Going from prototypes to a real product has taught me a few things:
1. What founders want to build is not always what users need—especially when those users are not tech-savvy.
2. Building a product people love still depends heavily on the founders’ taste. so the best thing you can do is to stop coding and visit an art exhibition.
3. Run AI agents codereview while you sleep.
4. Technical and non-technical co-founders should spend one day each week doing each other’s job.
15
We often hear that AI should not be involved in mental health counseling.If that means replacing therapists, I completely agree.
AI can never replace the human judgment, empathy, and trust at the heart of therapy.
But it can become the best assistant working behind the scenes.
Too often, therapists spend two to three hours a day handling insurance-related documentation.
When sessions are back-to-back, they have only a few minutes to flip through previous notes and remember where they left off with the next client.
General AI tools can help to a certain extent—but they weren’t built for clinical workflows, and they raise serious questions about how sensitive client information is handled and protected.
This is the problem and, at the same time, the reason we built BeeThere.
BeeThere is an AI assistant that designed for therapists by therapists.
It helps therapists prepare for sessions, create structured clinical documentation, reflect on complex cases, and connect treatment plans with progress notes through one continuous “Golden Thread.”
It is not here to make clinical decisions. It is here to handle more of the work surrounding therapy, so therapists can focus on the work that only they can do.
We’re now inviting a small group of therapists, counselors, and clinical social workers to join our beta and help shape BeeThere around real clinical practice.
BeeThere is free during beta, requires no credit card, and takes less than five minutes to set up.
beethere4u.com
We’ll BeeThere for you, so you can be there for your clients.
#MentalHealth #Therapists #HealthTech #ClinicalDocumentation
In 4 years, my coding workflow changed 5 times:
Hand-code + Stack Overflow → ChatGPT copy/paste → Cursor → Codex/Claude Code with parallel agents → Conductor with isolated worktrees.
This evolution is unstoppable and isn’t just better AI. It’s moving to orchestrating agents in parallel. so what's the next ?
11
一直觉得crewAI这样的公司核心就是角色扮演,不同的prompt给到同一种模型,好歹用不同模型之间沟通都能理解。cosplay在本质的没有并不能提升模型能力,还是要通过解决kv cache / context window 的问题来提升最终表现
72
Simon Fan retweeted
8 RAG architectures for AI Engineers:
(explained with usage)
1) Naive RAG
- Retrieves documents purely based on vector similarity between the query embedding and stored embeddings.
- Works best for simple, fact-based queries where direct semantic matching suffices.
2) Multimodal RAG
- Handles multiple data types (text, images, audio, etc.) by embedding and retrieving across modalities.
- Ideal for cross-modal retrieval tasks like answering a text query with both text and image context.
3) HyDE (Hypothetical Document Embeddings)
- Queries are not semantically similar to documents.
- This technique generates a hypothetical answer document from the query before retrieval.
- Uses this generated document’s embedding to find more relevant real documents.
4) Corrective RAG
- Validates retrieved results by comparing them against trusted sources (e.g., web search).
- Ensures up-to-date and accurate information, filtering or correcting retrieved content before passing to the LLM.
5) Graph RAG
- Converts retrieved content into a knowledge graph to capture relationships and entities.
- Enhances reasoning by providing structured context alongside raw text to the LLM.
6) Hybrid RAG
- Combines dense vector retrieval with graph-based retrieval in a single pipeline.
- Useful when the task requires both unstructured text and structured relational data for richer answers.
7) Adaptive RAG
- Dynamically decides if a query requires a simple direct retrieval or a multi-step reasoning chain.
- Breaks complex queries into smaller sub-queries for better coverage and accuracy.
8) Agentic RAG
- Uses AI agents with planning, reasoning (ReAct, CoT), and memory to orchestrate retrieval from multiple sources.
- Best suited for complex workflows that require tool use, external APIs, or combining multiple RAG techniques.
👉 Over to you: Which RAG architecture do you use the most?
_____
Share this with your network if you found this insightful ♻️
Find me → @akshay_pachaar ✔️
For more insights and tutorials on LLMs, AI Agents, and Machine Learning!
Made a WebApp within 10 mins from ideaition to development to host by using @GeminiApp
I develop this app because I HATE WRITING COVER LETTER!
30
Simon Fan retweeted
ZeroNode is officially open!
Users who submitted their emails on the waitlist + NFT holders can sign up immediately, create an account and start running the extension.
Want to join? Our members have some ref codes to share!
Install now at kaisar.io
Simon Fan retweeted
The Chainbase AVS Mainnet is officially live!
This marks a huge milestone in decentralized data infrastructure. Let’s break down the key features of this launch and what it means for the Web3 and AI ecosystems.
Read the full blog: blog.chainbase.com/avs-mainn…
Simon Fan retweeted
下午课上,教授突然疾步走到角落拿起某同学的书包,问他:“可以吗?”那个男生迟疑片刻:“当然。”全班都很疑惑。然后她拉开书包的拉链说:“我可以看吗?”那个同学:“额……可以吧。”教授此时大声地问所有人:“我有权这样做吗?”大家才从这半分钟的闹剧里回过神来说:“没有。”

















