Try it on infranodus.com Timecodes: 0:00 Introduction 1:12 💡 Case 1: Analyzing the main topics in relation to a search query on conversations 2:35 Trying to use ChatGPT and Perplexity AI for the same task… 4:10 Zooming into specific topics 5:22 ❗New feature: Filter a specific topic 6:57 Making AI think with you, not for you 7:21 Zooming into another topic, filtering the rest out (new feature again) 8:02 Resetting the filter to zoom out the discourse 8:18 ❗New feature: remove specific topics you’re not interested in** 9:30 💡 Case 2: Analyzing demand for a certain topic (related search queries) 10:28❗ New feature: save graph selections to notes** 11:37 💡 Case 3: Retrieve the main topics from multiple PDF documents 12:23 Retrieving the main topical clusters for a collection of documents 13:20 Zooming into specific subtopics of a cluster selected 15:05 💡 Cool feature: asking AI to elaborate on its own statement (manual agentic flow) 17:35 💡 Bonus content; Cognitive variability workflow - zooming in / zooming out / exploring / focusing
I can see myself and many others using it for research, learning, data mining, various business needs etc. The UX isn't bad, it's just colorful and feature-rich. But without a free plan, it's tough. I know you invested time and money into this, but think of a freemium model as a money MAKER, not a money losing strategy. You already built the tool, so 'freemium overhead' is minimal - data storage, bandwidth, some compute, etc but the upsell / conversion possibilities are very good with free users. I don't think having only a free trial on a paid plan is the way to go. Much respect, good luck!
That is good feedback. But I think that freemium model only works when you have a VC backed product and care more about showing growth than about sustainability. In our case I actually enjoy growing it organically. And I also think that it’s kind of like an exchange in commitment. I committed 10, actually 14 years of my life (if you count research) to make that happen, so I just ask for a small financial commitment in return. I just feel it’s fair and I also believe the value that the product brings is much more than the monthly fee I’m charging. Also, if the product is useful you’ll probably end up spending at least 4 hours with it every month, and with an average hourly rate in the west the subscription fee is only marginally increasing your expense (considering your time is also a resource).
2 หลายเดือนก่อน +5
Also freemium attract many users that won't pay. while a direct paywall wont stop people getting value and willing to pay for this value.
I just started my trial, I feel a fresh breeze of air, away from all the stupid prompting needed for something that makes sense. I work on SEO and this tool looks amazing! Thank you for creating it!
Great, I’m really happy to hear it! Please, let me know what you find more useful and which features you lack when you have time to play around with it. 🙏🏼🙏🏼
Fabulous content and possibilities for exploration. I use Evernote which has quite a number of notes and tags. I could see an application for appropriate tagging of notes and then a search within the collective notes for ideas, linkages, gaps etc. I wonder if this is an API integration or needs a dump of Evernote to Obsidian to work.
THANK YOU. Gpt summaries imo have been pretty useless for me because it glosses over a lot. Very interested in network graphs and this seems amazing for seo research! Do you know if there is a companion software that integrates numeric analysis? I would love to be able to track habits and symptoms and biodata to find connections.
@@Wavewave583 thank you! Yes it is great for SEO. Regarding numeric analysis - could you tell me a bit more about the source data and the insights you’re looking to get? I might have some ideas for you.
@@noduslabs Certainly! I currently track a lot of habits and activities throughout the day on a regular habit tracking app (that has yes, no, skip, and notes. I use the notes for each habit daily to track most of the data and just make connections myself mentally) Some are yes/no completion (eg worked out with personal trainer), some are numeric (eg number of awakenings throughout the night) some are time based (bedtime, wake time, time asleep, length of meditation etc), some use a numeric scale eg 1-6, 1-10, some of the data is semantic (content of dreams, type of running shoes used during runs, certain physiologic symptoms) I use a smart ring and track macros in MyFitnessPal. Some things I track have multiple data entries of diff types (1h spent with person x, or x medication x number of pills) I guess the other issue is that I track things throughout the day using my phone. I have come across things like neo4j but it feels pretty intimidating as someone who is a non coder.
Ah yes I forgot to mention it. Thank for brining it to my attention. It’s called The Conversation Book - here’s where you can find it: circadian.co/product/the-conversation-book/
I wonder if this app has its roots in qualitative research? I was involved with it when I was writing the master's thesis for my girlfriend in psychology.
@@minimal3734 not really as the roots are in text network analysis but it definitely can be used for qualitative research because of the topic modeling feature and the ability to slice off the top layers of ideas to get to the deeper latent topics.
Do you have educational licenses? I am an educator and we have no commercial interest. But I would like to test and experiment with your software... Thanks
Sir I have been periodically watching your videos for I'd say about a month now, I have one question if you don't mind. Is there any intention on your behalf to make this seamless as possible with Obsidian? Primarily regarding ease of import.
Went to infranodus, clicked live news and tried to get any sensible results on "Ukraine" and "economy". No luck. Lot of colorful nodes and edges to click, little useful results. Zooming into this topic ... not worked. Wanted some coherent summary for the conditions for the loans given to Ukraine and who will be the beneficiaries... search in graph did not work, could not force infranodus to zoom into this topic either. Am I stupid or is this tool lacking?
The live news graph sample uses the feed from the top 5 newspapers analyzing the headlines only. They are not writing about the loans today, there's much more news on North Korea getting involved in the picture plus the Hamas leader assassination. I understand that for you personally Ukraine may be the most important piece of news, but there are a lot of other things happening in the world and the graph reflects that objective picture. If you need a more in-depth info on a specific subject like loans to Ukraine, you need to get the news sources (or any other source data) that contain the information you need. For example, you could create a custom RSS feed that contains info on Ukraine and loans, like this one: rss.feedspot.com/brandmonitoring/search/5BnGt2Y=/ukraine%20loans/4bc6d8a1a23c54fc363183aea49effe7/rss - as you can see it only contains 13 items ranging from 1-week to 1-month old, so it's not a big topic on western media. Then you import this RSS feed into InfraNodus and you get a nice graph: infranodus.com/webinar/rss_ukraine_loans?background=dark&show_analytics=1&most_influential=bc2&maxnodes=150&labelsize=proportional&edgestype=curve&drawedges=true&drawnodes=true&labelsizeratio=2&dynamic=highlight&cutgraph=1&selected=highlight - which provides you the information you need. You can then use the AI box at the bottom to make the specific queries you mentioned. For instance, I asked about the conditions of the loans and it said: "The loans aim to repair Ukraine's power grid and other critical infrastructure, signaling a strategic collaborative effort among global leaders and agencies, despite challenges like Hungary's opposition to expanding sanctions." You can then use the built-in AI module to ask the questions you need. Regarding search and zoom in, they work perfectly fine, you couldn't find anything on the loans, because there was not anything on that topic in the main news headlines. Hope this helps!
I’d suggest to talk to a UX researcher and Product Designer to create a better user experience for day to day usage. There are so many features coming to the user that I don’t even know where to start or where to focus on.
People do use it though :) and sometimes some products are just not for everyday users and I’m absolutely fine with it. We made it as simple as it can be without reducing the functionality. And there is a simplified extension version and obsidian plugin.
Thank you for the update. I would be interested to hear your thoughts on this related video. NEW LLM & Knowledge-Graph Fusion: GIVE (UC Berkeley, Penn) th-cam.com/video/iGQLG0bWDxE/w-d-xo.html It looks like Open Source is finally catching up with some of your pioneering work.
I think that this would get more popularity if you'd open source it. At least some part of it. Keeping the cloud version. Just like modern solutions do.
What would be the advantage of being more popular? I still have bills to pay :) but I’d be curious to hear what would be the concrete advantages of making this open source
@@noduslabsI mean, more popularity, more cloud based users, more subscriptions, more money. Check how others open source projects grow with their cloud based solutions, while being open-sourced.
@@noduslabs it's a fair suggestion. Earning traction and traffic by sacrificing short term revenue, like what Meta did with llama. It's just a trade off.
Hi, Ca you add features like Highlight the summarised sentences based on the summery of main article inorder to retract summery point like a reference point.(I am keen to know how LLMs agents summaries articles without basic logical rules) Why LLMs agents have no well defined hardcoded static logical rules within their analysis for consistency as we know data can be misleading thus there should be defined parameters within LLMs to distinguish fraudulent data entries & discard them during their analysis with heuristics(rule of thumb) understanding of basic principles ?
In fact, InfraNodus highlights the top statements identified based on its topical analysis already (in the essential statements tab, when it makes the summary).
Try it on infranodus.com
Timecodes:
0:00 Introduction
1:12 💡 Case 1: Analyzing the main topics in relation to a search query on conversations
2:35 Trying to use ChatGPT and Perplexity AI for the same task…
4:10 Zooming into specific topics
5:22 ❗New feature: Filter a specific topic
6:57 Making AI think with you, not for you
7:21 Zooming into another topic, filtering the rest out (new feature again)
8:02 Resetting the filter to zoom out the discourse
8:18 ❗New feature: remove specific topics you’re not interested in**
9:30 💡 Case 2: Analyzing demand for a certain topic (related search queries)
10:28❗ New feature: save graph selections to notes**
11:37 💡 Case 3: Retrieve the main topics from multiple PDF documents
12:23 Retrieving the main topical clusters for a collection of documents
13:20 Zooming into specific subtopics of a cluster selected
15:05 💡 Cool feature: asking AI to elaborate on its own statement (manual agentic flow)
17:35 💡 Bonus content; Cognitive variability workflow - zooming in / zooming out / exploring / focusing
I can see myself and many others using it for research, learning, data mining, various business needs etc. The UX isn't bad, it's just colorful and feature-rich. But without a free plan, it's tough. I know you invested time and money into this, but think of a freemium model as a money MAKER, not a money losing strategy. You already built the tool, so 'freemium overhead' is minimal - data storage, bandwidth, some compute, etc but the upsell / conversion possibilities are very good with free users. I don't think having only a free trial on a paid plan is the way to go. Much respect, good luck!
That is good feedback. But I think that freemium model only works when you have a VC backed product and care more about showing growth than about sustainability. In our case I actually enjoy growing it organically. And I also think that it’s kind of like an exchange in commitment. I committed 10, actually 14 years of my life (if you count research) to make that happen, so I just ask for a small financial commitment in return. I just feel it’s fair and I also believe the value that the product brings is much more than the monthly fee I’m charging.
Also, if the product is useful you’ll probably end up spending at least 4 hours with it every month, and with an average hourly rate in the west the subscription fee is only marginally increasing your expense (considering your time is also a resource).
Also freemium attract many users that won't pay. while a direct paywall wont stop people getting value and willing to pay for this value.
@@noduslabsagreed
Exactly
@@CoachTu411🙏🏼
I just started my trial, I feel a fresh breeze of air, away from all the stupid prompting needed for something that makes sense. I work on SEO and this tool looks amazing! Thank you for creating it!
Great, I’m really happy to hear it! Please, let me know what you find more useful and which features you lack when you have time to play around with it. 🙏🏼🙏🏼
Dimitri one of your best videos ever. So well explained. THANK YOU.
this should be standard part of AI in general its amazing and will save so many lives thank bro :)
Nice work! Very interesting
*Nice 👍 video 📷 Thankyou so much for this video*
*Lot's of love 💕 from India.*
Just amazing! Very interesting and useful. Thank you for your valuable work. I will try it!
Let me know how it goes please!
Fantastic this really changes the game! Is the api exposed for this? - . Can this be used for vector based RAG?
Yes, it has an API - infranodus.com/api
Fabulous content and possibilities for exploration. I use Evernote which has quite a number of notes and tags.
I could see an application for appropriate tagging of notes and then a search within the collective notes for ideas, linkages, gaps etc.
I wonder if this is an API integration or needs a dump of Evernote to Obsidian to work.
There is both Evernote import and obsidian plugins available.
After commenting I watched another video and u clicked on a spot and then I saw the Evernote logo so will check it out. Thanx for quick response.
THANK YOU. Gpt summaries imo have been pretty useless for me because it glosses over a lot. Very interested in network graphs and this seems amazing for seo research! Do you know if there is a companion software that integrates numeric analysis? I would love to be able to track habits and symptoms and biodata to find connections.
@@Wavewave583 thank you! Yes it is great for SEO. Regarding numeric analysis - could you tell me a bit more about the source data and the insights you’re looking to get? I might have some ideas for you.
@@noduslabs Certainly! I currently track a lot of habits and activities throughout the day on a regular habit tracking app (that has yes, no, skip, and notes. I use the notes for each habit daily to track most of the data and just make connections myself mentally) Some are yes/no completion (eg worked out with personal trainer), some are numeric (eg number of awakenings throughout the night) some are time based (bedtime, wake time, time asleep, length of meditation etc), some use a numeric scale eg 1-6, 1-10, some of the data is semantic (content of dreams, type of running shoes used during runs, certain physiologic symptoms) I use a smart ring and track macros in MyFitnessPal. Some things I track have multiple data entries of diff types (1h spent with person x, or x medication x number of pills) I guess the other issue is that I track things throughout the day using my phone. I have come across things like neo4j but it feels pretty intimidating as someone who is a non coder.
What's the name of the book? Hard to tell from the video :) Awesome tool, thanks for sharing the tutorials.
Ah yes I forgot to mention it. Thank for brining it to my attention. It’s called The Conversation Book - here’s where you can find it: circadian.co/product/the-conversation-book/
I wonder if this app has its roots in qualitative research? I was involved with it when I was writing the master's thesis for my girlfriend in psychology.
@@minimal3734 not really as the roots are in text network analysis but it definitely can be used for qualitative research because of the topic modeling feature and the ability to slice off the top layers of ideas to get to the deeper latent topics.
Do you have educational licenses? I am an educator and we have no commercial interest. But I would like to test and experiment with your software... Thanks
Sure you can get the cheapest license then, it's pretty affordable!
Does it only have paid version?
Yes because it costs money to develop it. But there is a free trial.
thats some sexy UI/UX... which software is that?
@@pushingpandas6479 InfraNodus
Plan to open source it?
@@beckbeckend7297 what for?
Sir I have been periodically watching your videos for I'd say about a month now, I have one question if you don't mind.
Is there any intention on your behalf to make this seamless as possible with Obsidian? Primarily regarding ease of import.
There is a plugin for obsidian. Directly integrated. Very easy to use. infranodus.com/obsidian-plugin
Went to infranodus, clicked live news and tried to get any sensible results on "Ukraine" and "economy". No luck. Lot of colorful nodes and edges to click, little useful results. Zooming into this topic ... not worked. Wanted some coherent summary for the conditions for the loans given to Ukraine and who will be the beneficiaries... search in graph did not work, could not force infranodus to zoom into this topic either. Am I stupid or is this tool lacking?
The live news graph sample uses the feed from the top 5 newspapers analyzing the headlines only. They are not writing about the loans today, there's much more news on North Korea getting involved in the picture plus the Hamas leader assassination. I understand that for you personally Ukraine may be the most important piece of news, but there are a lot of other things happening in the world and the graph reflects that objective picture.
If you need a more in-depth info on a specific subject like loans to Ukraine, you need to get the news sources (or any other source data) that contain the information you need. For example, you could create a custom RSS feed that contains info on Ukraine and loans, like this one: rss.feedspot.com/brandmonitoring/search/5BnGt2Y=/ukraine%20loans/4bc6d8a1a23c54fc363183aea49effe7/rss - as you can see it only contains 13 items ranging from 1-week to 1-month old, so it's not a big topic on western media. Then you import this RSS feed into InfraNodus and you get a nice graph: infranodus.com/webinar/rss_ukraine_loans?background=dark&show_analytics=1&most_influential=bc2&maxnodes=150&labelsize=proportional&edgestype=curve&drawedges=true&drawnodes=true&labelsizeratio=2&dynamic=highlight&cutgraph=1&selected=highlight - which provides you the information you need. You can then use the AI box at the bottom to make the specific queries you mentioned. For instance, I asked about the conditions of the loans and it said: "The loans aim to repair Ukraine's power grid and other critical infrastructure, signaling a strategic collaborative effort among global leaders and agencies, despite challenges like Hungary's opposition to expanding sanctions."
You can then use the built-in AI module to ask the questions you need. Regarding search and zoom in, they work perfectly fine, you couldn't find anything on the loans, because there was not anything on that topic in the main news headlines.
Hope this helps!
It looks fancy, but not usable
I’d suggest to talk to a UX researcher and Product Designer to create a better user experience for day to day usage. There are so many features coming to the user that I don’t even know where to start or where to focus on.
The data visualisation would be an amazing design for a T-shirt tho 😂ngl it looks fancy
People do use it though :) and sometimes some products are just not for everyday users and I’m absolutely fine with it. We made it as simple as it can be without reducing the functionality. And there is a simplified extension version and obsidian plugin.
Not everyone cares about the looks
Thank you for the update.
I would be interested to hear your thoughts on this related video.
NEW LLM & Knowledge-Graph Fusion: GIVE (UC Berkeley, Penn)
th-cam.com/video/iGQLG0bWDxE/w-d-xo.html
It looks like Open Source is finally catching up with some of your pioneering work.
Thank you for the link! I’ll find time to watch it and will let you know! What are your thoughts on it?
I think that this would get more popularity if you'd open source it. At least some part of it. Keeping the cloud version. Just like modern solutions do.
What would be the advantage of being more popular? I still have bills to pay :) but I’d be curious to hear what would be the concrete advantages of making this open source
@@noduslabsI mean, more popularity, more cloud based users, more subscriptions, more money. Check how others open source projects grow with their cloud based solutions, while being open-sourced.
@@noduslabs Most companies just love one-click solutions, but small startups love open-sourced ones. One brings money, another brings popularity.
@@noduslabs it's a fair suggestion. Earning traction and traffic by sacrificing short term revenue, like what Meta did with llama. It's just a trade off.
Hi,
Ca you add features like Highlight the summarised sentences based on the summery of main article inorder to retract summery point like a reference point.(I am keen to know how LLMs agents summaries articles without basic logical rules)
Why LLMs agents have no well defined hardcoded static logical rules within their analysis for consistency as we know data can be misleading thus there should be defined parameters within LLMs to distinguish fraudulent data entries & discard them during their analysis with heuristics(rule of thumb) understanding of basic principles ?
In fact, InfraNodus highlights the top statements identified based on its topical analysis already (in the essential statements tab, when it makes the summary).