This Will Change How We Work: The Power of AI ⚡️and Unstructured Data 🚀

With real Exhibits and actual Demos/Videos

Our industry cannot stop talking about AI's ability to handle unstructured data and communication. And rightly so as more than 80% data in any organization is unstructured. And it is the key factor in the differentiation and growth of AI companies. There is a also a growing volume of unstructured data and traditional structured systems/software struggle to process it. but why traditional software can’t deal with unstructured data? and how AI deals with it?

But wait… what is unstructured data? — text, images, PDFs, audio, video, etc.

When I was building Prove’s India business (acquired my company), I noticed something that bothered me a bit. I was building various teams but I noticed the sales team was spending a lot of time on tools like Salesforce. One of the biggest problems in Salesforce was that not only do you have to teach a new salesperson how to use it, but it is a software that requires you to understand its data structure. SaaS solutions like Salesforce rely on structured databases (predefined fields/forms). There is a particular way in which you do everything, like:

• create accounts and opportunities (each CRM is different)

• there is a particular way in which you report the stage/progress of each potential customer in the pipeline? what is the probability of closing this customer and how big could it be?

• there's a particular way in which you do entries in each of the predefined fields, and then it is able to take that input and show a dashboard to the senior mgmt based on the data you've fed.

• The problem was that it takes a lot of data and the feeding of data takes a lot of time.

• Every week, a salesperson in my team would spend 4-6 hours, where he was trying to go through those hoops, understand, and after every call and meeting, would input the data in a particular format in Salesforce.

• The end result was a dashboard that I could use and the management could use.

  
  This doesn’t need to happen in the GenAI era anymore.

  all of this time spent — the learning curve, data input, constant updates, etc. can be saved. AI can go directly to the source and record the entries. For example, the calls that the salesperson is having, the emails they are sending or receiving, the messages that they are exchanging with their customers can all be directly recorded, transcribed, processed, and utilized by AI agents to understand what is going on with each of the customers. Another AI agent can then create a dashboard and show you the analysis of where we are in the pipeline with respect to each customer. Custom built dashboards on the fly depending upon the queries. We don't need a structured software flow in the middle, we don't need to input in a structured way, we don't need to understand that software (and the learning curve), we don't need to spend hours and hours every week to input the data. Its just far more efficient and make the sales team more productive by using natural english language commands. And utilizing the actual communication across mediums plus things like voice notes that the salesperson could take.
  
  I’ve noticed that while many people are talking about AI, very few are actually walking the talk. So, I decided to back up my theory with some real examples—actual footage, aka exhibits. :)
  
  Exhibit 1:

• PDFs - here I am showing you the pdf report to excel sheet using a simple AI tool. think about the gazillion documents that the companies deal with every month.

• Imagine an investment bank or a hedge fund/PE investing in public markets. they receive so many annual/quarterly reports and earning call transcripts and financials of the companies in PDFs and other formats. and their models are in excel. so they look at these PDFs and input the data in the excel sheets. There are people doing this work day in and day out. this takes time and prone to errors. what if this could be handled by AI agents? like this one agent below:

Exhibit 2:
Another form of unstructured data is videos. This is a third-party AI tool wherein you can feed a 1 hour long youtube video and it will convert it into 90 sec clip, with auto captions and focus camera on the speaker(s). This one has not been built by us.

• Youtube - Opus clips - input and output screens

Exhibit 3:
A caveat - this is our own product at gAI ventures. Consumer version is called Stella and B2B version for US wealth mgmt mkt is called FastTrackr

• Let’s see a quick journey of one of the features in the consumer version first. from a voice input to text to taking action and interaction with a software. voice to a calendly type link that you can share

(sorry for the bit of an audio issue 🙏)

• and below is the B2B version of this product. if you are in the wealth management or RIA space and want to see the demo of this product….feel free to ping me or

  Vineet Mohan

Note - these exhibits I am showing you are usually a small part of a feature or a product. As they say a feature is not a product and a product is not a company.

In this essay and demos, you saw how AI turns text, images, PDFs, and videos into business gold. Unstructured Data and AI are match made in heaven. CEOs and founders of software and SaaS companies are not sitting idle. The AI version of crunchbase is being built by crunchbase itself :)

• This is the reason why Marc Benioff is spending so much time at Salesforce and specifically on AI. This is why he is promoting Salesforce as agent force. If rumors are true he may name the company “Agentforce”. he recently shared that of the 36,000 customer support inquiries Salesforce gets every week, 31,000 (86%) are now resolved by an AI agent. A month ago, only about 26,000 could be resolved by the AI agent.

• Infact the opportunity and the threat of AI is so real to their businesses that all the bigtech founders are back to the drawing board. For e.g.,

• You must have heard of the announcement by Klarna that they are replacing Salesforce for their CRM needs? I don't know if it they have already done it, but it is quite possible that all the structured software and systems that have been created over the last 20 years will get replaced. That is why founders like myself and Kushal in the AI space are also excited about the future. If we can take away couple of categories owned by previous generation SaaS companies it will be awesome. or if we can create new categories of software that were not possible before. it will be because of our ability to deal with and process unstructured data which was not possible before.

Our belief is that if you combine unstructured data handling with AI and domain expertise, it can be a force multiplier. For e.g., We are building an AI auditor for the US market. it is also way cheaper than how much audits used to charge. This is a human-in-the-loop solution. just to be clear, I'm not talking about autonomous audit agents yet. The co-founder of our company is a deep domain expert - with 18+ years in companies like Goldman Sachs and FIS and startups in compliance/audits. The audit report has to be reviewed and signed by someone like him. apart from other involvement across steps. But the big value add is that it can do audits in 2 weeks which is way lesser than what it used to take - 8 to 10 weeks by GT’s and PWC’s of the world.

Another example of a differentiated vertical SaaS AI dealing with unstructured data could be Healthcare image analysis (like AI for pathology slides). would require a couple of domain experts and a world class AI team.

Let me end this essay with how we should think about Opportunities if most of the above is true:

• Predict how industries will evolve with unstructured-first approaches.

• Speculate on what AI tools will do that SaaS platforms cannot (e.g., cross-modal analysis, self-learning insights).