High level Inter-agent protocols


Hello Sovrin,

I’d like to know where is Sovrin standing with regard to agent interaction. Which standards/protocols are there being defined for such communication to be made in a meaningful way?.. I mean, I know there are certain protocols being considered/implemented already at the session/presentation layers (Sovrin TLS, for example) for securing communication, but are there any other protocols being designed or in consideration at the application layer? (the application being in this case the agents themselves)… In which ways can we structure the interaction flow between agents, given that the communication channel is already secured. Does it make sense to establish such high level agent interaction protocols?

Is my question clear enough?


Hello Carlos,

One of the protocols for inter-agent communication and data exchange will be XDI, there are several projects now underway based on Sovrin (as the lower-level ledger and trust layer) and XDI (as the agent-to-agent protocol). This will then enable various flows and use cases such as the ones that are shown here: http://xdi2.org/demos.html

Also, on the last Sovrin Agent WG call, we learned that Evernym is designing an “extension” architecture that will allow agents to speak multiple protocols and expose a diverse set of APIs. So there won’t be just a single protocol that agents will use to communicate. But XDI will be one of them.


Hey Markus,

Yes, I was guessing XDI would be surely involved in this matter.

This “extension architecture” idea sounds awesome. Is there any public info available about that?


Carlos, I for one certainly believe there is a great case for those higher-level agent-to-agent interaction protocols. In full disclosure, it would be hard for me NOT to as co-chair of the OASIS XDI Technical Committee—we’ve been developing a semantic data interchange protocol for this very purpose since 2004.

But I agree with the Agent WG chaired by Nathan Georger ( @nage) who feel XDI is overkill for many much simpler and lighter-weight interactions. So the Agent WG is working on lighter protocols that would still be considered application-level protocols, especially in the domain of specific APIs.

So I expect this to be a very active topic of discussion among the Agent WG for much of 2017.


Awesome, Drummond. I’m intuiting that most interesting things will come out of discussions around this topic.

The way I see it, it’s in this layer that complex and innovative distributed schemes can be implemented around identity. Looks like the meaningful core of Sovrin if I’m not mistaken, where “intelligence” arises.

I’m eagerly looking forward to get into these protocols, even such simple ones.