Timing for Release of 2.0 and Alternatives for Mid-Call Actions | Thoughtly Community

Timing for Release of 2.0 and Alternatives for Mid-Call Actions

Any idea when 2.0 is coming Torrey Leonard? Days, weeks? I'm debating whether I should build a complex set of agents and use call transfer webhooks (as a substitute for mid-calls actions), or I should just wait for 2.0. Ideally, 2.0 is released before I hack together that flow, but if we're not days away from the release, I guess I'd have no choice (reluctantly) 😅 (1.8 is marginally better than 1.5, but 1.0 is still far ahead of both of them. It works great. I just need the mid call actions to access data on-demand, but obviously they're not available for 1.0)

9 comments

Torrey Leonard
·
·
We're still working on collecting feedback! What makes you like 1.0 better than 1.8?
Chris Guzman
·
·
I'm still liking 1.0 after 1.8 roll out. One feedback I could mention is when i train the AI to optimize their tone in prompt.. rather than changing tone, it adds additional words in the script (which is not what I'm really trying to do)
Sebastien R.
·
·
V1 's flow was a bit better on the following points:
You could do a loop between outputs and previous say steps. now you no longer can.
Special characters are no longer allowed in the instructions
Outcome must now he 10 characters long
✅1
Sebastien R.
·
·
Btw, 1.8 sometimes is bugging by reading the instructions out loud, instead of interpreting it. Torrey Leonard
✅1
Shay R
·
·
1.0 just seems smarter, more improvisational, and understands variables/information better. 1.8 just seems dumb, almost like you have to hold their hand for everything. For example, I'm setting up an appointment booking agent and 1.0 works better matching appointment requests with availability, whereas 1.8 doesn't understand how to contextualize availability and spits it out in a very informal way. The BIG problem: 1.0 relies on old availability since no mid-call actions exist Torrey Leonard
Shay R
·
·
Sadly, it looks like 2.0 is further away than I had hoped, so I'll have to figure out a complex agent routing solution (to simulate mid call webhooks) 🥲
Alex Casella (Thoughtly)
·
·
Hey Shay R can you send us links to v1.8 calls that didn't go as expected?
Alex Casella (Thoughtly)
·
·
Sebastien R. do you have verbatim enabled on the node? If yes then it will read out the instructions as designed
Shay R
·
·
Alex Casella (Thoughtly) I'll retest 1.8 and 1.0 with the same flow (the new agent builder checked/unchecked) and the same responses and DM you the links to show you the difference