This type of book is the best example of classic content that never gets old. Methods change but concepts are the same.

I read a buzz word "Knowledge AI Platform" but I did not see any specific feature helpful for knowledge management like verification or transparency. It is more like any generic Agent builder. Maybe it meant to justify building something internally.

None of these have been especially impressive.

I think that’s because none of them go past “I’ve set up agents to be orchestrated this way” and that’s about as impressive as “look at my cloudformation template”.

To be honest perplexity does nothing to make sure it's answer are correct let alone the citations. They just look plausible. For anything little bit serious I use nouswise or nblm that sometimes abstain instead of making things up.

Yes this paper even tries their enterprise case. Perplexity performs worse. https://arxiv.org/html/2604.17843v1

I was wondering what the voice agent or engine behind it is. It sounded better than 11labs.

This is an active area of research to inject humility into llms in order to create some kind of knowledge boundary. You can look this paper from nouswise https://arxiv.org/html/2604.17843v1 and the product build on top it to try the humility.

How would the right format for you look like? Specially be monologue?

It could be monologue or dialogue, but it should be less full of verbal "syntactic sugar" than NotebookLM podcasts. More to the point, more detail-oriented, can pronounce math.

Can pronounce math is the real showstopper for me. Last time I tried, NotebookLM would try to say the TeX out loud. Like underscore dollarsign...

It's been sometime I have been using an app called nouswise. I switched right after they added nblm notebooks in Gemini. I knew they cannot stand another app competing with their flagship. But anyway, it do have audio recap and it definitely have way less of "syntactic sugar". I haven't tried but because it's agentic you can ask it to generate a monologue for you. But I assume you should be in deep mode. I so far only used the side bar for this.

The backend has changed people experience more hallucinations. You can see their subs.

It's cumbersome to have llm duplicate all quotes and still get good response. Else you get no visibility into your sources.

Best closest I stumble on is nouswise. Lets you contain an agent with even more sources and generate almost all outputs.

No pricing, "call for demo". Not sure this is for personal use.

Pricing with a long table comparison for the 3 plans https://nouswise.com/pricing

Are you looking for this?

I use it for personal use. They have 3 plans, 1 free 2 paid.

Like how many sources? And what you mean by sources (tables pdfs...)?

I have several projects with lots sources (800+). I mostly work with webpages, papers and youtube videos. It also supports mcp if you want to connect something.

Most VCs avoided application layer believing it is too risky with few player would emerge as winner over application layers calling them GPT wrapper (now called Harness) and pouring money into infra layer. Would love to see your opinion about how this thesis would turn out going forward.

Not my experience. I think most VCs thesis is around the application layer - not much around the infrastructure.

That being said, while I am biased, there is a lot of work around infrastructure so calling it "just a wrapper" massively underestimates the effort - this is purely from my own experience building this space.

Besides, if it is true how come OpenClaw is spending so much money on a open source project. Salaries alone will cost 7 digit sum for a harness and I have first hand experience dealing with companies doing exactly this.

Shameful plug - we are building cbk.ai, better known today as chatbotkit.com.

Your post title is misleading and false. They didn't archive overnight, they raised the money last year, the founder explains in comments.