nickk81

It goed to show how negative HN is

Built an HTML wiki based on ideas from: The Unreasonable Effectiveness of HTML (@trq212)

Which way do they face?

- this is version 2 - you can now chat with the knowledge base - you can browse/view the wiki's - there's a nice graph/diagram

I built this for a security KB with a limited number of sources... But creating the Wiki's was fully automated, and it's easy to add/edit them with AI or manually.

On what topics would you like to see a Wiki LLM + knowledge base?

not open source, and early days yet... it's hosted online and not locally.

will work on a version 2 now.

I built a system to create a Wiki LLM (see tweets from Karpathy). Here's a public view of the results.

This is just based on 4 PDFs related to security (NIST CSF, security by design). It works on any input given (via PDF, web pages, etc.).

I implemented Live Chat, it works nicely. I didn't use my knowledge base for it yet... If I do it slows the whole thing down. So now it's just a limited prompt.

I built a "Both sides AI" inspired by Karpathy (https://x.com/karpathy/status/2037921699824607591).

- Will AI help the global economy? - Is Messi the best footballer ever? - Will we have AGI by 2030?

This tool does 3 passes of analysis for both sides of the argument.

- multiple models at once - rich content - retrieve web pages (for additional chat context)

You mean, how much would this cost for Enterprise?

It can be competitive, it's basically just the base model cost plus some markup.

Try AI chat with Rich Output (maps, locations, follow up questions, check lists, etc.)

Get rich content (output) instead of just text output in AI chat.

I built an AI chat that shows helpful Rich Content along with the typical chat.

It supports the following: - Gemini Flash - Gemini Flas lite - Grok 4.1

More models to come!

What other Rich Content types should I add?

Now at version 4... I've added some more features to my analysis of "Is my job safe from AI?"

- Treepmap of all occupations - Guage of risk per occupation - Sharing options - AI analysis from 3 scenarios - AI analysis from first, second & third order impact - Data from BLS

Each occupation gets a 1–10 risk score. You can search by job title or browse by category. There's also an interactive treemap where cell size = employment count and color = risk level, which makes it easy to see where the exposure is concentrated by volume.

The part I found most interesting to think through: I didn't want to pretend there's one answer, so I built three scenarios: pessimistic (AI displaces without replacement), moderate (net-stable, winners and losers), and optimistic (AI expands economic activity faster than it destroys roles). The treemap switches between all three.

Inspired by Andrej Karpathy's US Job Market Visualizer. Built on top of 99helpers.com, my AI customer support tool — the job risk checker is a free standalone tool.

Is my job safe from AI?

Inspired by Karpathy, I analyzed jobs to determine whether they're at risk from AI.

The interesting part:

I used 3 different scenarios:

Scenario 1 — AI Eliminates Jobs AI takes jobs; few replacements created

Scenario 2 — AI Transforms Jobs Some jobs lost; new ones created

Scenario 3 — AI Creates Opportunity AI generates new demand and job types

And I added first, second & third order effects: First, Second & Third Order Effects How AI disruption cascades through this occupation, the broader industry, and society at large.

I took it a step further in V2.

What else should I add?

Hey HN, I built a free AI Answer Generator as part of a growing suite of tools on 99helpers.com.

You type a question, pick a language (14+ supported) and a tone, and get an instant AI-generated answer. No signup required.

A few things I'd love feedback on:

- Is the response quality useful, or does it feel too generic? - Any languages or tones you'd want to see added? - How does the UX feel, anything friction-y?

Try it here: https://99helpers.com/tools/ai-answer-generator

The full free tools suite (FAQ generator, fact checker, paraphrasing tool, etc.) is at https://99helpers.com/tools

Thanks for any thoughts!

GREAT Feedback. It really is. I like the concept of a C suite as well.

To be fair, currently it's very simple, and no the data is not accurate. But it would indeed be nice to tie it to real numbers.

I'll work on a better response (i.e., if you select all, only the most relevant pop up, and only relevant short answers given).

built a demo where you can chat with an entire AI leadership team — CEO, CFO, CTO, COO, a lawyer, a dev, an analyst, and an intern. Each has a distinct personality, expertise area, and communication style. You can talk to them individually or throw a question into a group chat and watch them respond from their own perspectives.

It's built on top of my customer support platform (99helpers.com), but I made this mostly to explore a question: what happens when you give AI characters persistent roles and let them interact with the same prompt from different angles? The group chat mode is where it gets interesting — ask something like "should we raise prices?" and you'll get the CFO's spreadsheet logic clashing with the COO's operational concerns.

Would love to hear what people think about multi-persona AI interactions as a UX pattern.

Chat with the DIGITAL TWIN AI C-suite: I thought it would be interesting to have like an AI digital twin c-suite that you can chat with, using the knowledge base of the company.

Would you like your own AI c-suite?

Thanks, this mini tool is mainly an SEO play (long term), and it's a basic version right now. So I want to improve it over time.

I have a basic answer generator. What should I add?

Well, ads are not 'nice' as such, this is just better than not having an option to talk to someone.

This is more a demo of the future of Ads.

I built a chat interface that extracts context (locations, budgets, goals, preferences) from natural conversation and visualizes it in a live side panel.

The idea: most AI chat is a flat wall of text. But when you're planning a trip or scoping a project, the structured information matters more than the prose. So instead of burying it in paragraphs, the UI pulls it out and displays it — maps for locations, cards for budgets and preferences.

I recently had another submission here on HN: https://news.ycombinator.com/item?id=47205890

Limits:

- Only supports a pre-defined user profile

- Note that the underlying AI model is simple, to keep costs down, this is about the concept of Visual Contextual Chat.

FIXED IT!!

I SEEM TO HAVE HIT THE RATE LIMITS FOR THE DAY

You can still have 10 different providers with advertising.

Although I agree more competition will act as a counter to spoiling the experience with advertising.

imho, the hosted solution will always be better, the major players will offer better integrated chat, and they'll have budgets to do so, as long as advertising income is available

Fight fire with fire