Always stuck with Github Copilot since the beginning. It gradually brings all the latest features in the ecosystem and the release notes and youtube channel are also easy to follow to track the latest advancements.
honestly, doom scrolling + surrounding myself with people intensely interested in the topic. I basically just info overload myself on an hourly basis and since I find it interesting enough it tends to make its way into the sticky part of my memory.
to give you an idea of what I mean, I'm checking bluesky, reddot, HN, youtube, instagram, tiktok, discord basically every hour. every morning, lunch, and evening I'm checking my email where I receive about a dozen or so newsletters either weekly or daily. a few times a week (usually weekends) I'm browsing github and arxiv for the whatever random thought or interest is the flavor of the week (lately its been on EOE techniques and classifiers). This has lead me to collect the names of a good number of authors I trust and that appear to be thought leaders in the space.
I don't think this approach is good at all, in fact, it feels neurotic and I think its an unhealthy obsession borne from my adhd, but its kept me going for the last 3ish years and I don't feel the slightest bit burnt out. personally I find the whole field of AI/ML to be immensely exciting so I think that part of it is a huge driver for how I personally stay up-to-date.
ideally I would offload a lot of these activities to a research agent and tbh I "plan" to eventually, same way I plan to gut short form videos and social media from my life and get 8hrs of sleep. Eventually.
it sounds like we share a vice for information gluttony. i have a similar workflow where i continuously scan the digital horizon for useful ideas and continuously feed them into my personal toolkit. i've kind of always been like that as a person and an engineer (the two aren't really separable for me), and claude code has been like switching from coffee to meth, and, um, there's some downsides to doing meth. it's hard not to build a house of cards that you end up abandoning when it starts feeling wobbly and you see a greenfield next door
I'm always so hesitant to do that, I really value getting news within the first 3 hours of it becoming available so doing weekly digests makes me feel like I'm losing something valuable, but I guess that could be solved with a personalized feed.. hmmm
i don't even try. the good stuff will be absorbed by the corps and harness providers (claude code, etc). the less useful stuff will fall away naturally. current harnesses (claude code) suit my needs very well.
i haven't yet seen tooling that makes me think, "wow, this is the missing link, i need this" -- not since claude code started working very well with sub-agents.
Cloud agents + workflows in the future. Escaping single player mode on a machine. Cursor has background agents. I build Ito.ai to do code reviews that actually run code. There's many more but the future is escaping your local machine.
as an example, i built tooling that lets me open claude code and say 'add MSP430FR2476TRHBR to the library' and a few minutes later there's a schematic symbol and footprint in my kicad library with parameter fields consistent with my organization's internal standards. i'll be waiting a while for anthropic to add that feature to claude code.
Always stuck with Github Copilot since the beginning. It gradually brings all the latest features in the ecosystem and the release notes and youtube channel are also easy to follow to track the latest advancements.
https://news.smol.ai/
Twitter - discord - lobsters - HN.
These 4 really.
honestly, doom scrolling + surrounding myself with people intensely interested in the topic. I basically just info overload myself on an hourly basis and since I find it interesting enough it tends to make its way into the sticky part of my memory.
to give you an idea of what I mean, I'm checking bluesky, reddot, HN, youtube, instagram, tiktok, discord basically every hour. every morning, lunch, and evening I'm checking my email where I receive about a dozen or so newsletters either weekly or daily. a few times a week (usually weekends) I'm browsing github and arxiv for the whatever random thought or interest is the flavor of the week (lately its been on EOE techniques and classifiers). This has lead me to collect the names of a good number of authors I trust and that appear to be thought leaders in the space.
I don't think this approach is good at all, in fact, it feels neurotic and I think its an unhealthy obsession borne from my adhd, but its kept me going for the last 3ish years and I don't feel the slightest bit burnt out. personally I find the whole field of AI/ML to be immensely exciting so I think that part of it is a huge driver for how I personally stay up-to-date.
ideally I would offload a lot of these activities to a research agent and tbh I "plan" to eventually, same way I plan to gut short form videos and social media from my life and get 8hrs of sleep. Eventually.
it sounds like we share a vice for information gluttony. i have a similar workflow where i continuously scan the digital horizon for useful ideas and continuously feed them into my personal toolkit. i've kind of always been like that as a person and an engineer (the two aren't really separable for me), and claude code has been like switching from coffee to meth, and, um, there's some downsides to doing meth. it's hard not to build a house of cards that you end up abandoning when it starts feeling wobbly and you see a greenfield next door
an agent that runs weekly to summarize news in this domain would actually be a really healthy way to stay up to date without constant attention.
I'm always so hesitant to do that, I really value getting news within the first 3 hours of it becoming available so doing weekly digests makes me feel like I'm losing something valuable, but I guess that could be solved with a personalized feed.. hmmm
i don't even try. the good stuff will be absorbed by the corps and harness providers (claude code, etc). the less useful stuff will fall away naturally. current harnesses (claude code) suit my needs very well.
i haven't yet seen tooling that makes me think, "wow, this is the missing link, i need this" -- not since claude code started working very well with sub-agents.
Cloud agents + workflows in the future. Escaping single player mode on a machine. Cursor has background agents. I build Ito.ai to do code reviews that actually run code. There's many more but the future is escaping your local machine.
no one knows what that link looks like but you
as an example, i built tooling that lets me open claude code and say 'add MSP430FR2476TRHBR to the library' and a few minutes later there's a schematic symbol and footprint in my kicad library with parameter fields consistent with my organization's internal standards. i'll be waiting a while for anthropic to add that feature to claude code.
sure, I program little routines or skills all the time. but for general-purpose stuff, claude code takes care of me.