I've read on how one could have his ajax content crawl-able by Google bot using the #! (hashbang?) protocol which delivers HTMl to the bot.
Yet, I couldn't figure the point of having ajax indexed. IMHO the only relevant indexing one might want is HTML (container page) + AJAX as a whole.
If http://someexample.com#!value=top-sell outputs
{ "value" : "carrot" }
and google index it, what good it is to the user?
It has no meaning outside of a HTML page which could looks like this:
[...]
<script src="item-handler.js" />
</head>
<body>
Our store best selling item is <output id="top-sell"></output>!
</body>
</html>
Here's how I picture the every day use case of #!:
User: Searches "carrot" on Google
Google: Display http://someexample.com#!value=top-sell in the result
User: Clicks the link
Server/Browser: Renders
{ "value" : "carrot" }
User: "What kind of gibberish is that?!"
User: Sends a email to web-master: "You suck!"
And I definitively don't want to do the latter...