Show HN: Choosing the right SaaS is a chore, so I built something to fix that | Hacker News Reader

Show HN: Choosing the right SaaS is a chore, so I built something to fix that

gralio.ai

Cool! How do you ensure the accuracy and neutrality of its recommendations, and are there plans to incorporate user feedback into the AI's learning process for even more precise matches

We rely on a multi-layered validation approach, where multiple independent sources must confirm any given data point before it’s presented. Aravind described this well on the Lex Fridman podcast. Behind every green dot is a chain of LLM prompts, reviewers, and “critics” ensuring accuracy. On top of that, user feedback and browsing patterns continuously refine the system’s weighting, so recommendations get better over time. We’re also working on a feature to show which tools your competitors are using. I think this will be really great.

Wow that's a time saver! But Claude or Perplexity can also give me a table, no? Not sure I see use in going to a dedicated tool.

Sure, Perplexity can present a simple table, just like it can list hotels, but you’d still visit Expedia to finalize your booking. There’s more to making informed decisions than just seeing a list of options. Right now, we’re focusing on surfacing deeper insights: detailed features, aggregated review summaries, and company health. Our goal is to go beyond a basic search and provide all the data points you need :)

Hi HN, In my previous role as a CTO at an Uber-like scaleup, I had to manage and evaluate around 60 different SaaS subscriptions. It was painful. As software engineers, we’re often asked to help select all sorts of software—payroll, payment processors, marketing automation, CX tools, call center software—tasks really meant for domain experts. This problem is universal, affecting businesses of all sizes. The SaaS landscape, unfortunately, is drowning in marketing fluff, biased paid reviews, and general noise.

So I built an AI to cut through it. Under the hood, we classify 4 million reviews to extract real signal. We use GraphRAG and a chain of LLMs to ensure quality. Similar to Perplexity, multiple independent sources must confirm each piece of information before we trust it.

Like Garry Tan (YC) suggests, chat interfaces might not be the future. Users prefer familiar UX, and the next evolution of the internet should be intelligent sites delivering personalized content. You’ll still get the benefits of LLMs—just in a more digestible format.

We started by tackling comparisons first. No two SaaS tools are the same, each with unique features and target markets. Our comparison engine alone can save you hours of research. Next, we plan to add a product discovery module to guide you to the perfect tool.

I believe that until AGI is widespread, business will continue as usual. Vertical AI agents will need to integrate with familiar SaaS products—how else can they coordinate work? Even AIs will need Trello, Gmail, Slack, Your CRM and the existing boring toolbox. Meanwhile, the amount of new software hitting the market will soar, making comparison and distribution harder than ever.

We’re VC-backed, and our guiding principle—our “don’t be evil” policy—is to provide unbiased software recommendations. Our long-term vision is to offer premium buyer features and competitive intelligence. Eventually, we’d love to integrate with AI providers for seamless migration, integration, and customization support.

Some examples: Linear / Trello https://gralio.ai/compare/Trello-vs-Linear Notion / Clickup https://gralio.ai/compare/Notion-vs-ClickUp Gusto / Rippling https://gralio.ai/compare/Gusto-2-vs-Rippling Webflow / Retool https://gralio.ai/compare/Retool-vs-Webflow

Let me know what you think!

Reply on news.ycombinator.com