@Sumeyye Bozkus oh, not heard of the age verification scandal. I'll have to look that one up. I mention Discord because I think the founder of Ultravox is active on it.

@Sumeyye Bozkus I had a quick read at the airport. I get it for some things to prevent fraud and identity theft, but not for something like Discord. Operating system users!

I'm doing a fixed price per month for up to x amount of minutes - same price for all clients. Plus some incentives and guarantees and optional add on extras.

Hi @Nir Simionovich and @Eric Klein , First off, I couldn't make the training but to be honest, at the time I wasn't sure I needed OPBX. So, I was wondering if there was a link to a recording or transcript? Basically, I would like to know the following: 1. Does the SaaS version expose the full REST API? — Can we programmatically create orgs, AI assistants, route DIDs from our own dashboard? 2. How does an "AI Assistant" in OPBX connect to external platforms like Retell/Ultravox? — What do we actually configure when creating one? 3. With the SaaS, do we use our own Cloudonix account or is it fully managed? 4. Per-org dedicated agents - does each AI Assistant map 1:1 to a platform agent with isolated concurrency? Thanks. p.s Below is a little solution explainer to a random problem I found this morning using Cloudonix. This got me thinking about Cloudonix as the telephony layer in the same way that we are using Langfuse as our observability layer.

@Nir Simionovich Thanks Nir, got all that down. Question 4 - let me put it in my own words rather than have Claude spit out technical jargon. Let's assume I have the following clients: - Bella Spa gets, - Los Naranjos Golf Club - Best Burger Let's assume they all have inbound agents. Let's assume they all get calls at 10am Bella Spa 10 calls Los Naranjos 15 calls Best Burger 20 calls (they do late breakfasts). If I am using Ultravox - they say no hard cap on concurrency. Retell AI has 20 and LiveKit - not sure yet. But I have agents on all three. From what I can see in the Cloudonix docs, there's a cap at 10 concurrent calls - but this is per account (my account), not per domain - (assuming each domain is each phone number). So if my understanding is correct, any incoming calls above 10 won't get through because of my account limit. And if so, what options exist for higher concurrency? Perhaps I have read the docs wrong? But here I am talking about three organizations but in reality, I'm hoping to get a lot more.

@Abderahil Mazouz Was looking at LiveKit Cloud pricing this week. It seems to be aimed at single tenant enterprises rather than developers building platforms for multiple tenants. At $50 dollars a month you get up to 4 agents. At $500 dollars a month, you'd think you'd get 40 agents. But you only get 5! So, if you need 10, 20, 100 agents for multiple organizations it seems as though self hosting is the only option if you want to build with LiveKit. Of course, you could use agent configs and just have just a few agents pulling in prompts and tools from a database but then you might run into concurrency issues I'd love to hear more about LiveKit Cloud pricing too in case I am missing something obvious.

@John George Going to try that now - inside the booking tools function: Add session.say() filler to create_booking and check_availability. Ok, done it works much better!! Much cleaner. Thanks!! Also added caching for the availability so it's checking event availability - response is almost instant.

@John George ah, good point. Basically, it checks every 10 seconds so if that happens, the worse that can happen is that when the agent goes to make the booking, it might come back as failed due to having already been taken. So, the next available slot is selected. To be honest, I hadn't thought about that. Appreciated.