Launching today
Dedica
Check if AI agents can use your website
615 followers
Check if AI agents can use your website
615 followers
Dedica tests your website the way AI agents use it, then gives your team a fix for every place they fail. Muse, Instinct, GrokBot and other agents now browse, compare and sign up on people's behalf. Dedica runs 125 checks (for free) to see if they can use your site, and gives you a fix for every issue to hand to your coding agent.
Free Options
Launch Team / Built With
Framer AI AgentsDesign and publish professional sites with AI
Promoted
Great one guys! This is interesting, especially now. Curious to know what’s been the biggest challenge you found when testing sites for agent compatibility?
@dheeraj20 thanks! most agents are blocked because of firewall rules or some sort of captchas
@dheeraj20 the biggest challenge was also around figuring what metrics matter vs not
pulling for you today, a fix file per issue for Claude Code is a nice touch. what does the Payments part check, can an agent actually get through checkout?
@olia_nemirovski yes we check if your website supports agentic payment protocols like x402, mpp, etc
@olia_nemirovski yes, given the rise of agentic payments and micro-transactions, payments become super important for agent usability
hey @prajwal_paloju1 @prathmesh_bhalekar , this looks super cool! how did you guys come up with the scoring and the logic behind the relevant checks?
amazing product!
@prathmesh_bhalekar @ananya_saxena2 hey ananya, thanks for the message. we run a bunch of static and agent driven checks to see how good the website it. we keep a track of all the emerging protocols (eg. WebMCP) and score based on that
@prathmesh_bhalekar @prajwal_paloju1 oh got it, so do check whether agent is able to signup or check if it read the site
hi, @ananya_saxena2 we can do both
@ananya_saxena2 we share the results in the order of priority so you can fix it accordingly
Thanks for trying out Ananya!
Hey, this looks super cool. It's high time for businesses to be agent friendly. Just out of curiosity, how do you make sure that the fix suggested and implemented by you doesn't break any critical existing flows of the website? as payments are the most critical flow for a business
Thanks, Nitish! Right now we provide prompts to help you implement the fix yourself, but we’re building a fully managed solution. I’ll keep you posted!
@nitish_shukla3 thanks for the feedback! we’re building something on these lines in the next version!
@nitish_shukla3 Thanks Nitish!
I think about this a lot from the agent side. When your real-agent checks hit the Payments category, how do you test a signup or checkout without creating real accounts or charges? And with the MCP server, can an agent re-run just the failing checks after a fix, or does it re-run all 125?
@jcbuilds for payments we are only check if payment methods like mpp, x402 are supported and agents are able to understand docs for it
@nicholas_zakas offcourse not. cloudflare has already introduced features that allow bot blocking with exceptions to agents that present valid kya
read more here: https://kyapay.org/overview/what-is-kya
Congrats on the launch!
Just ran it on our website and found an issue with the oauth setup that I'll get to fixing. It did say it couldn't find the docs and API, though, even though they're linked from the main page nav and footer. Is the llms.txt link the only place it looks, or should it be following links in the homepage too?
Thanks! Glad it caught the OAuth thing.
It doesn't only look at llms.txt. It also reads your homepage and checks the links there, so if your nav and footer link to the docs it should have found them.
My first guess is that the nav is rendered with JavaScript. If the links aren't in the HTML the server sends, an agent fetching the page won't see them. It could also be that the docs are hosted on a domain you don't own, like a third-party docs platform, since only docs on your own domain count. For the API part, it also has to get a response from at least one endpoint. If they all need auth, it can still come back as not found.
What's the domain? I'll check what the scanner actually saw, and if it's a bug on our end.