I’m building The Tank — a “collective AI” that can chat + build in one place!
Hey Indie Hackers — I’m building thetank.io because I’m tired of the same loop:
• Ask AI for help → it generates code
• You paste it somewhere else → preview breaks / context is gone
• You bounce between 6 tools + 12 tabs → the “build” is fragmented
• You come back later → the code/preview/decisions are lost
What the Tank is
thetank.io is one product that does two things really well:
- Conversation — requirements, iteration, decisions, debugging
- Building — generates real code + shows a live split-pane preview (v0-style)
The core idea is “Collective AI”:
Instead of betting everything on one model, the Tank routes across multiple models (via Vercel AI Gateway) to pick the best brain for the job — planning, coding, debugging, and writing.
What it’s trying to feel like
If the best parts of chat AI and app builders had a baby:
• you describe what you want
• it scaffolds the app (front + back)
• it renders it live
• it remembers the build so you can return later and keep going
Where I want feedback (and eyes)
I’m currently focused on making the builder experience bulletproof:
• stable, interactive preview (no flaky fallbacks)
• persistent code + preview per chat (so revisiting a build actually works)
• clear “chat → build” behavior so it knows when to generate a real project structure
If you’ve built anything involving live previews / sandbox rendering / multi-file project execution:
What’s the cleanest architecture you’ve seen for reliable previews AND persistent builds?
If you want to poke at it (and roast it), here’s the link:
https://thetank.io
Really interesting project! 👏 TheTank has a clean and modern look, the focus on helping projects ⚡ pitch, validate, and grow feels timely, especially in an era where founders need actionable feedback more than a generic “like.”
The UX feels uncluttered, which is important for a platform meant to connect founders with opportunities. One thing I’m curious about is how you’re approaching conversion flow for new users like what the first experience looks like after signup and how it guides them to take action.
Overall, it feels promising great job getting the concept looking sharp!
Thanks! I am still working on trying to figure out the best conversion flow. I would love any recommendations!
Thanks, I really like that you’re thinking about conversion this early.
If I were refining the flow, I’d focus first on making the outcome extremely clear in the hero section. For your audience, the message should instantly communicate speed and relief something that makes a technical founder think, “This saves me weeks of setup.”
After that, I’d reinforce the promise with a simple before/after contrast (months building infrastructure vs. launching in a weekend), then build trust by highlighting clean architecture and flexibility before pushing the main CTA.
Finally, I’d make sure the primary CTA is outcome-driven rather than generic something aligned with your speed positioning.
Small structural tweaks like this can significantly improve clarity and conversions. Happy to share more specific suggestions if helpful. 🚀
The "chat → build" continuity is the right problem to solve. I built something similar but realized most people don't want to build — they want the thing built for them.
The multi-model routing is smart. I've found GPT-4o for planning, Claude for code, and a cheap model for summaries is the sweet spot. One brain rarely does it all well.
For persistent previews: have you considered treating each conversation as a "project" with a git-like history? That way revisiting = checking out a branch, not regenerating.
Curious: are you targeting builders who want to learn, or delegators who want the result? The UX diverges sharply based on that answer.
Exactly, that! I want to target builders who want to learn!
Cool concept — but right now “a collective AI that can chat & build” reads more like a wishlist than a differentiated product, especially with so many multimodal AI tools already out there.
Without clear boundaries on scope, reliability, and real user problems you’re solving (beyond “it can do everything”), it’ll be hard to convert early adopters into paying users.
If you want feedback that actually helps you tighten your value prop and messaging so people don’t just nod — but want to buy — I’d be glad to jump in.
Appreciate the feedback. I just made some improvements! I've created a dashboard functionality in the builder that allows users to connect to their databases and use the tank as a hub for their connected dashboards.
Nice! Connecting dashboards is a strong move — now it’s all about showing how this actually solves real user problems so early adopters can’t resist.
Strong framing on “chat → build” continuity. For persistent builds, the cleanest pattern I’ve seen is versioned project snapshots + deterministic preview: each chat step commits a build artifact (files + deps + env), preview spins from that snapshot, and you keep a lightweight intent log for replays. Helps avoid “flaky preview” drift. Are you storing a per‑chat project root + dependency lockfile to keep previews deterministic across sessions?
I am a beginner in Saas just prepared a product and actively looking for ways to get customers, so my thought might not be valuble , but I will suggest to focus on this specific segment and keep the other opportunities into a seperate portal i.e front end alone , as the cutomer would love it if it focusses on thier problem rather than generic. :)