You land on https://asterwise.dev/ (Aster by AsterWise) for the first time. What's unclear, what makes you leave, and what would make you sign up?

Lyle Swiftneutral

The core concept of automatically routing tasks to the best LLM is interesting. My main concern is how much control I actually have. The page mentions "intelligent auto routing" and "we train Aster to recognise what your task needs," which sounds like a black box. I'm used to carefully managing prompt versions and model choices for cost and latency, especially with my SMB clients who have tight budgets. While the cost savings are appealing, I'd need to see more about how I can influence or override the routing decisions, and what the actual pricing model is beyond just showing estimates for 100 turns. I'm not sure I'd sign up without more clarity on that.

Simulated interest: 40/100

Teresa Dickensneutral

The core concept of intelligent auto-routing is exactly what I'm grappling with. My primary concern is the fragmentation of model usage within my team – it's causing significant overhead in cost reconciliation and debugging. The idea that Aster can dynamically select the best model for a task is compelling. What isn't immediately clear is the level of control I'd have over the routing policy and how effectively it handles custom internal tools and bespoke models, beyond the curated collection. While the cost savings are attractive, I'd need to see concrete proof of performance and a clear integration path before signing up, especially regarding governance and auditability.

Simulated interest: 60/100

Sergio Kilbackneutral

The core idea of automatically routing tasks to the right AI model is intriguing, especially the promise of cost savings and better performance. However, it's not entirely clear how "intelligent" the routing actually is, or if it truly understands the nuances of a task without me needing to specify the model. The comparison charts are helpful for cost and success rates, but the "unverified estimates" caveat makes me pause. I'd need more concrete proof of its effectiveness and a clearer understanding of the setup process before signing up. Specifically, I'm curious about how it handles exceptions or tasks that don't fit neatly into pre-defined categories.

Simulated interest: 40/100

Usha Deshpandeneutral

The core concept of intelligently routing tasks to the right AI model is compelling, especially the promise of optimizing cost and accuracy. As someone who constantly juggles different AI models for client projects, this is a huge pain point. What's a bit unclear is how granular the control is for custom workflows and if it integrates seamlessly with Make or similar platforms. The pricing table showing cost savings over 100 turns is interesting, but I'd need to see how it applies to my typical project scope. To sign up, I'd want a clear integration path for my existing tools and perhaps a more detailed breakdown of how the 'Aster Context Method' works in practice for non-coding tasks.

Simulated interest: 60/100

Cedric Pourosneutral

The core concept of "intelligent auto-routing" for AI models is intriguing, especially the idea of matching tasks to the most suitable model to balance cost and performance. However, the page doesn't immediately clarify *how* this routing works in practice for someone like me, who prototypes workflows with a small engineering team. I'd leave if I couldn't quickly grasp the practical integration or see concrete examples relevant to our use cases, like automating internal tools. To sign up, I'd need to see a clear, simple API integration guide and perhaps a sandbox environment to test the routing logic with our own prompt variations. The claims about cost savings and task success are compelling, but I’d need more than "unverified estimates" to be convinced.

Simulated interest: 60/100

Timmy Simonisneutral

The core concept of intelligent routing for LLMs is immediately appealing. My team juggles ChatGPT and Anthropic, and the idea of an automated system selecting the best model for a given task, especially to optimise costs and performance, is exactly what we're grappling with. What's unclear is the actual integration process – how seamless is it to plug in our existing scripts or custom tooling? I'd leave if the setup seemed overly complex or required a complete overhaul of our current workflows. To sign up, I'd need to see more concrete examples of how it handles context for complex coding tasks and clear documentation on the pricing structure beyond the high-level cost-per-turn estimates. The promise of reduced costs and maintained performance is very attractive, but I'd need to validate it with our own data.

Simulated interest: 70/100

Patsy Fadelpositive

Landing on Aster, I'm immediately drawn to the promise of "less choosing, more doing," which totally resonates with me as a no-code builder. The idea of intelligently routing tasks to the right AI model without me having to figure it out is super appealing, especially for cost savings. The cost breakdown showing Aster's significantly lower estimated API costs compared to others is a huge plus, given my focus on monthly software spend. What's unclear is how seamless the integration would be with my current workflow and if it truly handles all the 'choosing' without me needing to tweak settings constantly. I'd sign up for a trial to test its actual performance and see if it simplifies my project setup, especially for those fixed-fee projects where efficiency is key.

Simulated interest: 75/100

Indira Naikneutral

The core idea of Aster, intelligently routing tasks to different AI models, is very appealing. It directly addresses my current challenge of manually selecting the best AI for each specific reporting or anomaly check task. What's unclear is the "custom agents, workflows and general work" aspect of Aster-Work; how granular can I get with defining specific workflows and custom agents for my financial reporting needs? The cost comparison table is interesting, but I'd need to see how that scales with more complex, multi-step financial anomaly detection rather than just coding turns. To sign up, I'd need a clearer understanding of the customization options for Aster-Work and perhaps a trial to test its effectiveness with my specific data and reporting formats.

Simulated interest: 70/100

Claude Padbergneutral

The core concept of intelligently routing tasks to different AI models based on their needs is exactly what I've been exploring to optimize our studio's workflows and costs. It's unclear how "training the decision" actually works from a user perspective or what specific "provider keys" I'd need to bring. I'd leave if I couldn't quickly understand the practical setup for integrating this with our existing API usage. To sign up, I'd need a clearer path to seeing this in action, maybe a trial or a demo that shows how it handles summarizing research or drafting copy, and a straightforward pricing model. The promise of cost savings and better performance is compelling, but I need to see the implementation details.

Simulated interest: 70/100

Janis Malekneutral

The core concept of intelligently routing tasks to different AI models is intriguing, especially the promise of cost savings and optimized performance. However, the page is a bit vague on the specifics of how this 'intelligent routing' actually works. It mentions reinforcement learning and task analysis, but I'd need to see more concrete examples or data to validate those claims. The claim of "Estimated total API cost across a growing coding conversation" and "Estimated share of tasks completed successfully" is interesting, but the "Unverified estimates" label for task success is a significant concern for me. I need to know how these costs and success rates are benchmarked. To consider signing up, I'd want to see a clear, transparent breakdown of the routing logic and, crucially, a way to run my own controlled tests comparing direct API calls versus Aster's routing. If the cost savings are demonstrable and the model selection process can be audited, it might be worth exploring.

Simulated interest: 30/100