Yeah, it seemed like there was more than necessary, but I haven’t played around with metaprompting much so got some nice ideas.
@yoheinakajima
-
Prompt Rewriting and Ontology-Based Example Generation for LLMs
By
–
Pretty interesting. I like the prompt rewriting approach they have. I think I also saw they use the ontology to have the LLM generate examples (for context) that match the user set ontology, which is nice and makes sense.
-
AI Market Specialization and Consolidation Strategy
By
–
I suspect many will start to specialize initially (SMB, API, verticals, consumer, etc) so they can capture large market share in a smaller market before expanding back out and also consolidation or being bought into cos that add this to their existing offerings.
-
Glasp: Import Kindle highlights into your second brain
By
–
You might like @_Glasp
! Import all of your Kindle highlights into a second brain that you can ask questions too. -
Agents and RPA: Automation Intelligence Integration
By
–
👉👈
— Yohei (@yoheinakajima) 1 juillet 2024
👉👈
Agents RPA pic.twitter.com/YeyOxUTJnkAgents RPA
-
RAG limitations: moving beyond data retrieval to pattern summarization
By
–
Also as some ppl pointed out, this is harder to get right so quality varies across solutions and none working perfectly yet.* People don’t want to just find specific pieces of data, but want summaries and patterns. This is more than basic RAG. *tell me if you find one!
-
Market Space Becoming Increasingly Crowded in AI
By
–
*I'm excited about these, this is just an observation about how crowded the space is getting.
-
Talk to Your Data Startups Trend Analysis
By
–
When was the "talk to your PDF" trend? I'm now seeing dozens of "talk to your data" startups.*
-
Robust Router Design for Complex Agent Chaining Flows
By
–
I think you could do this for agents, but the chaining/prompting means it would need to be a much more robust (complicated) router… probably home grown and paired with efforts to fine-tune parts of the flow.
-
Dynamic prompts for different AI models optimization
By
–
Good point on mimicking what Claude uses in their system prompt. Does it mean I need dynamic prompt based on model I guess?