Ben Gendler

Congrats on the launch! Tracking spending while traveling is one of those problems that sounds simple until you're splitting costs across three currencies with four friends. Curious - do you handle shared expenses with groups? I've found that the messiest part of travel spending isn't tracking your own costs, it's figuring out who owes what. Especially when half the group Venmos and the other...

Creating personalized landing pages and deal rooms is nice, but knowing whether the VP you sent it to actually looked at it and which sections they spent time on - that's where the real sales intelligence is. We're building a B2C product but we also talk to investors and partners, and being able to create a personalized page for each conversation and track engagement would be really useful....

LinkedIn is a big channel for us as founders and the problem is very real. I know I should be posting consistently but the time it takes to write something that doesn't sound like generic AI content is painful. How does Linkji learn your voice? Does it analyze your previous posts or do you set up a style profile manually?

Sounds great! We're pre-launch so our AI visibility is probably close to zero right now, but setting this up early so we can track it from day one makes a lot of sense. How often do the visibility scores update, and do you see big swings between model updates?

We're about to start running ads and this would have saved us a lot of wasted budget on creatives that "looked good" internally but nobody actually looked at the CTA. The attention heatmap is really clever - knowing where people's eyes go before you spend money is way better than A/B testing in production with real budget. The variant comparison feature is huge too, we always argue internally...

I've been coding 12+ hour days lately and my eyes are paying the price. The problem with the 20-20-20 rule is that nobody actually follows it without something forcing them to stop. Having the Mac go into rest mode at intervals is way more effective than a notification I'll just dismiss. The "smart" part is key too - I don't want it interrupting me mid-deployment or during a call. Does it...

Most dashboards just show you numbers and leave you guessing. We're about to start running Meta and Google ads for our launch and the idea of getting actionable recommendations on creative and budget allocation from day one instead of burning through budget while we figure it out is really appealing. How much ad spend data does Predflow need before the recommendations become reliable? Like does...

I've seen too many ecommerce brands use AI to create product photos that look cool but completely bomb in ads because they weren't designed with performance in mind. Starting from a single product image and getting launch-ready creative that's actually informed by what converts is way more useful than pretty pictures. We're not ecommerce specifically but we need a lot of ad creative for our...

This is becoming more and more necessary. We use AI coding agents daily and the amount of code that gets generated without proper review is honestly scary. Can you define custom policies, like "never ship code that touches auth without a manual approval" or is it more of a general quality check?

It's a big deal for accessibility. The focus on underserved regions instead of just adding more European languages is the right call - there's a massive gap there. How does Tiny Aya perform on Hebrew specifically? And is it practical to fine-tune on domain-specific data at this size, or is 3.35B too small for meaningful customization?

Most automation tools assume you already know what needs automating, which is half the problem. We're a small team using Slack, Notion, Google Workspace, and a bunch of other tools, and I'm sure there are workflows we repeat every week without realizing they could be automated. Having AI analyze the actual patterns and recommend automations instead of building them from scratch is a much better...

Having an agent that can write content, deploy apps, and call REST APIs from inside a chat thread is powerful for small teams who don't want to context-switch between 10 tools. How do the agents handle content that needs approval before publishing? Is there a review step built in or does it just ship?

I use a bunch of different AI-powered web tools and every single one wants my API key, which means I'm trusting random apps with my credentials and have zero control over which model they actually use. Having a browser extension that acts as a wallet where I approve each request and pick the model myself makes a lot more sense from a security standpoint. Does it work with local Ollama models...

Most language apps make things too easy so you feel good but don't actually learn. How many languages does it currently support, and does the pronunciation mode work well for non-Latin script languages?

Localization is one of those things that always gets pushed to "later" and then becomes a nightmare when you finally need it. How does it handle context-dependent translations? Like in our app, the word "network" means something very specific - does it learn domain-specific terminology or do you need to manually define a glossary?

I use Claude Code and Cursor daily and I've learned that you can't just trust the output blindly - agents will tell you they implemented something and you'll find out later it was half done or the tests were never actually run. How does Vet handle cases where the agent made a reasonable interpretation of an ambiguous request? Does it flag those as potential issues or only catch clear deviations?

Congrats! Are you planning to add any kind of filtering by tech stack or AI tool used? Would be great to see all the projects built with Lovable vs Claude Code vs Cursor for example.

For startups where each employee can wear different hats, it is very relevant! How does it learn the team dynamics - like does it pick up on who handles what, or do you need to explicitly configure roles and permissions?

Congratulations! Can the AI generate multi-step forms with branching logic from a single prompt, or does it mainly handle simpler flat forms and you add the logic after?

Being able to replay the exact failing request with the same inputs and IO locally is huge. How does it handle replaying requests that involve third-party APIs that are rate-limited or have changed their response since the original failure?