Ofer Mendelevitch

📚 Full list of books: aiexplainsbooks.com
On Amazon: www.amazon.com/stores/Alexi...

I’d love your feedback: Are these actually useful resources? And what topics should I tackle next? You can share ideas here: docs.google.com/forms/d/e/1F...

I hope some of these books help you learn something new.

AI Explains: ADHD for Adults Amazon.com: AI Explains: ADHD for Adults eBook : Enciso-Steinberg, Carmen, Piani, Alexis: Kindle Store

Can AI write books people actually want to read?

I’ve been testing that through my side project "AI Explains" - concise 100–150 page ebooks, written with AI & curated by humans.

Most recent book about Adult ADHD, with Carmen Enciso-Steinberg www.amazon.com/AI-Explains-....

I'm excited to share our new episode of the "Generation AI" podcast, with our special guest Matt Abrams - a great discussion about AI and Trust, AI regulation and how to gain the benefits of AI without losing empathy.

Enjoy!

t.co/BQZNjRRDEB

Announcing Vectara Agents: Enterprise AI That Works Introducing Vectara Agents

www.vectara.com/blog/announc...

I'm happy to share this repository: designed to be a community-curated list of agent failure modes, techniques to mitigate failures, and related resources - so that we can all learn from each other how agents fail, and build better agents.

PRs/Contributions welcome.

github.com/vectara/awes...

If you have built AI agents in the last 6+ months you know they are (unfortunately) quite fragile and can easily fail in production and at scale.

Agents can misinterpret the task, choose the wrong tools, hallucinate in their response, etc.

One difference that remains is that GPT-5 maintains low hallucination rate - super important for RAG and agentic flows.
We just added GPT-5-High to the @vectara.bsky.social hallucination leaderboard:

github.com/vectara/hall...

Responding to RFPs with Enterprise Deep Research

Super excited to share the news: RFPs no longer take days and weeks, but with @vectara.bsky.social enterprise deep research it's quick and easy.

www.vectara.com/blog/respond...

Introducing Enterprise Deep Research

Enterprise Deep Research = the agentic AI killer app for business. It's like "consumer" Deep Research, but grounded in your own private data. There are so many valuable use-cases: compliance monitoring, sales enablement, onboarding, etc..
See our recent blog post: www.vectara.com/blog/introdu...

Context Engineering: Can you trust long context?

LLM context windows grew from 8K tokens to 1M in the span of 1-2 years. But there is a problem: they don't work very well at longer context.
This makes "context engineering" important for RAG and agentic workflows.
www.vectara.com/blog/context...

Is Your RAG Consistent?

Does your RAG return consistent results?
Consistency in RAG is a mandatory requirement for regulated industries. I'm glad to share the new "consistency score" in Open-RAG-Eval.

blog: vectara.com/blog/is-your...

Vectara's Hallucination Corrector AI hallucinations create significant business risks and erode user trust. Vectara's Hallucination Corrector (VHC) identifies inaccuracies, suggests fixes, and provides essential guardrails for your AI...

Announcing @vectara.bsky.social Hallucination Correction. This is truly ground-breaking for enterprise RAG.
www.vectara.com/blog/vectara...

LLM Joke:
Why did the transformer go to therapy?
It needed more attention.

YouTube video by AI Engineer open-rag-eval: RAG Evaluation without "golden" answers — Ofer Mendelevitch, Vectara

Having fun in Day 2 of AI Engineer World Fair in SF.
Check out my online talk about @vectara.bsky.social's open-rag-eval and how to do RAG evaluation w/o golden answers:

www.youtube.com/watch?v=1cQl...

GitHub - vectara/open-rag-eval: Open source RAG evaluation package Open source RAG evaluation package. Contribute to vectara/open-rag-eval development by creating an account on GitHub.

Open-rag-eval
Rag evaluation without golden answers

github.com/vectara/open...

Vectara's Hallucination Corrector AI hallucinations create significant business risks and erode user trust. Vectara's Hallucination Corrector (VHC) identifies inaccuracies, suggests fixes, and provides essential guardrails for your AI...

Announcing @vectara.bsky.social Hallucination Correction. This is truly ground-breaking for enterprise RAG.
www.vectara.com/blog/vectara...

GitHub - vectara/open-rag-eval: Open source RAG evaluation package Open source RAG evaluation package. Contribute to vectara/open-rag-eval development by creating an account on GitHub.

Sharing the latest release of open-rag-eval 0.1.5 with connectors to @vectara.bsky.social @langchain.bsky.social and @llamaindex.bsky.social

github.com/vectara/open...

We finally have an animal selected for our upcoming O'Reilly book "Hands-on RAG for production" - the Brown Hyena. Two first chapters (as early release) expected in May.

Excited about the Data Council 2025 event next week.

We finally have an animal selected for our upcoming O'Reilly book "Hands-on RAG for production" - the Brown Hyena. Two first chapters (as early release) expected in May.

End RAG Sprawl: The Case for Platform Standardization As enterprises rush to integrate Retrieval-Augmented Generation (RAG), many are encountering RAG Sprawl—a fragmented, resource-intensive approach that leads to security risks, inefficiencies, and moun...

Do you have RAG Sprawl?

www.vectara.com/blog/end-rag...

GitHub - vectara/open-rag-eval: Open source RAG evaluation package Open source RAG evaluation package. Contribute to vectara/open-rag-eval development by creating an account on GitHub.

Happy to share the new open-rag-eval: a new approach in RAG evaluation.

Repo: github.com/vectara/open...
blog: www.vectara.com/blog/towards...

GitHub - vectara/open-rag-eval: Open source RAG evaluation package Open source RAG evaluation package. Contribute to vectara/open-rag-eval development by creating an account on GitHub.

Happy to share the new open-rag-eval: a new approach in RAG evaluation.

Repo: github.com/vectara/open...
blog: www.vectara.com/blog/towards...

a.co/d/5NEWr7W

End RAG Sprawl: The Case for Platform Standardization As enterprises rush to integrate Retrieval-Augmented Generation (RAG), many are encountering RAG Sprawl—a fragmented, resource-intensive approach that leads to security risks, inefficiencies, and moun...

Do you have RAG Sprawl?

www.vectara.com/blog/end-rag...

Announcing Vectara’s Python-SDK A Python-native interface for Vectara, built for developers.

Check out the new Vectara python SDK

www.vectara.com/blog/announc...

Introducing the Knee Reranking: smart result filtering for better results

We just launched knee-reranking - the ability to cut off irrelevant results automatically before sending to the generative step in #RAG.
Check out the details here: www.vectara.com/blog/introdu...

I'm excited to share that @vectara.bsky.social's API is now available on Postman's network, making it even easier to use for developers.

check it out here: postman.com/vectara
Blog: vectara.com/blog/announc...

Do u mean use LLMs to write protocol translation code?

@anthropicai.bsky.social has MCP, IBM is doing ACP.

OpenaAI has Responses API, and Google with AgentSpace) seem to focus on their closed ecosystem instead of an open standard (at least for now).

Where are @metaai.bsky.social and @microsoft.com in all of this?