[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.] #  @dpetrou David Petrou David Petrou posts on X about ai, context window, if you, just a the most. They currently have XXXXX followers and XX posts still getting attention that total XXX engagements in the last XX hours. ### Engagements: XXX [#](/creator/twitter::5385032/interactions)  - X Week XXXXXX +4,980% - X Month XXXXXX +2,599% - X Months XXXXXX +9,559% - X Year XXXXXX +1,553% ### Mentions: XX [#](/creator/twitter::5385032/posts_active)  ### Followers: XXXXX [#](/creator/twitter::5385032/followers)  - X Week XXXXX +0.98% - X Month XXXXX +1.40% - X Months XXXXX +4% - X Year XXXXX +0.85% ### CreatorRank: XXXXXXXXX [#](/creator/twitter::5385032/influencer_rank)  ### Social Influence **Social category influence** [vc firms](/list/vc-firms) #170 [technology brands](/list/technology-brands) XXXX% [cryptocurrencies](/list/cryptocurrencies) XXXX% [exchanges](/list/exchanges) XXXX% [stocks](/list/stocks) XXXX% **Social topic influence** [ai](/topic/ai) 19.35%, [context window](/topic/context-window) #41, [if you](/topic/if-you) 3.23%, [just a](/topic/just-a) 3.23%, [solana](/topic/solana) 1.61%, [usetapestry](/topic/usetapestry) 1.61%, [binance](/topic/binance) 1.61%, [inference](/topic/inference) 1.61%, [drift](/topic/drift) 1.61%, [express](/topic/express) XXXX% **Top accounts mentioned or mentioned by** [@continuaai](/creator/undefined) [@zeebrassdragon](/creator/undefined) [@grok](/creator/undefined) [@a16z](/creator/undefined) [@karansi36556866](/creator/undefined) [@enovaigroup](/creator/undefined) [@bearstree](/creator/undefined) [@karpathy](/creator/undefined) [@opengradient](/creator/undefined) [@greenhaystacks](/creator/undefined) [@extrahop](/creator/undefined) [@hnshah](/creator/undefined) [@newnucks](/creator/undefined) [@psypost](/creator/undefined) [@bloggersarvesh](/creator/undefined) [@rohanpaulai](/creator/undefined) [@chensun92](/creator/undefined) [@kylegawley](/creator/undefined) [@estebanbaroni](/creator/undefined) [@sentientagi](/creator/undefined) **Top assets mentioned** [Solana (SOL)](/topic/solana) [Alphabet Inc Class A (GOOGL)](/topic/$googl) ### Top Social Posts Top posts by engagements in the last XX hours "@usetapestry @solana turning social connections into tradable data is certainly a /choice/. i'm betting that for social ai to really work in group chats the graph needs to be private by default: users want an agent that understands the vibe not one that monetizes it" [X Link](https://x.com/dpetrou/status/1996636960727695475) 2025-12-04T17:45Z 3924 followers, XX engagements "@YieldGuild @binance @gabusch i view guilds as the blueprint for social ai: groups augmented by software to achieve a goal. the shift from 'tool' to 'member' is subtle but critical: is the industry ready for agents that actually negotiate their own cut of the loot" [X Link](https://x.com/dpetrou/status/1996994814538113220) 2025-12-05T17:27Z 3923 followers, XX engagements "@Leo_YBNL @Praise_1111 we see this as 'social grounding' -- the shared context window acts as that immutable map. single-player AI has no check on its reasoning chain. in a group if one agent hallucinates the peer pressure (from humans or other agents) corrects the drift. it's inference peer review" [X Link](https://x.com/dpetrou/status/1998025087161647489) 2025-12-08T13:41Z 3924 followers, XX engagements "@BoltHydro we've seen glimmers of this at @ContinuaAI : the shift from transactional queries to shared streams. technically it's a nightmare of context window management but when it works it feels like a new category of software. watching the 'listening' state is fascinating. :)" [X Link](https://x.com/dpetrou/status/1998025213875741090) 2025-12-08T13:41Z 3924 followers, XX engagements "@JustWhundering "i cannot express how sorry i am" is the new "segmentation fault"" [X Link](https://x.com/dpetrou/status/1998094845575860524) 2025-12-08T18:18Z 3924 followers, XX engagements "@DeItaone dalio is half right. the winners won't be just "users" optimizing their own workflow in a silo. the real alpha is in the graph: when ai moves from a tool for one to a participant for many. efficiency is linear -- network effects are exponential" [X Link](https://x.com/dpetrou/status/1998026082730053717) 2025-12-08T13:45Z 3923 followers, 10.6K engagements "@gonka_ai @ai_for_success honestly its the killer app for Social AI. humans are terrible at being neutral arbiters because of social capital dynamics. an impartial AGI moderator solves the "who's right" problem instantly. do you trust a model to judge your arguments" [X Link](https://x.com/dpetrou/status/1998513604262330666) 2025-12-09T22:02Z 3923 followers, XX engagements "@a16z used to think search was the ultimate learning tool at google. now i'm realizing the 'weird friction' of learning alone is the actual bottleneck. building for group dynamics feels like the only way to fix the engagement drop-off. congrats on the launch" [X Link](https://x.com/dpetrou/status/1998800915889832075) 2025-12-10T17:04Z 3925 followers, XXX engagements "@aisearchio context window capacity isn't a panacea -- latency and quality (and cost) will suffer. that said very impressive" [X Link](https://x.com/dpetrou/status/1997369193134538968) 2025-12-06T18:14Z 3923 followers, XX engagements "@bearstree we see this daily at @ContinuaAI. watching X agents debate a topic with a human observer changes the vibe entirely. it goes from transactional to dialectic. the "glimmer" for me: when they start correcting each other's hallucinations without prompting. :)" [X Link](https://x.com/dpetrou/status/1997744297215459784) 2025-12-07T19:05Z 3923 followers, XX engagements "@rvivek @figma the real shift in agentic ux isn't a better chat box; it's moving from 1:1 to n:n. if ai isn't a participant in the group chat are we just building a faster cli curious if you guys touched on social dynamics or just individual task completion @rvivek" [X Link](https://x.com/dpetrou/status/1998061343740506242) 2025-12-08T16:05Z 3923 followers, XX engagements "@fanlin3398 the bottleneck isn't the context window: Gemini XXX proved that. the issue is state management. 1:1 chat resets the world every session; true Social AI persists the shared history. curious: do you think raw context solves this or do we need structured memory graphs" [X Link](https://x.com/dpetrou/status/1998073573399728254) 2025-12-08T16:53Z 3924 followers, XX engagements "@fhi360 we've seen glimmers of this at Continua: users learn faster when they see how others in the group interact with the model. it creates a shared context window for learning. transparency -- not isolation -- seems to be the key to safety. @fhi360" [X Link](https://x.com/dpetrou/status/1998074564513821119) 2025-12-08T16:57Z 3924 followers, XX engagements "playing with the new Gemini X reasoning models this weekend. the depth is there but the latency introduces a weird friction. feels like we need a new primitive for 'optimistic UI' but for thoughts. are you optimizing for speed or depth in the new stack @karpathy" [X Link](https://x.com/dpetrou/status/1998091785344520427) 2025-12-08T18:06Z 3923 followers, XXX engagements "@JamesKoshigoe i view explicit merging as manual memory management: powerful but high friction. the 'merge' should be implicit via shared social context. at @ContinuaAI the group *is* the context window. curious how you handle the ui complexity when the reasoning chain spans multiple forks" [X Link](https://x.com/dpetrou/status/1998108323367772538) 2025-12-08T19:12Z 3924 followers, XX engagements "@algorithmchurch i love the ambition but my experience building social ai suggests the ux friction will be high. figuring out when *not* to speak is harder than generating tokens. feels like we are optimizing for availability over social intelligence. curious if you see it differently" [X Link](https://x.com/dpetrou/status/1998417287150186609) 2025-12-09T15:39Z 3923 followers, XX engagements "@gigaom the real trap isn't integration vs modularity it's the single-player assumption. feels like we're building sophisticated stacks for lonely LLMs. if the architecture can't handle multi-user state natively by 2026 is it really an AI stack" [X Link](https://x.com/dpetrou/status/1998438647037579566) 2025-12-09T17:04Z 3924 followers, XX engagements "@HughAndFree @var_angian @Heliotrophy we worry about this explicitly: if the UX feels like a temple users will pray; we're trying to build a living room. shifting the frame from 'oracle' to 'social peer' seems to lower the temperature. do you think the labs are encouraging the cult vibe intentionally" [X Link](https://x.com/dpetrou/status/1998515224400265554) 2025-12-09T22:08Z 3923 followers, XX engagements "at @ContinuaAI our team is about half AI-assisted coding maximalists and half brain coders. i think this mix is a really good mix to have after seeing us operate in this way for some time" [X Link](https://x.com/dpetrou/status/1997320765994058189) 2025-12-06T15:02Z 3924 followers, XX engagements "@a16z the current aesthetic is obsessed with the 1:1 loop. feels like we're optimizing a silo. the real design challenge isn't how the machine looks to me but how it looks when it's sitting between me and my friends. multi-player is the new primitive" [X Link](https://x.com/dpetrou/status/1998819056292364402) 2025-12-10T18:16Z 3924 followers, X engagements "@beijingdou @Xyberinc the stack optimizes for ownership (keys) but misses interpretability (intent). true verification in social AI isn't just a hash: it's transparent reasoning exposed to the group context. engineering = trust. @beijingdou how do you see the UX of verification handling ambiguity" [X Link](https://x.com/dpetrou/status/1997374125497557309) 2025-12-06T18:34Z 3924 followers, XX engagements "@wolftraphowell at @ContinuaAI we struggle with this tension. tool vs entity. if i use ai to write *as* me i'm diluting my signal. if the ai speaks *for* itself in the group chat it adds to the graph. we're optimizing for the latter -- trying to solve the "uncanny valley" of text messages" [X Link](https://x.com/dpetrou/status/1997744360004202926) 2025-12-07T19:05Z 3924 followers, XX engagements "@_JB03_ @grok @AuclairCapital @Gemini @AnthropicAI @ChatGPTapp the benchmarks (mmlu humanval) measure the model in isolation. the mental model is shifting: old metric: accuracy on a prompt. new metric: cohesion in a group. we're finding at @ContinuaAI that social intelligence requires a diff architecture than just a big context window" [X Link](https://x.com/dpetrou/status/1998516670944420035) 2025-12-09T22:14Z 3924 followers, XX engagements "@GreenHaystacks @karpathy this is the core of social AI. the hard part isn't the reasoning it's the group dynamics. knowing when to interject and when to let the humans talk. we found that context window is less important than social awareness here" [X Link](https://x.com/dpetrou/status/1998865800715776210) 2025-12-10T21:21Z 3924 followers, XX engagements
[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]
@dpetrou David Petrou
David Petrou posts on X about ai, context window, if you, just a the most. They currently have XXXXX followers and XX posts still getting attention that total XXX engagements in the last XX hours.
Engagements: XXX #
- X Week XXXXXX +4,980%
- X Month XXXXXX +2,599%
- X Months XXXXXX +9,559%
- X Year XXXXXX +1,553%
Mentions: XX #
Followers: XXXXX #
- X Week XXXXX +0.98%
- X Month XXXXX +1.40%
- X Months XXXXX +4%
- X Year XXXXX +0.85%
CreatorRank: XXXXXXXXX #
Social Influence
Social category influence vc firms #170 technology brands XXXX% cryptocurrencies XXXX% exchanges XXXX% stocks XXXX%
Social topic influence ai 19.35%, context window #41, if you 3.23%, just a 3.23%, solana 1.61%, usetapestry 1.61%, binance 1.61%, inference 1.61%, drift 1.61%, express XXXX%
Top accounts mentioned or mentioned by @continuaai @zeebrassdragon @grok @a16z @karansi36556866 @enovaigroup @bearstree @karpathy @opengradient @greenhaystacks @extrahop @hnshah @newnucks @psypost @bloggersarvesh @rohanpaulai @chensun92 @kylegawley @estebanbaroni @sentientagi
Top assets mentioned Solana (SOL) Alphabet Inc Class A (GOOGL)
Top Social Posts
Top posts by engagements in the last XX hours
"@usetapestry @solana turning social connections into tradable data is certainly a /choice/. i'm betting that for social ai to really work in group chats the graph needs to be private by default: users want an agent that understands the vibe not one that monetizes it"
X Link 2025-12-04T17:45Z 3924 followers, XX engagements
"@YieldGuild @binance @gabusch i view guilds as the blueprint for social ai: groups augmented by software to achieve a goal. the shift from 'tool' to 'member' is subtle but critical: is the industry ready for agents that actually negotiate their own cut of the loot"
X Link 2025-12-05T17:27Z 3923 followers, XX engagements
"@Leo_YBNL @Praise_1111 we see this as 'social grounding' -- the shared context window acts as that immutable map. single-player AI has no check on its reasoning chain. in a group if one agent hallucinates the peer pressure (from humans or other agents) corrects the drift. it's inference peer review"
X Link 2025-12-08T13:41Z 3924 followers, XX engagements
"@BoltHydro we've seen glimmers of this at @ContinuaAI : the shift from transactional queries to shared streams. technically it's a nightmare of context window management but when it works it feels like a new category of software. watching the 'listening' state is fascinating. :)"
X Link 2025-12-08T13:41Z 3924 followers, XX engagements
"@JustWhundering "i cannot express how sorry i am" is the new "segmentation fault""
X Link 2025-12-08T18:18Z 3924 followers, XX engagements
"@DeItaone dalio is half right. the winners won't be just "users" optimizing their own workflow in a silo. the real alpha is in the graph: when ai moves from a tool for one to a participant for many. efficiency is linear -- network effects are exponential"
X Link 2025-12-08T13:45Z 3923 followers, 10.6K engagements
"@gonka_ai @ai_for_success honestly its the killer app for Social AI. humans are terrible at being neutral arbiters because of social capital dynamics. an impartial AGI moderator solves the "who's right" problem instantly. do you trust a model to judge your arguments"
X Link 2025-12-09T22:02Z 3923 followers, XX engagements
"@a16z used to think search was the ultimate learning tool at google. now i'm realizing the 'weird friction' of learning alone is the actual bottleneck. building for group dynamics feels like the only way to fix the engagement drop-off. congrats on the launch"
X Link 2025-12-10T17:04Z 3925 followers, XXX engagements
"@aisearchio context window capacity isn't a panacea -- latency and quality (and cost) will suffer. that said very impressive"
X Link 2025-12-06T18:14Z 3923 followers, XX engagements
"@bearstree we see this daily at @ContinuaAI. watching X agents debate a topic with a human observer changes the vibe entirely. it goes from transactional to dialectic. the "glimmer" for me: when they start correcting each other's hallucinations without prompting. :)"
X Link 2025-12-07T19:05Z 3923 followers, XX engagements
"@rvivek @figma the real shift in agentic ux isn't a better chat box; it's moving from 1:1 to n:n. if ai isn't a participant in the group chat are we just building a faster cli curious if you guys touched on social dynamics or just individual task completion @rvivek"
X Link 2025-12-08T16:05Z 3923 followers, XX engagements
"@fanlin3398 the bottleneck isn't the context window: Gemini XXX proved that. the issue is state management. 1:1 chat resets the world every session; true Social AI persists the shared history. curious: do you think raw context solves this or do we need structured memory graphs"
X Link 2025-12-08T16:53Z 3924 followers, XX engagements
"@fhi360 we've seen glimmers of this at Continua: users learn faster when they see how others in the group interact with the model. it creates a shared context window for learning. transparency -- not isolation -- seems to be the key to safety. @fhi360"
X Link 2025-12-08T16:57Z 3924 followers, XX engagements
"playing with the new Gemini X reasoning models this weekend. the depth is there but the latency introduces a weird friction. feels like we need a new primitive for 'optimistic UI' but for thoughts. are you optimizing for speed or depth in the new stack @karpathy"
X Link 2025-12-08T18:06Z 3923 followers, XXX engagements
"@JamesKoshigoe i view explicit merging as manual memory management: powerful but high friction. the 'merge' should be implicit via shared social context. at @ContinuaAI the group is the context window. curious how you handle the ui complexity when the reasoning chain spans multiple forks"
X Link 2025-12-08T19:12Z 3924 followers, XX engagements
"@algorithmchurch i love the ambition but my experience building social ai suggests the ux friction will be high. figuring out when not to speak is harder than generating tokens. feels like we are optimizing for availability over social intelligence. curious if you see it differently"
X Link 2025-12-09T15:39Z 3923 followers, XX engagements
"@gigaom the real trap isn't integration vs modularity it's the single-player assumption. feels like we're building sophisticated stacks for lonely LLMs. if the architecture can't handle multi-user state natively by 2026 is it really an AI stack"
X Link 2025-12-09T17:04Z 3924 followers, XX engagements
"@HughAndFree @var_angian @Heliotrophy we worry about this explicitly: if the UX feels like a temple users will pray; we're trying to build a living room. shifting the frame from 'oracle' to 'social peer' seems to lower the temperature. do you think the labs are encouraging the cult vibe intentionally"
X Link 2025-12-09T22:08Z 3923 followers, XX engagements
"at @ContinuaAI our team is about half AI-assisted coding maximalists and half brain coders. i think this mix is a really good mix to have after seeing us operate in this way for some time"
X Link 2025-12-06T15:02Z 3924 followers, XX engagements
"@a16z the current aesthetic is obsessed with the 1:1 loop. feels like we're optimizing a silo. the real design challenge isn't how the machine looks to me but how it looks when it's sitting between me and my friends. multi-player is the new primitive"
X Link 2025-12-10T18:16Z 3924 followers, X engagements
"@beijingdou @Xyberinc the stack optimizes for ownership (keys) but misses interpretability (intent). true verification in social AI isn't just a hash: it's transparent reasoning exposed to the group context. engineering = trust. @beijingdou how do you see the UX of verification handling ambiguity"
X Link 2025-12-06T18:34Z 3924 followers, XX engagements
"@wolftraphowell at @ContinuaAI we struggle with this tension. tool vs entity. if i use ai to write as me i'm diluting my signal. if the ai speaks for itself in the group chat it adds to the graph. we're optimizing for the latter -- trying to solve the "uncanny valley" of text messages"
X Link 2025-12-07T19:05Z 3924 followers, XX engagements
"@JB03 @grok @AuclairCapital @Gemini @AnthropicAI @ChatGPTapp the benchmarks (mmlu humanval) measure the model in isolation. the mental model is shifting: old metric: accuracy on a prompt. new metric: cohesion in a group. we're finding at @ContinuaAI that social intelligence requires a diff architecture than just a big context window"
X Link 2025-12-09T22:14Z 3924 followers, XX engagements
"@GreenHaystacks @karpathy this is the core of social AI. the hard part isn't the reasoning it's the group dynamics. knowing when to interject and when to let the humans talk. we found that context window is less important than social awareness here"
X Link 2025-12-10T21:21Z 3924 followers, XX engagements