When someone asks ChatGPT, Claude or Gemini to find or buy a product, the assistant has to read your site to do it. If an agent can't find your pricing page, can't read your product copy, or stalls at your payment form, the sale goes to whichever site it can handle. Dedica, which launched on Product Hunt on 28 October 2026, is a quick way to find out which of those problems your site has.

What it is

Dedica is a website checker. Its homepage says it "tests your site the way AI agents such as ChatGPT, Claude and Gemini use it". You type in a domain and it runs 125 checks across four layers: Discovery, Access, Usability and Payments. You get a score out of 100 and a list of fixes, with the ones that move the score most at the top. If you pay, you also get one instruction file per issue, written for a coding assistant to apply.

The makers listed on Product Hunt are Prathmesh Bhalekar and Prajwal. The product launched this week and its site names no company history or customers.

Who it's for

Dedica suits solo builders and creators who sell from a site they control: a course on your own domain, a SaaS with a self-serve checkout, a digital product shop where you can edit the templates. Those are the cases where a list of fixes is useful, because you can make the changes yourself or hand them to Claude Code or Cursor.

It is less useful if you sell through a hosted storefront where you can't touch robots.txt, page markup or the checkout. A score will tell you where agents struggle, but you may not be able to act on most of it. If you're still deciding whether agent buyers matter for your business at all, start with our look at whether listing your tool where AI agents shop brings in users.

What it actually does

The four layers map onto the steps an agent goes through before it can buy something:

  • Discovery: can an agent find your site and the right pages on it?

  • Access: once it arrives, is it allowed in, or do blocks and bot rules turn it away?

  • Usability: can it read the content and work the interface?

  • Payments: can it complete a purchase?

The homepage names the layers without listing what sits inside each one, so the descriptions above are the plain reading of the layer names rather than Dedica's own definitions.

The free report gives an overall score and grade, a per-layer breakdown and up to 20 example findings, each with evidence and a short explanation. Scores are published at a public address in the form getdedica.com/score/yourdomain.com, and the homepage says a check takes under a minute.

The developer page shows more of how the system is built. There is a free, read-only public API that returns a cached report as JSON, rate-limited to 30 requests per minute per IP, with no key needed. There are also two MCP servers: one with tools to check or run a scan from inside an agent, and one that answers questions about Dedica's documentation. The sample response shows each layer carrying a score, a maximum score and a weight. The page doesn't say how those weights combine into the overall number, and it doesn't list the 125 checks.

What we could and couldn't test. We tried to run the free check on hackrlife.com today. The score page loaded and showed a "Checking the latest report" screen, but our fetch tool can't run the page's live scan, and Dedica's report API refused our tool's requests under its robots rules. So we don't have a score to show you, and nothing in this review is based on hands-on results. Running it yourself in a normal browser is free and needs no signup.

Pricing (as of 29 October 2026)

Dedica has two options and no subscription, according to its pricing page:

Option

Price

What you get

Free check

$0

Score and grade for one domain, layer breakdown, up to 20 example findings. No signup.

Fix kit

$30 until 30 November 2026, then $50

Every finding for the domain, one instruction file per fix for Claude Code, Codex, Cursor or GitHub Copilot, the full unredacted report, a .zip download and an emailed report link.

The fix kit is a one-off purchase for one domain at one point in time. Checking a second domain, or getting an updated kit after you've changed your site, means paying again. Payments go through Stripe, and there is a 30-day money-back guarantee with no reason required.

Getting started

A sensible first session takes about fifteen minutes:

  1. Go to getdedica.com and enter the domain your customers buy from. If your checkout lives on a separate domain or subdomain, check that one too, since the score is per domain.

  2. Note the overall score and the per-layer split. A weak Payments layer matters more to a shop than a weak Discovery layer does to a site that already gets plenty of agent traffic.

  3. Read each of the free example findings and the evidence attached. Sort them into three piles: things you can change today, things your platform controls, and things you're unsure about.

  4. Test the top two findings yourself. Ask ChatGPT, Claude or Gemini to find your pricing page and describe what you sell, then ask an agentic browser to get as far through checkout as it can. If the agent trips where Dedica said it would, the finding is real for your site.

  5. Fix what you can from the free findings, then run the check again and see whether the score moves. Re-running the free check costs nothing.

Only after that would we consider the fix kit, and only if the free findings proved accurate in your own tests.

Alternatives

A manual agent test. Ask two or three assistants to find, summarise and buy your product, and write down where each one fails. It's free, uses the real agents your customers use, and gives you evidence you can trust. Pick this if you have a small site and an hour to spare. It won't give you a checklist or a score to track over time.

Google Lighthouse. The audit built into Chrome's developer tools covers accessibility, SEO basics and page structure. Agents lean on the same clean markup and labelled buttons that screen readers do, so Lighthouse overlaps with Dedica's Usability layer. Pick it if you want a free, well-documented audit and don't need anything on payments or agent access.

An SEO crawler or agent-ready SEO data. Tools in this category crawl your site for broken links, blocked pages and missing metadata, which overlaps with Discovery. Our review of OpenSEO, open-source SEO data your agent can use, covers one option. Pick this route if being found and cited is your main problem; our piece on publishing your own data so ChatGPT cites you goes deeper on that side.

Limitations

  • No published scoring method. Dedica lists the layers and shows that each carries a weight, but it doesn't publish the 125 checks, how they're weighted, or any evidence that a higher score leads to more agent sales. That makes the number hard to interpret and hard to compare between sites.

  • You pay again to re-check. The kit is a snapshot. Fixing your site and wanting an updated kit means another payment, which adds up if you iterate.

  • Instructions only. Dedica writes the fixes for a coding assistant; it doesn't apply them. You need access to your site's code and a developer or coding agent to use the kit.

  • A very new product. It launched this week, from a small team, with no track record on the site. That's normal for a launch, and it's one more reason to lean on the free tier first.

  • Hosted platforms limit what you can fix. If your shop runs on someone else's checkout, many findings may sit outside your control.

Verdict

Worth the free check if you sell online; hold off paying for the fix kit until Dedica publishes how its checks are scored.

The free check costs nothing, takes a minute, and gives you a structured list of places where agents might struggle on your site, with evidence for each. That's a useful prompt to run your own agent tests, and the public API and MCP servers make it easy to fold into an agent workflow later. The $30 kit is cheap, and the 30-day refund lowers the risk. Without a published method, though, you can't tell whether the score reflects how ChatGPT, Claude and Gemini actually behave on your site or how much any single fix matters. Use the free findings, check them against real agents, and come back to the kit once Dedica shows its working.

Comments

Avatar

or to participate