Subscribe
Sign in
Octomind QA Agent — AI-powered end-to-end testing
An agent-powered QA tool that knows what to test without you telling it. Your very own junior QA that builds your tests from scratch and runs them to check your app for bugs. You build your app, we make sure it works.
68
Replies
flo merian
Hunter
📌
Meow Product Hunt!
A few months ago, I hunted Octomind, the AI-powered testing tool for web apps.
Since then, the team has kept shipping continuously.
Today, Octomind introduces QA Agent — a tool that knows what to test without you telling it.
Over to you! Go to octomind.dev and start shipping with confidence ✌️
Share
@fmerian @germandrummer92 @shivamchhuneja Thank you so much! Being able to release any time is part of our engineering culture here at Octomind. And decent test coverage is one of the stepping stones to make this happen :)
Congrats on launching @marc_mengler @germandrummer92 @ma_zah @daniel_rodler and the entire Octomind Team! 🎉
An AI-powered testing tool for web apps that catches bugs before users do is such a valuable addition for developers. The fact that Octomind automatically writes and updates tests without needing manual input is a massive time-saver for teams, and the ability to integrate it into CI/CD pipelines makes it even more powerful. Exciting stuff!
A couple of questions about the future of Octomind:
Handling complex app testing: As your AI evolves, how do you plan to address the needs of more complex web applications with dynamic elements and integrations? Are there plans to train your AI to handle intricate, real-time data or user interactions, which tend to be more difficult to test?
Test coverage and customization: Does Octomind allow developers to customize or tweak the tests it generates, or is it purely autonomous? It would be interesting to know how much flexibility dev teams have to adjust the AI-generated tests to fit specific use cases or edge scenarios that are unique to their app.
For some feedback:
The concept of automating tests with AI is fantastic, but I believe offering clear test coverage reports (e.g., what’s been tested, what potential gaps exist, etc.) would be incredibly helpful. This would give developers greater visibility into how thorough the AI’s testing is and ensure that they’re covering critical areas of their apps.
Suggestion: Consider enhancing the platform with detailed test coverage analytics to allow users to review and feel confident that key aspects of their web apps are being thoroughly tested. This could help teams prioritize which areas need manual intervention or further scrutiny.
I also noticed there’s no live chat support on the site. Adding a customer support chatbot could provide quick answers for users setting up their tests or trying to integrate the tool into their CI/CD pipeline. It would improve onboarding and make troubleshooting much faster for new users.
Overall, Octomind is tackling a significant pain point for developers, and I’m excited to see how it grows from here. Keep up the amazing work! 🙌
@marc_mengler @germandrummer92 @ma_zah @uvkewate
Complex app testing
As long as interactions stay inside the browser we will be able to handle them. There are of course areas which are more challenging like canvas or real time communication. We have to see how frequently people are asking for it. If a serious need arises, we are all in to tackle these challenges.
Customisation
We always start with a prompt. The prompt can either come from our system or you can enter your own prompt - or copy it from your test management system. This is the first level of customisation.
Once the first flow is generated you have full control over it. You can review and modify every single step it took. You can also add additional steps if the need arises. Or you can further refine your prompt and regenerate either all steps or you keep those which are already good. This is the second level of customisation.
Should you be unhappy with the selectors the system came up with you can easily directly enter Playwright locators. This is the third and most technical level of customisation.
Basically you are always in full control.
Coverage
I love the idea. Not an easy one, I have to say. But definitely super helpful if it can get solved in a good way. It is on our list.
Support Bot
Also a good idea. Thank you. I think we have to build up a more comprehensive knowledge base first to make it really helpful. I am sure we will install one once we are at this point.
Other than that: Thank you so much for sharing thoughts and feedback! This is much appreciated!
@marc_mengler @germandrummer92 @ma_zah @daniel_rodler most welcome and i am glad you found it useful. For support bot you can look into https://primecx.ai/
🧐 Good find
Would be interesting to hear from the founders what this new initiative involves? What is this new QA Agent angle all about versus what was launched half a year ago. Seems like an upgrade but im not sure what's better over the old version. the old version was already best in class when I tested it. thanks
@sentry_co Hi André, thanks for your kind words! In fact we streamlined the whole onboarding process to lower the adoption barrier and we made it much simpler and a lot faster to increase your test suite. In addition our AI agent is now much more capable of handling all the different HTML we keep finding around the internet :)
that is insane 🤯
the test cases it generated were pretty good! I needed to update a selector to be something less transient (it picked "has text: 178 images" in a image gallery 😂) but that was easy enough to do.
This could definitely make testing life less painful. Thanks!
@jonnymiles thanks for trying it out and providing feedback! Yes, we have to work on better selectors. The question for us is, what is the impact of a bad selector if you have auto-maintenance available? And can we make auto-maintenance understand that it fixed a selector for the 3rd time and that it should maybe choose a better one? Just some thoughts. We are on it :)
Let us know if you need any help with your testing!
@arjunrajkumar I am glad you like it! If your mobile app is build on react native and accessible via URL, yes, it should work! But if it's a native Android or iOS app, we don't support it yet. On which of these are you developing your mobile apps?
@killiandunne1 thank you very much! Did you try it out? Was there anything you liked or didn't like? Any comment would help us to improve.
Octomind is incredible. They've have finally built a QA agent that actually works. Marc and team are great to work with!
Ocotmind is fantastic - Maria and her team are doing a great job and it's been an inspiration to see how fast they can ship new features. 80 releases since February is wild!
@nishkarsh_srivastava Thank you so much. This is exactly the idea. Help people to get rid of the annoying part of e2e testing so that they can focus on building!
@marc_mengler great work with the QA agent, really innovative and an AI-based solution that's not only fluff!
Out of curiosity, what kind of auto-generated tests usually are most liked / kept / used by your customers?
@marc_mengler @vdelitz This largely depends on the type of app being tested. For many apps, it's often clear what needs to be tested because common applications tend to follow standard workflows. However, things become less straightforward when you move into the SaaS space, where apps are much more domain-specific.
@vdelitz I'm glad you like it! Indeed, our first two attempts at nailing the AI part failed, but the third one was successful. It's easy to build a demo, but turning it into a robust product is difficult and always takes longer than initially estimated.
The most common auto-generated tests deal with account creation & management and different payment options, so the critical, security and cash-critical parts!
Hey Makers!
We’re back! After 60,840 generated tests since we went public with beta, we’re launching Octomind with even faster test auto-generation to take your testing chores off your plate.
1. You build.
2. We generate tests.
3. You refine.
4. We run them to catch bugs before your customers do.
Back in February, we opened the Octomind beta to real users. Letting real-world apps with all their edge cases and challenges use Octomind gave us insights we couldn’t have gathered otherwise.
Over the past 6 months, we’ve been rethinking and fine-tuning. We realized our AI agents perform better with more guidance, and you like more control over the tests they generate.
1. We start with a base of auto-generated tests.
2. From each one, you can tell Octomind to create three more sequential tests.
3. Click by click, you’ll build a comprehensive test suite, all under your control.
Right now, Octomind covers your basic app functionality tests and runs them for you. You can handle the complex stuff in your current stack, deciding what gets automated and what remains manual.
Octomind saves you serious time by automating test creation and execution. How much time?
On average, it takes about an hour to outline, code, and verify a test case. Since our last launch on Product Hunt, we’ve saved a total of 60,840 hours for teams to focus on building great products.
And there’s more! Soon, we’ll also maintain your tests automatically - detecting changes in your app and updating the tests accordingly.
To celebrate, we’re offering a 20% discount on our paid tier for the next 12 months with the Product Hunt promo code CATMEETSOCTO. If we get 500+ votes today, we’ll bump that discount to 30% for everyone!
Let me know what you think, I’m always listening!
Marc
The UI looks super intuitive!
Happy to never write tests again :D
How do you manage updates on the front or back end?
@canger Hi Can, since we are running the test cases we also see them fail. If this happens we are starting an extensive analysis process to find out if the test needs update (maintenance) or if it is a real problem with the frontend or the backend. This is what we call auto-maintenance. However, we only have it ready in an experimental version. If you want to use it, let us know, we can turn it on and demo it. Auto-maintenance for the general public might be the next Product Hunt launch :)
@artagel Thanks so much, Dan! We're continuing to push forward with our mission, and I’m already excited about the next set of features we’ll be rolling out soon. We’re on a mission to redefine how testing is done!
@artagel same here, used it from the beta. The discovery has gotten way better, where they crawl your app and generate UI tests. It's already caught a few bugs before they hit production, where I accidentally broke something on a page. One time it was on my billing page, so it's already paid for itself.
@harry_sin1 so far we want to nail web apps at a level we're satisfied with and ship AI test auto-maintenance. Once we have that in place, mobile is the next step.
Hey Octominds, awesome tool! It looks really promising and could save a ton of work.
How well would you expect it performs today on more complicated webapps and workflows? Especially considering user interaction through text or files?
Thank you and enjoy your launch day!
@paul_von_berg Great question. I think we currently still struggle with files unfortunately but we have gotten that request a few times so will definitely get to it eventually.
What do you mean by text, just natural language inputs, or rather form data? It really depends a little on the complexity of the process and UI, but of course the best way is to just try it.
We love looking into the pesky complex websites that our Agent still struggles to understand, the web is really a crazy place!
@germandrummer92 awesome, that makes a lot of sense & it is for sure a place of endless complexity.
I was thinking of just natural language inputs but both -also text fields- seems super relevant to have proper test cases for because there obviously comes in a lot of volatility/ attack potential.
I will check it out, thanks Octominds ;) and looking forward to test that feature!
Launching soon!
@fmerian Congratulations on this wonderful launch! I was wondering how your QA agent reacts to changes in the app during development?