Blink Agent Builder - You can now vibe code agentic AI apps
by•
Blink is the first vibe coding platform that builds AI agents. Describe what you want — Blink creates an agent that thinks, uses tools, and completes tasks end-to-end. Built-in web search, code execution, vector database, sandbox, and 180+ AI models. We used it to recreate Cursor, Perplexity, and Shortcut in minutes. Now it's your turn.
Replies
Triforce Todos
This is honestly one of the coolest launches I’ve seen here in a while. The idea of just describing an agent and having it work is wild.
Hey team, congrats on the launch. This looks very promising.
I’m a non-technical founder evaluating Blink for a real internal operations system, not a side project or experiment. Before committing longer-term, I want to make sure I fully understand how things work in practice.
The most important question for me:
Once an app is built and running, what does it actually cost to keep it running if we make no changes or only very small changes in a given month?
Put simply:
Do credits get used only when building or changing the app, or also during normal day-to-day use?
If the app is live and being used by our team, but we’re not rebuilding anything, does the cost stay flat on the subscription plan?
A few follow-ups:
I see there’s currently a 50% discount on annual plans. If that’s going to change, do users get a heads-up, or does it just end?
I’m planning to start on the monthly plan to test Blink properly, but would like the option to switch to annual before pricing changes.
On the annual plan: if we sign up yearly and decide to cancel after a few months, is the unused portion refunded, or does “cancel anytime” mean access continues until the end of the term?
When deploying an app, can it run on our own domain (for example app.ourcompany.com), or does it need to live on a Blink subdomain?
Appreciate the clarity! Trying to evaluate Blink responsibly for long-term use in a real business.
Hey ! Why is it different from Jynova for example ?
Super cool product, Kai! I love just how robust the execution is. One note on what I find is missing from a product like this is giving the user an option to be asked a couple of clarifying questions before hitting generate the first time. Sometimes I have a clear idea of what I want, other times I spend valuable credits giving course corrections because I didn't think through some of the obstacles it could knock up against. If there is a way to have Blink give the option of asking some clarifying questions that sparks more details and gives a more effective first pass, that would be great!
So the hard part isn't prompts, it's trust. Blink.new Agent Builder bundling web search, code execution, a sandbox, and a vector DB is a strong combo. How do you surface exactly what changed and what tools ran each step? A tight audit trail will win teams.