Send call recordings to Graph

[tomastiminskas]

Requirements

• When a call is started in a tribe that has a SecondBrain url set, we want to start recording call automatically on call launch and send call recorded file to Graph automatically

Context

• Recording call files are uploaded to S3 bucket where we can set up a lamba to send files to workflow from there.
• S3 receives a file where the name is the room id of the call
• Right now standard calls have the following format: https://chat.sphinx.chat/rooms/sphinx.call.765295315.760382 where sphinx.call.765295315.760382 is the room id. So the file uploaded to S3 will be sphinx.call.765295315.760382.mp3
• On a call with G we discuss about including some reference to the workflow on the room id, so the lamba in S3 will search for workflow reference in the room id and if present it will send it to it

Open Questions

• Ideally we should include the workflow_id in the room id of the call when starting it from the app having a url like this: https://chat.sphinx.chat/rooms/sphinx.call.w.1234.765295315.760382 where 1234 will be the wokflow id. This way S3 lamba will just checked the number after w. if present in the file name and send it to the workflow with that id
• The app needs a way to get the workflow_id of a graph based on its url. For example on repo2Graph tribe, the SB url is nav.swarm38.sphinx.chat. @tomsmith8 do you have an endpoint or can we create one to get a workflow_id from that url?

[tomsmith8]

What do you mean by workflow_id. Wouldn't all call recording call the same workflow_id no matter which graph they came from i.e

• repo2graph, graphmindset, machinelearning tribes
• Each start a call and its recorded
• Each get pushed to the same workflow_id (e.g. Analyze call recording using Whinx)

Also, are we saying that each tribe that has a SB should support storing the calls in there or to a single company graph?

[tomastiminskas]

@tomsmith8 the goal (based on my discussion with Paul) is to send each tribe call recordings to the tribe specific graph. So I thought each graph would have a different workflow id. But I might be wrong about that.

If the workflow will be always the same, then how can we let the workflow know in which graph it needs to be inserted?

[tomsmith8]

ok sweet makes sense. Let's add a toggle to the tribe to enable/disable call recordings for a tribe I guess then.

• One workflow_id to generate call recordings analysis
• When calling this workflow you'll pass params like the s3 link etc. in set_var request to stakwork
• Included in that should be 2b_base_url and secret

Example would be:

{
"2b_base_url": "https://boltwall.swarm38.sphinx.chat/api",
"secret": "{{SWARM_38_API_KEY}}",
// s3 link and any other data needed
}

How we get that data:

• Store in the tribe details?

[tomastiminskas]

Then I guess we don't need to store anything in the app.

If worflow_id and secret will be always the same, then that can be stored in S3 as it's right now.

The only thing that the app will need to send is the 2b_base_url. The app just hits the endpoint to record the call and the file name will be built based on room_id so I guess we need to define a format of the room_id to include the 2b_base_url

Something like:

https://chat.sphinx.chat/rooms/sphinx.call.-nav.swarm38.sphinx.chat-.765295315.760382

So file will be create with name sphinx.call.-nav.swarm38.sphinx.chat-.765295315.760382.mp3

Then lamba in S3 will check if there's a string between - in file name and in that case use that value as 2b_base_url param.

Does it sound good?

[tomsmith8]

secret will be different depending on the where it comes from.

also, 2b_base_url e.g https://boltwall.swarm38.sphinx.chat/api is different to the frontend e.g nav.swarm38.sphinx.chat but its the same graph.

Stakwork workflow will need to be aware of the api endpoint https://boltwall.swarm38.sphinx.chat/api AND secret