Suddersan Thevarajah

Investing in public equities building Aadi Capital.

[Suddersan Thevarajah @suddersan]

At what point is $UBER $LYFT #cybercab headless and instinct or Muse just books your ride from A to B (choosing best/cheapest option)?

a) Next 2 yrs
b) Next 5 yrs
c) Next 10 yrs

[Suddersan Thevarajah @suddersan]

Wouldn’t you say the rate change smarter more in terms of default rates than they do for funding costs or investment return? The biggest risk is NPL increasing, and significant losses on the credit book. I know allowances cover approximately 250% of the credit card and high interest financing - but that could blow out real fast if for example unemployment increases or the selic trends upwards.

[Suddersan Thevarajah @suddersan]

@InvestNorthwise What’s behind the 2027 loss? Assuming write-offs, how much $?

[Suddersan Thevarajah @suddersan]

@hridoyreh Tried this for local seo - any significant effects?

[Suddersan Thevarajah @suddersan]

Recently moved to LA and Looking to join a financial group/ community that focuses on public equity investing. Any recommendations on where to look?

#equityresearch #investing

[Suddersan Thevarajah @suddersan]

@velez_david Maybe with JEV, efficiency ratio drops even lower (already ridiculously low compared to incumbents, but still)? Perhaps the better capital efficiency that would come with this could be used to test new services that didn’t meet the ROI hurdle before?

[Suddersan Thevarajah @suddersan]

Believe neobanks (and banks who have started to implement AI) will benefit hugely from JEV in bringing down their token costs and improving their services such as KYC and onboarding.

$KLAR $NU

[Suddersan Thevarajah @suddersan]

https://t.co/t93gl1eqWb

Check out the desktop - lmk if you have feedback or simply disagree!

[Suddersan Thevarajah @suddersan]

With the introduction and increased adoption of personal Ai assistants like Muse; I believe we are set to see a significant step up in product and service discovery optimisation. Simply put - The ones with the best product and services will grow even faster. For neobanks: Sofi, Nu, Revolut.

Combine this with the fact that these banks already have been moving toward Ai first - and then now significantly reducing cost on tokens (unless they reinvest the spend and further improve or introduce new services).

Updated dashboard for Nu: https://t.co/VR0E3BcVIR

[Suddersan Thevarajah @suddersan]

My $NU Dashboard

Let me know what other metrics I should be tracking for this name. Increasingly building my first draft of the models using Claude - and end my workflow creating these type of dashboards as single source if truth for each name. The dashboard refreshes as I update my estimates in my excel model.

Love feedback on tips on how other investors have updated their workflows.

https://t.co/FJ70JSoidc

#equityresearch

[Suddersan Thevarajah @suddersan]

With Muse we are taking a significant step up in product and service discovery. Already been the case for the last few years for active ChatGPT and Claude users - but with Muse the diffusion is reaching another level.

-> concentration of demand in product and services (marketing won’t save your business anymore, at least until you can start to market the agents).

[Suddersan Thevarajah @suddersan]

I wonder if the introduction of JEV is a game changer for Harvey/Legora. Analyze huge amounts of text and check for true/false statments and feeding it back to an LLM to give you the output. Not my expertise, but I see a lot of marketer using it for SEO/AEO audits - before instructing an agent to act on the findings. Could a similar workflow ble repliacted for law? If so - Legora might have chosen the right option.

#Legora #Harvey

[Suddersan Thevarajah @suddersan]

Så hva kjøper man om man er long-only og ikke vil sitte på inverse indekser eller gull? USD som safe haven gir lite mening når hovedproblemet er en gjeldstynget amerikansk økonomi - eller finnes det argumenter for at den faktisk vil styrke seg under en større korreksjon?

Tenker payment services (naturlig hedge mot inflasjon med en % av hvor enn prisene er) og store forvaltere av kreditt kan være en grei plassering (APO). Massiv inflow av kapital for IG private credit - og når korreksjonen kommer, så vil man fortsette å investere i IG, så kanskje enda mer inflow (bare for andre typer fond). FRE og SRE står for nesten all inntekten (mao de tjener penger uansett om kreditten forfaller). De har direkte eksponering gjennom Athene - men er akkurat store nok til at de ikke bør gå under). Er ikke lenge til de får full tilgang til 401k kontoene i USA, blir det enda villere.

Litt den samme tanken som at det er bedre å eie banken enn å ha pengene i den (ikke lokale sparebanker). Lenge vært enig i at forsikringsselskaper er den beste typen banker å investere i, men tror nesten disse kredittfondene er enda bedre.

Skrev litt om de før Q2-26

https://t.co/ZNWlwTGxxl…

[Suddersan Thevarajah @suddersan]

@saranormous Talking about $BE ?

[Suddersan Thevarajah @suddersan]

Is it mostly suited for “simple” decision making? In the sense that there is little “grey area” or nuances. So that there always is an option with the probability of > .70?

Thinking it works best for grading, routing and classifying. Perhaps ideal for processes that require large volume of simple decisions; customer service, refunds, approvals/rejections etc.

@jasonlk @Benioff has emphasised how @salesforce data improved the deterministic output of the AI models. Reading this as most agents tasks and processes are largely repetitive and system 1 by nature (or at least can be modelled into system 1). Is JEV going to be significantly cost deflationary for Agentforce?

Tell me straight if I’m totally off.

BTW. Great work on 20VC.

[Suddersan Thevarajah @suddersan]

Perhaps a rookie question (not my expertise): will this mostly benefit companies that have adopted AI for repetetiv deterministic processes (system one)? I’m thinking customer service, refunds, approvals/rejection etc.

For a company like @salesforce token cost for their portfolio products should drop significantly, right? From what I understand most users create agents that handle specific tasks (repetitive manner). @Benioff has basically spent the last 3 months communicating how their data improves the deterministic output. Am I wrong to read that as most of their spend being system 1 related tokens?

If I’m right: SaaS - Re acceleration is on for real (for the companies that have adapted).

[Suddersan Thevarajah @suddersan]

Who’s to replace Siri?

Instinct or Muse?

#Instinct #Meta #Apple

[Suddersan Thevarajah @suddersan]

Solo investor looking for some feedback on my neobank thesis. #klarna #nubank
Love to discuss with someone who’s done the work on these names.

https://t.co/KDYqhI1H9O

[Suddersan Thevarajah @suddersan]

@drewfallon12 How does it compare to instinct? Guessing investors want to see some data on monetisation. Probably a good idea to see how much Watermelon is an improvement on the existing model family.