Hacker News new | past | comments | ask | show | jobs | submit login
Show HN: nbi.ai – Generative Business Intelligence (narrative.bi)
98 points by fromthegut on Nov 17, 2023 | hide | past | favorite | 27 comments



A pet peeve of mine, as someone who has been in the BI domain for a few years. There is no such thing as data driven decisions. What is there, is decision driven data, i.e. what you monitor indicates your desired decision making toolset.

This comment is not specifically targeted against this product, as other than that marketing part the rest of claims are very expected and valid. I do have reservations about whether generative AI produces much more value in this context than older rule-based NLG approaches. I.e. I would not expect the divergence between rule-based and statistical inference narratives to be significantly different in the context of traditional BI, where data unexpectedness is hamstrung in rigid data models.


Hi HN, Yury and Michael are here to answer your questions.

2 years ago, we launched Narrative BI to make data analytics available for everyone. Since then we helped 2,500+ teams get insights from their marketing data.

Our backstory is here: https://news.ycombinator.com/item?id=33627953

We started as a freemium product and built a unique business data set. We used this dataset to launch NBI.AI - a Generative BI platform that can be connected to virtually any structured data source (unlike our previous version designed for specific data connectors).


Hey folks, I'm a free customer here and a big-big fan of their project. I've been using narrative.bi for my project and sharing the results on Twitter [1] lately.

How I'm using narrative.bi: - A high-level report of WoW, and MoM changes on our website traffic that land in my email. - Alternative to GA4 UI, which drives me nuts.

What I don't get yet and I want to see implemented (I'll definitely pay for this): - Dimensional analysis: Which dimensions contribute to the topline change the most. I.e. Say unique users increased by 30% WoW, what top 5 dimensions have contributed to this change? Country? Device? Etc - Global event correlation: are there any global events that influence the metrics change? Maybe a holiday in the Country?

[1] https://twitter.com/avysotsky/status/1723403222906286185


Hey @vood, thanks for the feedback!

We actually improved our drill-downs recently. Check it out: https://www.narrative.bi/product-updates/purchase-insights-g...

It's totally OK to use Narrative BI in a no-UI way via email or slack, but our web UI provides more advanced insights so you may want to check it from time to time. Simply click on an insight in your email and you'll be redirected to a more detailed view.


Can you unpack: "Drive growth with AI-generated data narratives." Superficially, that sentence reads as fluff corporatespeak


Thanks for the feedback!

Historically, we've been focused on growth use cases (marketing, sales, PLG). "Data narratives" is how we call our natural language written insights (AI generated). So it might sound like a corporate jargon but actually describes our value proposition. Will think how to make it more clear. Thanks!


My interpretation: Use AI to understand the data, so that you can get more customers / serve your existing customers better.

Disclaimer: I’ve never tried their product


Your interpretation is accurate! The self service connectors we build are mostly marketing/ sales data sources, so the narratives we generate are focused on growth insights and recommendations.

Feel free to try: would love to hear your feedback!


> Feel free to try

I will, but not yet. I am building a product, slowly. But my only user so far is myself :p


Good luck then!


I thought it was a parody website until I saw the founder commenting here.


how do you ensure that the AI that you use does not hallucinate? (assuming you use some type of LLM)


We do a lot of pre-processing and post-processing with iterative generations and a voting system to ensure accuracy.

AI is mostly used for natural language generation.

For insight-generation, we built our own algorithm (ML-based).


Can you share some examples of what people are using this for today?


You can learn more about our solutions here https://www.narrative.bi/solutions

But let me describe the most popular use cases:

1. Report automation. People connect their marketing data sources and get automated reports on key metrics 2. Anomaly detection. We generate alerts when we see unusual patterns that affect your key business metrics 3. Natural language insights: we uncover interesting patterns and correlations in your data and provide recommendations on how you can improve/optimize your campaigns


Do you use that data for training? Given this:

> We started as a freemium product and built a unique business data set. We used this dataset to launch NBI.AI - a Generative BI platform that can be connected to virtually any structured data source

it sounds as if you have done so in the past.


We don't use customers' data (eg data from your data sources) for training purposes. What we use is meta-data like objectives, behavioral data, preferences and feedback loop (was it helpful y/n) to personalize the insights.


"we don't literally steal your data but we surveil you to harvest data from you like a skinner box."


"we don't use customer data, we process it and use the generated data"


The next logical step after ad tracking -- you will use their website and willingly give them your info for better surveillance.


Not a customer nor have I used the app, but I think one use case you're hitting is the following:

I'm an ad management agency that does google ads for 200x ecommerce companies. Every month I have my marketing specialist create a powerpoint presentation with graphs of their performance + a narrative about it (ROAS was up 23.5% but this was mostly due to on-brand search). Instead of my marketing analyst creating this presentation, I just generate it with nbi.ai.

Do I have that right?


This is so accurate, it feels as if our actual customer wrote it :)

We helped many agencies to automate reporting the way you described: https://www.narrative.bi/solutions/agencies


When will there be an Automator-like tool which will allow a user to create basic workflows?

I have a stack of several hundred invoices with checks which I take photos of and then need to rename to match the Invoice ID.

If I could also have it find the check #, amoutn, and date of the check and some other information and use all that to construct a filename --- that would be ideal, but just getting the Invoice ID would be a nice start.


OCR. Tesseract. Huginn. Python. Mo AI needed but add a LLama and you're home rolled. N


you may want to check https://www.veryfi.com/


Any plans for a shopify integration?


Yes, we plan to add Shopify, Klaviyo and many more integrations soon!




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: