I built a free tool to check if AI recommends your company — or your competitor's

Interesting idea. One thing I'm curious about: AI answers can vary a lot depending on the model, prompt, and even over time. How are you handling that variability when calculating the score? Is it based on multiple prompts/models or just a single snapshot?

Smart wedge. That “is AI even mentioning us” question is the new version of “are we on page one,” and most teams still cant answer it without vibes.

Shipping a free check with no signup is the right move for Indie Hackers too. Instant gut-check, low friction, easy to talk about. The part that’ll keep people around is the “here’s what to fix, draft already written” angle. Score alone is interesting. Actionable fixes are sticky.

I’m building Make it RAIN for a neighboring problem on the builder side, not “does AI recommend you,” more “once you exist, how do you actually get buyers and a monetization path.” Different layer, same feeling of shipping something I needed myself.

One question: when the model says a competitor and not you, how much of the guidance is content/SEO vs product positioning vs just weak web presence? That distinction seems important so people dont just spam more blog posts.

This is a smart angle, live-querying the model instead of guessing from static signals. We've been building the adjacent piece: a free checker that scores whether a site is technically parseable at all (crawler access, schema, llms.txt, answer-first structure) before it ever gets to the recommendation stage.

Ran a couple of benchmarks on that side and the two scores genuinely diverge. Same average technical score across two different verticals, totally different failure patterns underneath. Technical readiness looks like a prerequisite, not a predictor, your live-answer approach is closer to actually measuring the outcome.

Curious if you're seeing sites that pass your check but would fail a basic crawler-access test, or if by the time a site shows up in an AI answer at all the technical stuff is usually already solved.

Checking whether AI recommends your company vs a competitor is a smart niche — this is going to matter a lot more as people start using AI assistants for research/shopping decisions. How are you sourcing the "ground truth" of what AI models are actually saying about a brand?

Useful framing. One thing I'd add for B2B SaaS, especially infra/payments, is that AI visibility is not just brand mentions. It is whether the model can find enough verifiable detail to recommend you without hallucinating.

The pages I would test separately:

• pricing/fees and what is not included

• who the product is for / not for

• integration path and time-to-first-value

• security/reliability claims with concrete docs

• support, refund, and edge-case handling

For payment products in particular, vague copy like "simple checkout" is weak evidence. A model needs facts such as settlement flow, webhook behavior, order states, supported chains/tokens, and where funds custody starts/ends. Otherwise it will either ignore the product or recommend a safer incumbent.

I like the problem you are targeting, but I would be careful with what a single live AI answer can actually prove.

One response can confirm that a model mentioned a company under one specific set of conditions. It does not necessarily prove that the company is consistently visible or recommended across buyer journeys.

The result may change depending on:

• the model;

• the exact prompt;

• the date;

• the user location or language;

• the conversation context;

• repeated runs of the same query.

So I would separate three claims:

1. Mentioned — the company appeared in this response.

2. Recommended — the model explicitly endorsed it for a defined need.

3. Consistently visible — the result repeats across prompts, models, and runs.

Those are very different levels of evidence.

For me, the strongest version of an AI-readiness score would show:

• which prompts were tested;

• which models were queried;

• how many repeated runs were performed;

• whether the company was mentioned or actually recommended;

• which competitors appeared;

• how stable the result was;

• what evidence supports each proposed fix.

A single answer is a useful diagnostic sample.

A repeatable test protocol is what turns it into something a company can trust.

How are you currently handling variation between repeated runs of the same prompt?

Congrats on launching. One thing I immediately wondered about is confidence in the recommendations. An AI-readiness score is a compelling idea, but users will naturally ask why the platform reached a particular conclusion and what they should prioritize first. If the reasoning behind the score is as actionable as the score itself, I think that becomes a much stronger reason for marketers to trust the insights and come back over time.

I specialize in independent product validation for early-stage SaaS. I evaluate products from a genuine first-time user's perspective, uncover where trust, clarity, or usability break down, and produce professional walkthrough videos alongside structured findings that founders can use to improve onboarding, product adoption, and marketing.

I like that you're framing the problem around buyer discovery rather than treating it as another SEO metric.

I'll be interested to see which recommendations customers consistently choose to act on first. Those patterns will probably reveal whether the product is really about AI visibility, competitive positioning, or reducing uncertainty around how buyers evaluate vendors.