one thing that makes agents more complex is that they generally add capability over time which should also be tracked as part of the state
@yoheinakajima
-
Seeking elegant primitives for implementing stateful AI agents
By
–
stateful agents, decision traces, context graphs… talked about a lot, but has anyone seen an elegant primitive around how to actually implement?
-
User experience comparing Claude and ChatGPT workflows
By
–
I’m increasingly using Claude (prob 50/50 now) but ChatGPT still knows me better as I’ve been using it for longer. I did try the move over info prompt which helped but not quite there yet.
-
Email-based mission control prototype demo
By
–
okay this email based mission control i prototypes last night is pretty cool: > starts with email
> grab and summarize content about participants from crm/notes
> auto bucket emails into workstreams using context about participants
> auto bucket emails into specific project -
AGI: Anticipation or Anger Depending on the Language
By
–
In this embedding analysis, compared to other futuristic technology terms, AGI is more strongly associated with anticipation than fear, particularly in English, Japanese, Chinese, and Italian. It also exhibits stronger negative connotations in several European languages, such as anger in Spanish and Italian.
-
Comparaison des émotions liées aux tech futuristes en 12 langues
By
–
now i compared similarity of 12 futuristic tech terms with 12 emotions across 12 languages and then normalized them you can see charts like how does "AGI" feel across languages* *this is in relation to the other 11 terms
-
Traces d’agents auto-améliorants pour modèles
By
–
you talking about how traces of self improving agents can be fed into the retraining of self improving models?
-
Color-emotion similarity changes across languages
By
–
i compared embedding similarity between colors and emotions and normalized it to create this chart and then when i changed the language, the results changed (eg. 赤、愛) see: english, spanish, japanese, hindi
-
AI agent one-shots video with multi-tool workflow
By
–
just tried this out and it one-shotted* this video: "before the agent does anything"
— Yohei (@yoheinakajima) 13 mai 2026
*i generated the narrative using chatgpt and used that as a prompt. featuring: @e2b @runanywhereai @composio @mem0ai @firecrawl @browser_use @agentmail @covenantlabsai
some thoughts:
– i… https://t.co/X3RgWs9MOk pic.twitter.com/xjmXZF11HPjust tried this out and it one-shotted* this video: "before the agent does anything" *i generated the narrative using chatgpt and used that as a prompt. featuring: @e2b @runanywhereai @composio @mem0ai @firecrawl @browser_use @agentmail @covenantlabsai some thoughts:
– i -
Outils déterministes pour plus de sécurité des agents
By
–
there are many agent use cases locked behind "what if" fears because after we give a tool to an agent, we're relying on the prompt to limit behavior tools like @denieddotdev allow teams to manage capability as a separate deterministic layer more safety enables more