This was an fantastic event. Lots of insights and learnings whilst connecting with like minded people who are each leading their own amazing product brands!
Last night, Third Wave hosted a dinner at Gauthier Soho with Sam Brooks from Highland Europe and founders of DTC product brands.
We explored what makes a brand worth backing, from the team and customer loyalty to the realities of scaling.
What Makes a Brand Worth Backing?
Join Sam Brooks, General Partner at Highland Europe, for a Third Wave dinner for DTC product brand founders and CEOs. Explore what gives an investor confidence in a brand, and what founders should look for in return.
We’re opening a research preview of Claude Duet, a new app where Claude works as a proactive agent.
It comes to you with what needs your attention, completes tasks you ask of it, keeps working when you step away, and checks with you before it acts on your behalf.
@thsottiaux Does this apply to batched calls as well? I'd guess that batched traffic through the API is one of the dream types of loads to manage yield from capacity, but the economics of it needs to work.
This has been a frustration of mine for a couple of years now. I guess the root issue is that the models aren't self aware in regards of their own capabilities and judgement is trained on historic velocity (mostly human performance). Having said that, OpenAI could guide the models through the system prompt.
@thsottiaux A really useful thing would be letting us see how our existing workflows would run with the proposed 5hr limit to smooth demand. I voted for no 5h limits, even if it means not getting twice the usage on being worried that overnight tasks getting throttled would badly impact.
@aliasaria As a founder, I've grappled with many of these thoughts and questions. The most painful aspect is giving up when capital constrained for fear of a better capitalised competitor getting there first or scaling faster. Now, I'm very much focused on getting things shipped instead.
@Codexresets_ This just doesn't make sense! Sol 6.1 Fast on API at 32 TPS is slower than 6.1 Sol standard on API. Are you sure your tests have enough samples to give meaningful insight?
Here's one no one is talking about... Maybe @OpenAI and @thsottiaux are actually trying to get rid of their old Pro 200 subscribers as they make far more profit from the same capacity going to API users....
@hany99dev Not in the slightest. I'm becoming less convinced that the limits haven't been lowered for existing Pro 200 subscribers already. The left usage is going slower for sure, but so is overall velocity of tasks getting done.
The OpenAI situation keeps getting worse.
Yesterday Tibo announced a global reset for all paid ChatGPT accounts at 10am PST today. It's been over 2 hours. No reset.
Builders literally scheduled their whole day around this.
First they cut usage in half.
Now they promise a reset that doesn't show up.
You can't keep doing this to the people paying you.
@_architected I'm glad that it's not just me not seeing the reset. I totally agree that they shouldn't name a time and encourage everyone to burn their tokens and then fail to deliver.
@MatthewBerman Have a look at your token usage under your profile. I'm finding that the reason the usage isn't draining as fast is simply because there's far less tokens being used than before 6.1. It's proving to be very capable but incredibly slow 🙃
@thsottiaux - Got home all excited to use 6.1 Sol and can't see it on the model list in Codex. Is this a staggered roll out and if so, when will it be available in the UK?