concepts sitting in my head to life
Dexto retweeted
Anthropic shouldn’t do this and doesn’t have to
Would be great if it stopped
Dexto retweeted
Introducing Claude Opus 4.7, our most capable Opus model yet.
It handles long-running tasks with more rigor, follows instructions more precisely, and verifies its own outputs before reporting back.
You can hand off your hardest work with less supervision.
ALT Claude Opus 4.7 Benchmarks
