What if AI could run directly on a blockchain?
Hi Indie Hackers,
I’m Umang, CTO at fxis.ai and part of the core team behind Haveto.
For the last few years, we’ve been working on something simple to say but complex to solve:
Can AI actually run directly on a blockchain, at scale?
Most blockchains struggle when usage grows; fees rise, performance drops, and AI workloads become almost impossible. At Haveto, we’ve taken a different route: auto-sharding + on-chain AI execution. That means no external servers, no inflated costs, and a system designed to stay fast and affordable as demand increases.
Our mainnet and testnet are already live, and now we’re opening up conversations with developers, builders, and anyone curious about the future of AI + blockchain.
Instead of a pitch, I’d love to hear your thoughts:
Do you see real-world use cases for AI directly on-chain? Or do you think off-chain scaling solutions will always dominate?
Excited to learn from this community and share our journey along the way.
Really like the premise behind Haveto, a to-do that narrows focus to a few must-dos and then gets out of the way. That “start here” clarity is what most generic task tools miss. Curious on two fronts: what single activation best predicts someone sticks around (first “have to” completed, a planned week, or a streak hit)? And which channel is actually bringing paying users right now; search, community demos, or word-of-mouth? A tiny 30-sec “from chaos → done” demo and easy imports from popular lists could lower switching friction even more.
P.S. I’m with Buzz, we build conversion-focused Webflow sites and pragmatic SEO for product launches. Happy to share a quick GTM checklist if useful.
Thanks so much for taking the time to share your thoughts,. Really appreciate the depth of your feedback.
Just to clarify, Haveto isn’t a productivity or task management app. It’s actually a Layer-1 blockchain designed to run AI directly on-chain. The “have to” name might have made it sound like a to-do app, but our focus is on solving the scalability + cost problems of AI on blockchain.
That said, your perspective on activation clarity and lowering friction is spot on, even though we’re in a very different space; the principle is the same: make it simple for people to see value quickly. That’s something we’re also keeping front and center as we bring developers and early adopters onto Haveto.
Really glad you shared this. Insights from adjacent domains always spark new ways of thinking.
Cool product, I checked out Haveto and it strikes me as one of those tools that sits just where people need structure without losing flexibility. When you’re juggling ideas, UI, backend, PRs across multiple projects (as I am, with ScrumBuddy and Claude-powered orchestration), what usually kills velocity is the friction of tracking state and context. Haveto’s lean approach to “what you have to do” feels promising.
What I’d love to see is how Haveto handles context switching. Say I’m in backend task mode, then shift to UI, then revert, then get a bug fix reminder. If Haveto can preserve the thread of what I meant in each of those shifts, honoring codebase conventions, naming, priority order, then it won’t just be a TODO list, it’ll be an extension of my internal debugger.
Also curious: have you built any features yet around generating scaffold tasks automatically from user stories? In my orchestration setup with Claude, I feed in backlog items, and Claude helps me expand them into UI + API + PR subtasks, which saves hours in switching mental gears. If Haveto can lean there, I think it could hit a sweet spot between planning and shipping.
Thanks so much for taking the time to check out Haveto and share your thoughts. Just to clarify that Haveto isn’t a task management tool, it’s a Layer-1 blockchain that runs AI directly on-chain with auto-sharding for scalability and cost efficiency.
Your points around context-switching and orchestration are really interesting, though they highlight exactly the kind of developer use cases that on-chain AI could unlock in the future. Appreciate you bringing that perspective, it’s always valuable to see how people connect the dots across different domains.
Got it, thanks for clarifying, that makes Haveto even more interesting. Running AI directly on-chain with auto-sharding is a pretty bold move, and it shifts the framing completely. I was looking at it through the lens of orchestration pain I run into daily, but what you’re building is the kind of foundation that could actually make those higher-level workflows far more resilient. If devs can trust that models execute deterministically and scale without central bottlenecks, then context-switching, backlog expansion, even end-to-end orchestration like I’ve done with Claude and ScrumBuddy start to look very different. Excited to see how the blockchain layer evolves, because that unlock could ripple across the whole stack.
Really impressed with what you’re building, Umang! Running AI directly on-chain with auto-sharding is a bold move, and the scalability potential is exciting. Curious about two things:
What’s been the biggest technical breakthrough so far that makes on-chain AI execution viable at scale?
From a go-to-market perspective, are you seeing more traction from developer communities, enterprise pilots, or crypto-native builders?
I help brands build AI-powered, high-converting websites that drive measurable growth. Always interested in connecting with projects pushing boundaries like this—your approach could open up some exciting possibilities for brands and businesses looking to integrate AI at a deeper level.
Would love to see a quick demo or breakdown of a real AI-on-chain workflow too. That could really help people visualize the impact!
Really appreciate your kind words and thoughtful questions.
On the technical side, the biggest breakthrough has been combining auto-sharding with on-chain AI execution. That’s what allows workloads to scale without the usual congestion or cost spikes, something most blockchains can’t handle when demand rises.
From the go-to-market view, there’s strong interest across the board. Developers love the flexibility (any language via Docker), enterprises value transparency and cost efficiency, and crypto-native builders see the tokenomics and scalability as game-changing.
And you’re absolutely right, a demo brings all of this to life far better than words. If you’d like, let’s schedule a quick call and we can walk you through a real on-chain AI workflow. It’ll give you a firsthand look at how the pieces fit together and where this can create real value for brands and businesses.
Feel free to also reach out anytime at umang@haveto.com
I am always happy to connect, answer questions, or explore ideas further.
Reading your post reminded me of something I saw yesterday .... PAI3’s upcoming ‘AI Box.’ Their idea is to put compact but industrial-grade AI hardware in the hands of individuals, who then plug it into a decentralized network. It flips ownership of compute away from big data centers, just like your approach flips AI execution away from off-chain servers.
what you think: do you see on-chain AI (like Haveto) and decentralized hardware (like these boxes) as complementary — or are they competing visions of where AI infra should live?
The way we look at it, decentralized hardware and on-chain AI can both play important roles in shaping how compute and intelligence are distributed. Haveto’s focus is on keeping AI execution transparent, scalable, and trustless on the blockchain, while still leaving space for different kinds of infrastructure to plug into that vision.
The exciting part is seeing how these different pieces of the puzzle can evolve together and open up new possibilities for builders and users.
Yes..........we see on-chain AI and decentralized hardware as complementary, each enabling new possibilities for builders and users.