Matthew

lead AI initiatives at consulting firm. interested in why things are the way they are and which things could instead be other better things 🇬🇧

[Matthew @templative]

@Teddarific ‘Align middle’ and ‘align center’ always throw me for such a loop. I’ve concluded I’m better off not thinking about it and instead just trust my fingers to execute the shortcut I need.

[Matthew @templative]

@andrewchen Works when there is dominant verb for which the brand steps in - you only use Google to search => when people say ‘to google’ you know it means ‘to search for’. In the primitive chatbot era it wasn’t uncommon to hear ‘just ChatGPT it’ but now you can ask, build, create, etc.

[Matthew @templative]

@trq212 @travelingmunchi Hey Thariq - any timeline on native Windows sandbox support?

[Matthew @templative]

@businessbarista Claude

[Matthew @templative]

@nikitabier the share functionality differs when clicking link in bio vs link in post. When I find interesting company I can quickly hit the link in bio and share to a tracker. When I find an interesting article I want to save the workflow is more cumbersome - open in browser, share from there. It would be great if there was share option on links in post that shared the link (vs just share the post)

[Matthew @templative]

@travisk @tbpn I guess sharper question would be whether it’s price-maximizing to have multiple tip components? Tip to restaurant workers, tip to delivery person…

[Matthew @templative]

@travisk @tbpn Let’s assume you are right that two-component pricing is profit-maximizing vs one-component. Do we have enough evidence and/or confidence to say we should stop there? Could three be better? Or does the complementarity of fixed plus variable point to two as the optimal solution?

[Matthew @templative]

@BlasMoros @wabi did invites go out?

[Matthew @templative]

@BlasMoros @wabi Regularly input all my thoughts as short notes that form the context. Request brief on subject X created on-demand pulling from all that context

[Matthew @templative]

@rohanjamin typo in the red error text after code entry

[Matthew @templative]

@simonw Could it be simplified to ‘an LLM agent runs tools to achieve a goal’? If you are clear on what it means to achieve a goal then ‘in a loop’ feels superfluous and definition of agent dependent on complexity of problem / level of iteration required feels shaky

[Matthew @templative]

@tyler_agg memory

[Matthew @templative]

@devdigest @nickscamara_ @firecrawl 🔥

[Matthew @templative]

@jerryjliu0 Is it better to codify the workflow as series of outcomes or series of activities? Benefit of former is it opens aperture to avoid inefficiencies of current process but I guess risk that sharing the outcome alone is not always sufficient for agent to work out the right activities

[Matthew @templative]

@arakharazian @tryramp Is it right to extrapolate from ‘companies in sector X using Ramp’ to ‘all companies in sector X’ and how does that answer vary by sector? I could easily a common factor driving both Ramp usage and spend on AI…

[Matthew @templative]

@michael_reid11 You’re forgetting Stevie G

[Matthew @templative]

@ExaAILabs Is there a home for all these apps?

[Matthew @templative]

@nicochristie interested

[Matthew @templative]

@nicochristie Interested

[Matthew @templative]

@tyler_agg Content