I've been using this product for over a month and genuinely see the value in what it's trying to do. I wanted to give it 5 stars, but a couple of significant issues have chipped away at that score.

Issue #1 Prompt Optimizer Limitations

The prompt optimizer only works within the three major AI platforms (ChatGPT, Claude, etc.). If you've built your workflow anywhere outside of those, the feature simply isn't available to you. This was actually one of my primary reasons for purchasing, and I even asked about it before buying. The problem is that the limitations were never clearly communicated and there's an assumption baked in about how users work, and if you don't fit that mold, you're out of luck. That kind of ambiguity is frustrating, especially post-purchase. You can't even optimize prompts you have saved on the platform.

Issue #2 The AI Credits System (The Big One)

This is the dealbreaker for me.

I purchased the highest tier and still ran out of AI credits multiple times. Credits being consumed for vectorization is understandable, but having them drained by agent tool calls is where things fall apart fast. I've since downgraded to Tier 2 and scaled back considerably and I'm still hitting the wall.

Here's the real problem: when your daily AI credits run out, your memory is completely dead to your agent and the platform is essentially bricked for 24 hours. You can't add anything or do anything other ran read what you put in there. The glaze on top is when you run out the reset is not from a fixed reset time each day, but from the exact moment you ran out. So if you're working late one night and hit the limit at midnight (literally what I'm doing right now at almost midnight), you can't pick up where you left off the next morning. You're locked out until that same time the following night, meaning your agent has no memory of what happened and you lose all continuity should you decide to push forward. I must wait until 11:49pm tomorrow to provide my agent with the context of the work I did tonight. Or bring the documents over to another platform to continue.

That kind of unpredictability makes building any real reliance on this platform a dangerous move, which is honestly why I'm glad I downgraded before I became too dependent on it.

A Better Alternative I've Found

I picked up Diaflow a few months ago, and it's not better than diaflow but it handles this time limit thing much more gracefully (there is no 24 hour reset in place to avoid spikes in tool calls. Vectorization is built in and doesn't consume credits. And you can make your database accessible across all your agents not just the ones living inside that ecosystem.

My Suggestion to the Team

The core idea here is genuinely cool, but the credit exhaustion experience needs a rethink. Rather than completely bricking the platform, consider a graceful degradation model..maybe writing features get paused, or new documents revert to their original unvectorized format until credits refresh. Zeroing out something as critical as memory feels punitive. The current setup almost feels designed to maximize the pain of running out, ensuring users feel pressured to buy more credits rather than simply wait for the minimum 24 hour reset.

It would almost be better to not have AI built in at all.