Moneyweb Crypto

AI democratizes research for DIY investors while threatening traditional wealth managers

26 min Listen to the original Episode page
Artificial IntelligenceDIY InvestingWealth ManagementJSESEC FilingsTokenized AssetsAgentic TradingMarket Data

Summary

Host Kieran Ryan and guest Shiven Moodley of Novaque discuss how AI tools like OpenAI and Claude provide DIY investors with sophisticated research capabilities previously reserved for professionals. Moodley highlights that a 2025 Natixis survey found 52% of wealth managers view AI as a competitive threat, while a University of Chicago study showed AI outperforming humans in earnings predictions. He outlines practical workflows for analyzing JSE and US stocks using AI agents and public data sources. The conversation also covers Novaque's multi-agent ecosystem, insider trading strategies based on SEC filings, and the emerging landscape of tokenized assets and agentic payments.

Episode sections

  1. 0:12 – 1:48

    Intro and AI in Wealth Management

    Host Kieran Ryan introduces guest Shiven Moodley of Novaque, discussing how AI provides DIY investors with sophisticated research tools at low cost. A 2025 Natixis survey is cited, noting 52% of wealth managers view AI as a competitive threat.

  2. 1:48 – 3:28

    AI Tools for DIY Investors

    Moodley explains that AI models like OpenAI, Claude, and Grok allow investors to synthesize financial data from sources like SENS news for JSE stocks or US markets. This makes complex analysis accessible to beginners, moving them beyond 'vibe investing'.

  3. 3:28 – 6:19

    Practical AI Analysis Workflow

    Moodley outlines a workflow for analyzing JSE-listed companies: using OpenAI's 'Deep Research' for macro views, identifying stock candidates, downloading financials and price data from Yahoo Finance, and loading them into the AI for investment perspective assessment.

  4. 6:19 – 7:06

    AI vs Human Analysts and Hedge Funds

    The host references a University of Chicago study showing AI outperforms humans in earnings predictions. Moodley notes hedge funds are now buying data on which stocks AI agents are scanning, treating it like Google Trends to anticipate market moves.

  5. 7:06 – 9:35

    Limitations and Education

    Moodley argues that while AI improves research, sophisticated financial planning (tax, long-term wealth impact) still requires human judgment. He advises DIY investors to use AI to learn concepts like DCF (Discounted Cash Flow) using prompts like 'ELI5'.

  6. 9:35 – 14:41

    Novaque's AI Agents and Insider Trading Strategy

    Moodley describes Novaque's multi-agent ecosystem (100 agents) analyzing macro/micro data. He details a strategy using public SEC filings to track insider trading by top 10% holders and politicians (e.g., Nancy Pelosi), taking options positions based on this data.

  7. 14:41 – 16:07

    Performance and Weather Trading Pilot

    Moodley reports a ~40% gain for the year using this structured, risk-managed approach. He also mentions a pilot project using AI to predict NY weather 12 days out and betting on Polymarket, yielding a small $10-$20 profit.

  8. 16:07 – 26:12

    Tokenized Assets and Agentic Payments

    The discussion turns to tokenized stocks and crypto options markets on chains like Solana and Ethereum. Moodley advises focusing on a few assets to avoid burning tokens on API costs, noting that agentic payments enable 24/7 automated trading.

Assets discussed

AssetKindStanceContext
SolanaSOLcryptoneutralI I agentic payments have become one of the the big requirements from these big blockchains like Solana, uh Avalanche, um Ethereum as well, in terms of how of base, I know base as well, Coinbase.
AvalancheAVAXcryptoneutralI I agentic payments have become one of the the big requirements from these big blockchains like Solana, uh Avalanche, um Ethereum as well, in terms of how of base, I know base as well, Coinbase.
EthereumETHcryptoneutralI I agentic payments have become one of the the big requirements from these big blockchains like Solana, uh Avalanche, um Ethereum as well, in terms of how of base, I know base as well, Coinbase.

“Stance” is how the speakers talked about the asset, not a recommendation.

Claims worth verifying

A 2025 wealth survey by Natixis found 52% of wealth managers said AI was making automated advice a meaningful competitive threat.

Unidentified speaker (speaker_1) ·0:12

How to verify: Search for 'Natixis 2025 wealth survey AI automated advice' in financial news archives or Natixis press releases. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

University of Chicago researchers found ChatGPT outperformed human analysts at predicting the direction of company earnings when given anonymized financial statements.

Unidentified speaker (speaker_1) ·0:32

How to verify: Search for 'University of Chicago ChatGPT earnings prediction study' in academic databases or news outlets. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

Shiven Moodley was nominated as one of the top professional advisors of the year by the Irish Business Post.

Unidentified speaker (speaker_1) ·1:12

How to verify: Search 'Irish Business Post top professional advisors Shiven Moodley' on the Irish Business Post website. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

OpenAI's 'Deep Research' feature can be toggled on to provide macro views and list ideal stocks for a portfolio.

Unidentified speaker (speaker_2) ·3:42

How to verify: Test the feature in the OpenAI Chat interface or review OpenAI's documentation on 'Deep Research'. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

Hedge funds are approaching AI labs to buy data on the most common stocks being scanned by AI agents.

Unidentified speaker (speaker_2) ·6:56

How to verify: Search for news reports on hedge funds purchasing AI agent activity data or 'AI agent stock scanning data'. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

Novaque was nominated in the Irish Business Post as Digital Asset Initiative for Ireland in blockchain.

Unidentified speaker (speaker_2) ·22:40

How to verify: Search 'Irish Business Post Digital Asset Initiative for Ireland blockchain Novaque'. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

Novaque's AI-driven options strategy was up about 40% for the year.

Unidentified speaker (speaker_2) ·14:41

How to verify: Request performance verification from Novaque or check for public disclosures if Novaque is a registered fund. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

A pilot project using AI to predict NY weather 12 days out and betting on Polymarket yielded $10-$20 profit.

Unidentified speaker (speaker_2) ·19:37

How to verify: This is a specific personal anecdote; verification would require Novaque's internal records or a public blog post detailing the trade. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

Goldman Sachs released a statistic that 55% of wealth managers transitioning assets from parents to kids are likely to lose the relationship.

Unidentified speaker (speaker_2) ·20:52

How to verify: Search for 'Goldman Sachs wealth managers losing client relationship parents to kids statistic'. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

Shiven Moodley has a tool on GitHub that curates documents for agents to build startup or hedge fund architectures.

Unidentified speaker (speaker_2) ·24:34

How to verify: Search Shiven Moodley's LinkedIn or X (Twitter) for a GitHub link to his investment tool. For recent events, verify 2026-08-12 through 2026-11-12. Use another period only when the speaker explicitly names it.

Do your own research

  • Verify the 2025 Natixis survey findings on wealth manager sentiment toward AI automation.
  • Review the University of Chicago study on AI vs. human analyst performance in earnings predictions.
  • Assess the reliability of AI-generated macro views and stock picks against traditional fundamental analysis.
  • Check the current accessibility and API costs for JSE data versus US market data for automated workflows.
  • Evaluate the performance claims of Novaque's insider trading strategy using public SEC filings.
  • Research the regulatory status of tokenized stocks and crypto options on Solana and Ethereum.
  • Investigate the risks of relying on AI agents for speculative markets like Polymarket weather betting.
  • Consider the tax and long-term wealth impact implications of automated, high-frequency trading strategies.
Full transcriptexpand
0:04
Unidentified speaker (UNKNOWN): The
0:04
Unidentified speaker (speaker_0): Money Whip Crypto Podcast, where we discuss all things crypto related.
0:10
Unidentified speaker (speaker_0): Your host, Kieran Ryan.
0:12
Unidentified speaker (speaker_1): For DIY investors, artificial intelligence opens up some interesting opportunities to go it alone. AI gives you access to sophisticated Specificated investment research at very little cost, and investment houses are starting to take notice. There was a well-surveyed wealth survey done in 2025 by Natixis. 52% of wealth managers said AI was
0:32
Unidentified speaker (speaker_1): making automated advice a meaningful competitive threat. University of Chicago researchers found that ChatGPT, given anonymized financial statements and carefully designed instructions, outperformed human analysts at predicting the direction of company earnings. You can see where this is going. It raises a compelling question. How much of what
0:52
Unidentified speaker (speaker_1): investors pay professionals to do could they soon do themselves? Joining me to explore the opportunities, risks, and implications for investment houses is Chivan Moodley of Novake. He's no stranger to the Moneyware Crypto Podcast. Chivin, by the way, was recently nominated as one of the top professional advisors of the year by the Irish
1:12
Unidentified speaker (speaker_1): Business Post, and he talks to us today from Dublin. Ireland. Hi Shavin, good to talk to you again.
1:17
Unidentified speaker (UNKNOWN): Hey
1:17
Unidentified speaker (speaker_2): Karen, good to be back and uh good to uh share some insight with the audience.
1:22
Unidentified speaker (speaker_1): Congrats on your uh nomination there by by the Irish Post. Um it it uh I'm sure there's plenty more to come and we w want to get into this a little bit today. Just tell us how far this has gone. What can the ordinary investor do with AI today that couldn't be done in
1:38
Unidentified speaker (UNKNOWN): the middle of the day?
1:39
Unidentified speaker (speaker_1): past, or where they would have required a professional analyst uh or an expensive research subscription? So tell us where this is going and what AI does for the for the investor.
1:48
Unidentified speaker (speaker_2): Yeah. So right now is actually we're sort of in this really great space where the DIY investor that was previously maybe they didn't want to read as much research. They were relying on maybe
2:03
Unidentified speaker (speaker_2): vibe investing with what's happening on X and punting a little stock here and beyond easy equities. Now has the tools and availabilities through these new AI models, whether it be OpenAI, Claude, Groc,
2:18
Unidentified speaker (speaker_2): to synthesize a lot of information that sits within these within with the stocks, whether it
2:25
Unidentified speaker (UNKNOWN): be
2:25
Unidentified speaker (speaker_2): US market or South African. market, because within the South African market, you have you have an opportunity to load in maybe the sense news, maybe investor related news, and get a deep dive of the financial aspects and even break them down even further if you're not a sophisticated investor. You could summarize that into
2:45
Unidentified speaker (speaker_2): being eighteen year old and you're starting to invest for the first time. AI is able to synthesize that information a lot more easier, making it a lot more accessible to make take these long term bets in investment, uh making the DIY investor a bit more sophisticated um than where they were maybe five, ten years ago.
3:03
Unidentified speaker (speaker_1): All right, let's make this a little bit more practical for the South African investor. Uh if I wanted to access a This JSE listed company, how would you take me from its annual report, which I can get online to a useful investment assessment using AI? What tools and information would I need and what would it cost me?
3:28
Unidentified speaker (speaker_2): If you're choosing a little simple uh bot, maybe OpenAI for free, cloud for free. You're limited to credits there, but if you want a bit more intensive, hard hardworking working models, you'll maybe go into a paid version, which is $20
3:42
Unidentified speaker (speaker_2): for maybe upload or an OpenAI subscription. But working through that process, where you would fundamentally start is maybe looking at some deep research. So you can quickly toggle on in your OpenAI. Let's, for example, work with OpenAI. I'm a deep user of OpenAI. And um you switch on,
4:02
Unidentified speaker (speaker_2): you go into chat. you switch on deep research, you'd like, hey, give me a breakdown of all the macro uh that's impacting the JC stock market right now. It's gonna give you a little deep research output, you know, maybe ask you some questions, you know, build out that uh that overall macro view for you. You can then say, Great, based on this, what are the
4:21
Unidentified speaker (speaker_2): ideal stocks that I may want to look at for my long term portfolio? Because I may be risk adverse. I want something to maybe hold for ten years, fifteen years. And uh it'll give you a list. It'll give you a list of JC stocks. You can go into their website, you can pull their financials and you can load it up into the model and ask it to
4:41
Unidentified speaker (speaker_2): give you a perspective on is it a good investment or not? That's where sort of the DIY investor currently is going. In the US stock market, ideally if you invest in the US stock market, that data is a lot more easily accessible through sort of these API integrations with Claude
5:02
Unidentified speaker (speaker_2): and OpenAI that have these plugins to maybe public data sources. But JSE might not be as accessible, so you might take public and inf you might take those, download that information, throw it in. and then maybe download the price from Yahoo finance or from a trusted source of price to give you an idea of where that price
5:21
Unidentified speaker (speaker_2): actually is. But once you have that you kind of have an idea of where price is, what is influencing the stock right now and when and how should I get in. Uh This has been a very common use on the DIY investors but also the student investors that just are
5:41
Unidentified speaker (speaker_2): working a little small job and they're punting a little five hundred to a thousand Rand um into an account just as a long term horizon. And they're looking at that with maybe their tax free investment that they're managing themselves, or just a small equities account through any broker broke of off their the choice.
5:58
Unidentified speaker (speaker_1): Okay. So
5:59
Unidentified speaker (UNKNOWN): right
6:00
Unidentified speaker (speaker_1): for so we we established that we can definitely do analysis far better than we could as a DIY investor using AI. But better analysis doesn't necessarily mean better returns. So where is AI most useful in making investment decisions and and where does it still struggle? Because um I mentioned there the U
6:19
Unidentified speaker (speaker_1): University of Chicago study found that AI is actually making better earnings predictions than human analysts. So w we kind we're of moving in the d in the right directions. But how should investors check its figures, challenge the conclusions before committing money to suggestions that are made by an AI model?
6:36
Unidentified speaker (speaker_2): Yep, and and what we have as something that's interesting after that Chicago paper was hedge funds actually approaching these AI houses or these AI labs to buy what is the most common stocks that are being looked at or scanned at by these agents. Because hedge funds need to take those particular positions in which the
6:56
Unidentified speaker (speaker_2): AI is sort of pushing and people are more looking at. So it's exactly like looking at Google Trend Search. But when looking at
7:06
Unidentified speaker (speaker_2): is it making research better? It's made research better where we were to where we are now. However, the sophisticated financial planning decision not necessarily should lie with an agent. That's where it is quite tough. And a financial planner might be in a better position to actually look at how is
7:26
Unidentified speaker (speaker_2): tax implications working? What are those consequences? How does that long term impact my financial wealth and health? But when taking and trusting these agents, we've seen the speculative market, um, sort of these introducing brokers use these agents to
7:46
Unidentified speaker (speaker_2): sell signals, to sell a variety of other aspects of research to try and get people to buy and trust these outputs. Um, yes, it is challenging, and you shouldn't trust everything that you read, but you also should do a significant amount of research. So why while doing company research, you should be
8:05
Unidentified speaker (speaker_2): educating yourself on how does a DCF cash flow work? What is why are they having too much of cash? Where are they investing? And understanding the business for yourself is actually a great way to actually learn through the process. Uh, one of the really great aspects that I
8:25
Unidentified speaker (speaker_2): passed on to a friend. that was getting into investing was they were going through this process of DIY investing. And I told them use the prompt, um which I I'm not sure it exists on Claude, but it exists on OpenAI. And you can say use ELI five to explain this idea to me. And ELI five is a trigger
8:45
Unidentified speaker (speaker_2): in OpenAI uh that allows something to be broken down to explaining to you like you are five years old. So it's actually a great learning curve as you're going through this research.
8:58
Unidentified speaker (speaker_1): Right. And DCF of course is discounted cash flow. And a lot of people need to an understand what that is. All right. Now you've done a little bit of work in this area yourself. Uh you you have developed a number of AI bots, if I can call it that. One that does the you know looks at the macro picture. In other words,
9:18
Unidentified speaker (speaker_1): where where is the world market going? What what's the outlook for you know the the next few months? And then you've also got one, if I understand you correctly, that looks more at the micro picture. Take us through that. Where where what what kind of research does it do? How do you collect this data? How do you act on it?
9:35
Unidentified speaker (speaker_2): 100%. Hundred percent. So the agent has the situation same fundamental reasoning as everyone else. However, we
9:44
Unidentified speaker (UNKNOWN): I'm
9:45
Unidentified speaker (speaker_2): gonna work on two aspects. I'm gonna work on a professional that has experience and then a person that doesn't. So a person that doesn't may want to learn through the process and go through the process as as so forth. But for someone who is experienced who who's is maybe a trader leveraging AI or financial planner leveraging AI,
10:02
Unidentified speaker (UNKNOWN): you
10:02
Unidentified speaker (speaker_2): can start thinking and curating your agents specifically to your IP and then closed looping that to learn. So our macro agent has both crypto side and a traditional side, and we look at aspects of game theory, we look at aspects of
10:21
Unidentified speaker (speaker_2): macroeconomics and also the micro with households, firms, how they're spending why. they're spending. And we start to translate that information into different aspects. So one of
10:34
Unidentified speaker (UNKNOWN): a
10:34
Unidentified speaker (speaker_2): more sophisticated model that we run is a multi-agent ecosystem that has about a hundred agents with different personalities. And it runs, it takes, let's say, a news headline and it translates that into outcomes and forecasts. And that sort of gives us a picture of
10:54
Unidentified speaker (speaker_2): both macro and micro and how we should invest, both across a crypto aspect on are we going to be rubbed? Is it a is it a weak time? Or is it a really good investment for the long run? And that has been really interesting because we used one of these models in um insider
11:14
Unidentified speaker (speaker_2): trading. information. So how have the top ten percent actually bought or sold shares in their companies in the US? And when we used these intellectual models uh more sophisticatedly, we were able to take positions in the options market over medium to long-term long term durations.
11:34
Unidentified speaker (speaker_1): And when you talk about insider trading, that that of course has got a little bit of a negative uh connotation. We we're not talking about n knowledge that is not available to the broad market. You're talking about politicians, for example, who disclose by law what it is they're investing in the US. Am I correct?
11:51
Unidentified speaker (speaker_2): Correct. One hundred percent. So I am not an insider trader to the FSEA that might be listening.
11:58
Unidentified speaker (UNKNOWN): Right.
11:58
Unidentified speaker (speaker_2): Um but uh the we use public information and we scrape public information domains and filings to the SEC in the US um in order to identify what are these major holders doing. And when we start analyzing that on a micro scale and we look at a macro scale
12:18
Unidentified speaker (speaker_2): of what policies are coming, who's lobbying certain aspects of policy. And we know a big push of that was now the Democrats taking the House and Senate. Um so we start looking at how were their portfolios adjusted. And this idea came from um, you know, Nancy Pelosi, uh one of the best
12:38
Unidentified speaker (speaker_2): all time traders of the world, as we know her. Um, but the number one politician for known for insider trading, if you're following a portfolio you're you're up significantly. Um
12:51
Unidentified speaker (speaker_2): and uh we follow the same idea. We we use the insider trading uh sort of function and uh we take uh options positions and uh this is allows us to take a very structured approach to the market allowing us to harvest premium but also protect our
13:11
Unidentified speaker (speaker_2): downside in in the long run.
13:13
Unidentified speaker (speaker_1): Right. So an option your your downside is limited to your the premium that you pay, uh, but your upside it can be pretty substantial on that. And so is this completely automated once you've gathered all this information and hoovered it up, you've made a the the the bot makes a decision or do you manually execute?
13:31
Unidentified speaker (speaker_2): So I still manually execute. I put everything up until uh risk management. So if you were thinking about a hedge fund and the process of a hedge fund, you're going to work all the way from quant researcher to maybe risk manager, compliance, and then portfolio manager, and then your your
13:51
Unidentified speaker (speaker_2): execution or your dealer. You should be the execution and the dealer because all that information that you bold build with your agents should be curated to a decision or a rating or a percentage. out of 100 that gives it its confidence. And
14:08
Unidentified speaker (UNKNOWN): so
14:08
Unidentified speaker (speaker_2): once we get to that level and we understand, yes, the financials make sense, the thesis make sense, we go into the market and we look at duration
14:18
Unidentified speaker (speaker_2): of those option spreads, and we take positions in those markets. Very similarly happens in crypto as well. So if people are thinking about crypto,
14:27
Unidentified speaker (speaker_2): you can do exactly the same thing in terms of looking at on chain data and then going into the options market and then curating straight things.
14:35
Unidentified speaker (speaker_1): Right. And uh so how how has that uh automated process been working for you? Has it been successful?
14:41
Unidentified speaker (speaker_2): Yes, it has been successful. We're up about forty odd percent for this year, um, which is which is not bad. Uh we're not aiming to shoot the lights out. We're just aiming to uh stay within a risk rate that I am comfortable losing and growing the portfolio steadily. And um
15:01
Unidentified speaker (speaker_2): that's where the the long-term aspects of of this works. Another one you know we were talking about off air is using polymarket data and uh weather. So one of my
15:15
Unidentified speaker (speaker_2): habits has been coming to reading uh journals that are published on archive and uh using that to curate models and uh ended up using a model from there to predict twelve days out weather in New York and then taking bets in polymarket to harvest uh little small premiums.
15:35
Unidentified speaker (speaker_2): But this is what AI does. If you are if you have the capability of the knowledge of finance or the market, market structure, um, or you're mathematically inclined, you have a very good advantage right now. Whereas if you're starting out or an intermediate, you have a very good
15:55
Unidentified speaker (speaker_2): baseline to actually build something that you can test and you can make it yours. Um and that that's that's that should be the mindset going into playing and tinkering with these things.
16:07
Unidentified speaker (speaker_1): Right. Of course with tokenized stocks and tokenized everything coming, the ability to have these AI bots not only generate the research, give you a percentage ranking of confidence in the likely outcome of the trade, and then execute.
16:22
Unidentified speaker (UNKNOWN): You
16:22
Unidentified speaker (speaker_1): know, so what you would need basically is a a digital asset wallet. which is then connected to various different exchanges. If it's crypto, it could be Luno, Vala, Kraken. Um, if it's options, I don't know what options market you're using, but you would want to connect there. I I would imagine that you were getting to the point where these things
16:42
Unidentified speaker (speaker_1): could operate fully automatic. You know, you don't only need to check in on it maybe once a quarter or something like that. Is that where you see this going?
16:49
Unidentified speaker (speaker_2): Yes, I I I do. I I agentic payments have become one of the the big requirements from these big blockchains like Solana, uh Avalanche, um Ethereum as well, in terms of how of base, I know base as well, Coinbase. Um agentic payments and how these agents are
17:09
Unidentified speaker (speaker_2): really working with the environment, meaning twenty 24-hour four hour markets because tokenized stocks don't really close, have created an environment where you can constantly trade and your agents are constantly trading particular assets. And one thing I would recommend that don't focus on everything and everything under
17:29
Unidentified speaker (speaker_2): the sun. Stick to one particular asset or maybe two or three assets. So you don't burn tokens excessively. I ran up my token bill quite quickly because it was scanning the broader market of tokenization. Because we were testing to find arbitrage opportunities between
17:49
Unidentified speaker (speaker_2): exchanges pricing tokenized stocks. And uh
17:55
Unidentified speaker (UNKNOWN): you
17:55
Unidentified speaker (speaker_2): burn through credits very quickly. So it's important that you pick maybe an asset class or or a sector as well and hyperfocus on your agents learning those habits consistently.
18:08
Unidentified speaker (speaker_1): Right.
18:14
Unidentified speaker (speaker_1): So if you if you're scanning out a shotgun approach the whole thing, you're gonna burn through a lot of um computing power. So maybe uh focus on
18:23
Unidentified speaker (UNKNOWN): you
18:24
Unidentified speaker (speaker_1): know the top ten cryptos or the top ten AI stocks or something like that.
18:29
Unidentified speaker (UNKNOWN): Am
18:29
Unidentified speaker (speaker_1): I correct in saying that?
18:30
Unidentified speaker (speaker_2): Yeah, one one hundred percent. You only have a limit on your codecs or your Claude and if you're using their APIs, their APIs are are fairly expensive on on tokens. And tokens is that that compute power as you mentioned. So be mindful of how much of compute power you're actually using. Uh because for
18:50
Unidentified speaker (speaker_2): example
18:51
Unidentified speaker (speaker_2): I burnt through tokens on my weather, uh pulling weather data, but I could pull as much polymarket data as possible. So without narrowing polymarket data, my weather tokens burnt out in like an hour with it scanning the weather. So you don't want to rack up your bull on your third party provider,
19:11
Unidentified speaker (speaker_2): but you also don't want to burn through your your tokens on on your AI agents as well, which you won't know when it stops. So it's important to have those triggers that, hey, maybe to a Slack channel or to a telegram channel, hey, you're running low on tokens, you might need a top off. So
19:29
Unidentified speaker (UNKNOWN): keep
19:30
Unidentified speaker (speaker_2): those in mind. You need those triggers in in place.
19:33
Unidentified speaker (speaker_1): How did you do by the way, trading the weather in New York? I'm curious to know.
19:37
Unidentified speaker (speaker_2): We did fairly well. Hyper testing wise, we're up like ten dollars. Um we weren't really aggressively trading. It was like trading one cent consistently. Um and uh yeah, so ten to twenty dollars I'm I'm pretty happy with uh with that outcome.
19:54
Unidentified speaker (UNKNOWN): Oh
19:54
Unidentified speaker (speaker_1): that's it it sounds like a pilot. You you're you're running a pilot there, is that They right? own? Yeah. Okay.
19:59
Unidentified speaker (speaker_1): All right. Final question, Chivan. Uh let's go back to the the premise at the beginning. of this interview. We're talking about the big investment houses. Where do they maintain their edge in a market like this where AI is so freely available? You know, previously they used to be able to arbitrage their knowledge. They would they would be maybe a month or six
20:19
Unidentified speaker (speaker_1): weeks or a couple of months ahead of the market and understanding where the macro picture was heading or where a particular company was likely to. be going. That arbitrage has disappeared. So where do you think they get their edge going forward?
20:32
Unidentified speaker (speaker_2): I definitely believe their edge is coming from the relationship management, which is which is key. And it's also important to know there was a statistic that released, I think it was Goldman Sachs, that that put it out that fifty-five 55% percent of wealth managers that are transferred. transitioning assets right now
20:52
Unidentified speaker (speaker_2): from maybe the the parents to the kids are likely going to lose that relationship with the kids or the kids are going to move to a different financial manager. And that's a scary thought to believe because this arbitrage of information is existing. It's why am I paying you one point five to two percent on financial management if I'm seeing
21:12
Unidentified speaker (speaker_2): you once a year and you're not really adding value to maybe this this wealth process of I'm trying to generate. And the real the really important aspect now is how can these wealth managers leverage AI to not only stay ahead, but to keep this real-time
21:32
Unidentified speaker (speaker_2): analytics function off of their their clients to know who needs what when they need it without them dropping off. Because
21:41
Unidentified speaker (speaker_2): for wealth managers, particularly with middle income households that may come into strain right now, they may incur a problem that as interest rates maybe go up, the clients more exposed to more macro events, that they are likely to drop off on maybe policies that they have with them or or
22:01
Unidentified speaker (speaker_2): different products that they may have with them. And it's really important now that you can use AI in combination with this information to better align your customer to prevent them falling off over their long-term journey. However, there's a there's a catch in this. I've spoken to a lot of wealth
22:20
Unidentified speaker (speaker_2): managers, particularly with a company I was building called Via, uh we were were doing a blend of traditional and digital assets. We also were nominated in the Irish Business Post as Digital Asset Initiative for Ireland in blockchain. And what we when we were speaking to wealth managers in South Africa and Ireland, we
22:40
Unidentified speaker (speaker_2): identified that none of them are actually considering data or AI sovereignty. And why is that important? It's because when you use a perplexity, when you use an OpenAI, or you use a cloud, your data is passed on to those frontier models. It's not actually yours. And compliance teams with the
23:00
Unidentified speaker (speaker_2): FSCA haven't really thought about that as yet. On your information now sits in a cloud in Europe, it sits in a cloud in the US. However, it violates Appia Popier in this case of being accessible. Um we haven't really thought that far. And a lot of wealth managers or asset managers are like, oh yeah, I'm using Cloud to
23:20
Unidentified speaker (speaker_2): answer this question. But I'm like, you're not yes, it's a frontier model, it's great. But your data security in terms of that person's personal information is out and about. So for the advisors, it's important that you understand how your compliance is actually looking at these frontier models. For the
23:39
Unidentified speaker (speaker_2): retail customer, if someone's using an AI, you need to be sure you're asking them what AI are you using and are you sharing any information with these frontier models? How are you sharing information with these frontier models? Those are really key aspects because you don't want in a case your ID number being
23:59
Unidentified speaker (speaker_2): leaked on a frontier model. mark. Um And those are risks that are now coming through. So advise us to keep that in mind in as as this world develops.
24:09
Unidentified speaker (speaker_1): All right. And just to end off here, Shivan, uh I I believe you have uh uh a tool for anybody that wants to start experimenting and uh with with DIY investing. I guess the caution here is play around with it. Don't invest any money. See what you can do with it. Um
24:29
Unidentified speaker (speaker_1): train it. Um but it gives people a little bit of a head start. So how can they access that?
24:34
Unidentified speaker (speaker_2): Yeah. So the the m the tool sits on my GitHub, so you can reach out to me on LinkedIn or on X and I'll share that GitHub link with you. And so it's a skill that plugs into your agent and what it does is it curates the documents needed for your agent to go out and build the
24:54
Unidentified speaker (speaker_2): architecture of these uh of anything, whether it be a startup idea or whether it be a a financial hedge fund idea that you're playing around with and want to test. It gives you the necessary skills in which an engineering team, which would be your agent, is actually going out to execute. So um
25:14
Unidentified speaker (speaker_2): it's got an gotten really good reviews so far. I'm happy with it and I use it myself to very quickly think about ideas. So again, it's not for alive trading. Be sure you caution if you're trading your your own money. But it's a great head start to prep the
25:33
Unidentified speaker (speaker_2): documentation needed um to get you started.
25:36
Unidentified speaker (speaker_1): Terrific.
25:37
Unidentified speaker (UNKNOWN): We're
25:37
Unidentified speaker (speaker_1): gonna
25:37
Unidentified speaker (UNKNOWN): going
25:37
Unidentified speaker (speaker_1): to leave it there. Chivan Mudley of uh Novake, thank you very much once again for your time and look forward to talking to you again shortly, hopefully.
25:46
Unidentified speaker (speaker_0): Thanks for listening to the Moneyweb Crypto Podcast, hosted by Kieran Ryan.
25:52
Unidentified speaker (speaker_0): To listen to our other podcasts, go to moneyweb.co.za or the money MoneyWeb web app and follow money MoneyWeb web news for daily updates.
This page was produced automatically: the audio was transcribed with nvidia/parakeet-tdt-0.6b-v3+sortformer-v2.1 and interpreted by Qwen3.6-35B-A3B-UD-IQ3_S.gguf. Transcription and interpretation can contain errors; always confirm against the original episode. Nothing here is investment advice.
All insights