in the past few months, companies like @ perplexity_ai , @ browsercompany and now @ OpenAI have released there own agentic browsers.
and for them, it’s a natural progression to own the browser layer because now you control the whole OS and experience for the user.
whatever actions that you can do in these browsers are often centred around the read layer, i.e reading, explaining and summarising what’s on your screen, browsers like comet allow you to spin up browser agents to perform (write) actions on your behalf.

@ perplexity_ai comet with reddit
currently, if you want your agentic browser to trade for you the process looks like this:

the process above take takes roughly 10 seconds, out of which only ~2 second is taken to process the swap onchain.
the browser agent continously takes a screenshot of the present view, processes it and then performs actions on it via DOM, and this process becomes slower if the applications interface is complex.
in crypto, where every millisecond matters, you would want your agent to take actions as fast as physics allows. imagine you are unable to buy a token at right time because your agent (in this case browser) was trying to figure out an interface like bloomberg terminal.

a view of McD stats in 2016
any LLM would find it hard to navigate through the above interface and in crypto a lot of protocols and apps have bad interfaces.
agentic browser can’t leverage the most powerful aspect of crypto: the public-nature of data and protocols.
any agent interacting with crypto
it should be a fundamental right of any user to have access to such systems which allow them to create and execute complex financial tasks without worrying about writing scripts, programming or hosting there own infra just via writing prompts in simple english language.
one good prompt is worth more than a thousand line of code.
at SendAI, we are working to provide this fundament right to every user.





