Fintech 🧠 Food - 1033: Open Banking Endgame

Hey everyone 👋, welcome to 1033 Food; hitting send so you have this before Money2020. Join the 34,840 others by clicking below, and to the regular readers, thank you. 🙏

Hey Fintech Nerds 👋

It’s here. 1033 passed.

The Rant Section 2 has the bullet point version of everything you need to know about 1033. Hitting send early so you have this before Money 2020.

Will this fix Open Banking in the US? It’s a start, but it’s not what everyone hoped for. Read the 📣 Rant to find out why.

Goldman profits are down 33%, dragged down by paying $2.2bn for Greensky to sell for $1.7bn barely a year later. No wonder the CEO isn’t doing high-profile DJ sets anymore. More analysis in 👀 Things to Know

Synapse is in the middle of a long, public, messy divorce. This ugliness is a microcosm of a market where banks are under pressure from regulators, and some distribution models are coming under pressure. More in 👀 Things to Know.

👉 PS. I’m not at Money 2020 in Vegas, so have fun if you’re going! You can hit reply to this email to get in touch (and I won’t charge you $1 to tweet at me, but Elon might)

👉 PPS. Need a listen for Money 2020? I had a chat about all things Future of Fintech with CFTE, and you can find it on YouTube here or Apple Podcasts here.

👉 PPPS. Need a listen for Money 2020? I joined Alex Johnson for Fintech Takes, and you find that here.

Here's this week's Brainfood in summary

📣 Rant: 1033 Open Banking: Endgame

💸 4 Fintech Companies:

1. Ethos AI - Finance AI Data Governance Platform

2. Verisoul - Detect fake users

3. Rainforest - Better embedded payments for platforms

4. Oneslipstream - One API for Credit Bureau's (UK)

👀 Things to Know:

1. Goldman profits by 33% led by consumer BNPL business write down

2. Synapse Public Spat with Evolve. + Lays off 40% of Staff 

📚 Good Read: Unbundling AI 

If your email client clips some of this newsletter, click below to see the rest

Weekly Rant 📣

Open Banking: Endgame

1033’s proposed rule is here.

Open Banking in the USA is entering its End Game. 

While the CFPB has proposed a rule for Open Banking, connectivity is breaking as some incumbents like Fidelity ban third parties from accessing data.

These two events are connected.

As is this Rant to previous work by Jason Mikula, who first reported the Akoya news, and Alex Johnson, who raised some fantastic questions.

The Fintech industry has been waiting for a 1033 rule to pass like kids wait for Christmas.

(And for the impatient among you, skip to section 2 for what the rule says and lessons learned from the UK)

The will they/won't they ever pass Open Banking Regulation saga has been running since the Dodd-Frank Act passed in July 2010. 

In that time, an entire industry has been born that took screenscraping, open banking, and APIs and created a revolution. This means instant account verification, cashflow underwriting, and all your accounts in a single dashboard.

It has been a huge success.

"Plaid for x" is a category of company.

But a new problem emerged. "Conversion."

Open Banking is like this classic Anchorman meme.

60% of the time, it works every time.

There are plenty of banks, brokerages, and credit unions for whom Open Banking will "break" regularly.

A consumer tries to connect their account and it doesn’t work.

There are arguments on all sides as to why.

But "Open Banking" is broken, and now it's entering its endgame.

Can 1033 fix this? Will the CFPB put Akoya back in its box? Or will all aggregators have to pay the banks much more than they do today?

Let's do this 🏊‍♀️

Thesis: While long-awaited, regulation alone won't resolve the fundamental incentives issues underlying the challenges in US open banking.

1. A brief history of open banking in the US

   1. Open Finance is more successful in the US than anywhere in the world but breaks too often

   2. Some incumbents view Open Banking as not secure

   3. Aggregators are forced to pursue bi-lateral arrangements to improve conversion

   4. Standards exist but without regulation they lack adoption

   5. Issues persist, as you can see in Fidelity, blocking access.

2. What does the rule say?

   1. The rule requires data providers to not charge for API access and more.

   2. The proposed rule is in comment until December, live before the election?

   3. The rule says nothing about brokerage and does not impose technical standards.

   4. The rule aims to create competition, remove junk fees and prevent pricing games by incumbents

   5. Implications: Banks can’t charge for API access and if PNC copied Fidelity it would be in breach of the rule if it passed

   6. How it differs from the UK. The UK required the banks to fund an independent standards body and provide a standard for payment initiation

3. Will it work? Lessons from the UK regulation

   1. Connectivity works near 100% of the time

   2. Account switching has not increased

   3. Zero upside for incumbents is an important lesson

4. Incentives matter

   1. There must be an economic upside for all participants.

   2. Security needs to be a widely shared burden.

   3. Can we create an industry utility to help here?

5. The Road Ahead

   1. The deadline will matter. Will there be an implementation timeline? 

   2. Regulation needs to pass before the next election to survive political wrangling. 

   3. If that political appointee changes, will we see a watering down of rules or a scrapping? A lot can happen between now and the end of 2024.

   4. Expect a scramble for standards.

   5. Expect howls of pain and lobbying. 

6. What should you do about it

   1. Lean in

   2. Own the gateway

   3. Get creative.

7. The US and Open Finance: Perfectly Imperfect

1. A brief history of US Open Banking

a) The US Open Banking market is a massive success, but it doesn't always work. Open Banking is more successful in the US than in Europe. The EU has around 5 million regular Open Banking users, while the US has closer to 80 million. Visa says 87% of consumers link their accounts to another service. 

The market is still growing.

In the US, connectivity breaks; there are no standards, but every product is available, from brokerage to Crypto to credit and debit. Cashflow underwriting is a default, and you have payroll API access. 

The free market has delivered something broken, imperfect, but incredible in its own right. 

Yet. The market has evolved in a way that almost everyone has issues with.

b) Some incumbents view Open Banking as "zero upside all downside," harming conversion. In the early 2010s, the US Open Banking ecosystem happened to and not with the incumbent banks and brokerages. One day, they woke up and found companies screen scraping data; they were on the hook for managing. If that data is leaked or stolen, that could result in fraud, money laundering, or cyber risk for the incumbent. Incumbents now had to bear the cost of managing data, which was being provided to potentially competitive offerings.

Viewed through this (myopic) lens. Open Banking can be seen as hostile and not secure. A rational actor would block such a connection. That's a bad end-user experience, but it is logical for "security."

c) Aggregators are forced to pursue bilateral arrangements to improve conversion. Most aggregators offer API-only access for security, but this requires a relationship with the underlying bank. Some aggregators have signed bilateral agreements to secure this relationship where the banks get paid for that data access. 

The aggregators would argue some of their biggest customers and beneficiaries are the incumbents. But without regulation, they are stuck between some incumbents who break connectivity, trying to support various standards and regulation development, and are "caught in the middle." 

d) Standards exist but aren't enforced. The Financial Data Exchange (FDX) has worked for years to build a common API standard for consumer account access. However, the lack of enforcement around a standard creates a market where incumbents can avoid implementing the APIs with an aggregator and then complain about "screen scraping" as an excuse to shut them out. 

e) Issues persist, as you can see in Fidelity, blocking access. If you try to log in to a Fidelity account using a Fintech app, chances are you will see this message: 

Fidelity breaks any connection that doesn't use its captive Open Banking platform, Akoya.

Why?

Alex Johnson summed up neatly with the image below. The market changes if banks can force companies to access data through one specific platform. 

(Alex is the master of these diagrams)

Without a universal way to connect to accounts and some accountability around that, Open Finance is a mess.

Many Fintech folks are pinning their hopes on 1033 to sort this out.

Will it?

2. What does the rule say and mean?

a) What does the rule say? 

It defines four roles.

• A data provider is any credit, debit, prepaid, or deposit account provider.

• An authorized 3rd party is any app or service with consumer consent to use data held at a data provider (e.g., a Fintech wallet or Neobank)

• A data aggregator who connects to multiple data providers through a single interface (e.g., Plaid, MX, Finicity)

• A standards body sets technical standards for APIs and represents the views of all ecosystem participants (e.g., FDX).

The key responsibilities of data providers: 

1. Must make all transactions for the past 12 months available, including the most recent card authorization.

2. Must include any fees (!!), yield, or rewards data available.

3. Make account verification like name, email, and address available (but not date of birth).

4. Must set up secure APIs with clear, standardized documentation

5. Must guarantee 99.5% uptime and not improperly restrict access

6. Cannot charge fees for API access or API calls (!!)

Key responsibilities for authorized 3rd parties

1. Make disclosures to consumers about how data is used and processed and their privacy (this is something the banks lobbied for)

2. Use standardized APIs and security models and maintain written policies

3. Do not use data for targetted advertising, data re-sale, or cross-selling (although cross-selling can be done if implicitly opted in)

Key responsibilities for data aggregators are interestingly pushed to the authorized 3rd party and the data provider. The fact that data is being processed must be disclosed.

The most interesting part for me was the standards body.

Responsibilities of the standards body

1. Balances decision-making power across all parties

2. Has an appeals process

3. The CFPB authorizes it.

b) When does it go live? The rule is at the proposal stage and is open for comment until December. There is no timeline for when it will go live. If I had to guess, I’d say before the election. The rule does say

• Six months from publication, any institution with more than $500bn in assets (or more than $10bn in revenue) must comply.

• 12 months from publication, institutions with $50bn to $500bn in assets must comply

• 30 months from publication, institutions with $805m to $50bn in assets must comply

c) What does the rule NOT say? 

1. Nothing about investments or brokerage. That could leave Fidelity unscathed by any new rule. But PNC? And those banks that play games with connectivity and latency? That should "just work." 

2. Nothing about how standards work. But if you must offer an API and it must give free and equal access, two questions appear. 

   1. How should that API work? FDX is a standards body with momentum; will it have to change its governance or processes?

   2. How will the CFPB's new rule be enforced? This is the big unknown. If Fintech companies report that one bank that always breaks connectivity, what happens next?

3. Nothing about payments. Open Banking initiated payments can be a “killer app” for open banking. It is possible without regulation. (For example, Plaid already works with Tesla to help consumers make car deposits.)

d) What are the objectives of the rule?

1. Create competition. The release talks about empowering consumers to walk away from bad service.

2. Consumer data control. Trying to avoid any company misusing consumer data for advertising.

3. Avoid pricing and availability games. The CFPB suggested large banks could play games with availability, latency, or price. The press release positions this as a “junk fee,” which is very on-brand for the Biden administration.

4. Move to “fair industry standard setting.” Implies that the CFPB views the way standards are set today as unfair.

e) What are some implications?

1. The rule doesn’t let banks charge for API access so have to take the hit on cost? This will further reinforce the perception that open banking has zero upside for incumbents and all downside. (See lessons from the UK in next section)

2. It can’t be used for identity use cases. If the date of birth can’t be shared, then this isn’t an identity “rail.” It will work for account verification and payment initiation, however.

3. If PNC tried to copy Fidelity, it would breach this rule. The timing of the Fidelity and Akoya move probably wasn’t an accident.

f) How does this differ from the UK market? 

1. The UK doesn’t have open banking it has open checking. That means no brokerage, credit and lending product aggregation.

2. The UK required open access to the A2A (account to account) payments initiative via open banking. This is not included in the CFPB rule.

3. The UK required large banks to fund an independent body (open banking implementation entity) that defined the standards. The CFPB is leaving that to the market.

3. Will it work? Lessons from the UK

a) The UK does have open banking that works 99%+ of the time. When I try to fund a Fintech wallet or pay my taxes with open banking, it just works. The UK does not have a conversion issue but a use case and adoption issue.

b) The UK hasn’t seen more account switching The UK's Open Banking market hasn't increased or decreased consumer switching. I'd argue the biggest competitive landscape shift came from creating a new path to banking licenses and giving central bank payment system access. 

c) The zero upside for incumbents is an important lesson. The push to force incumbents to open up checking account data also made it a zero-upside move for the banks. They were also forced to bear the cost of creating the standard. Is it creating competition where incumbents cannot compete? The UK will shortly widen the available products, but without some upside for the incumbents, I fear that too will suffer the lack of adoption the earlier products have. (And you know I'm a big fan of UK regulatory innovation.)

In fact, as I write this, a Mastercard exec has noted that India's UPI has made payments economically unviable for market participants. 

Why would any rational actor want this if it's all cost and no revenue?

Everyone needs ROI.

We must think about that from the start to get the most out of any Open Finance framework. 

4. Incentives matter

a) There must be an economic upside for all participants. You could argue that with a level of "data sharing" playing field, a sufficiently competent incumbent has every opportunity to compete. The reality is that today, most direct deposits, IRAs, and 401ks exist with an incumbent. They lose, and new entrants win. 

A revenue-sharing model could allow banks to receive a small transaction fee when consumers leverage open banking through third parties. This could be capped to prevent pricing abuses but would offset infrastructure costs. Aggregators are now placed to offer this as a carrot.

The future of Open Banking needs to be more like correspondent banking. Banks who compete for end customers partner with each other to make payments internationally. By becoming the customer of my competitor, I have a natural win/win when they send me volume.

If we had a simple pricing structure for APIs, why couldn't we view it this way?

b) Security needs to be a widely shared burden. The weakest link in any security chain is usually at the last mile. Small Fintech companies often lack a large enterprise's sophisticated cyber security budgets, but large enterprises are exposed to the problems of the smallest company.

We see this today in BaaS, and how one bad apple with compliance can ruin the bunch. Incumbents are judged by their worst-performing partners. This looks and feels like another product opportunity. 

What about a SOC2 for data sharing? 

c) What if we created shared funding for a "utility" beyond membership to a standards organization to help support smaller organizations? 

Utilities like Early Warning Services help banks share data about suspected fraud and are co-funded. 

This utility could set a pricing framework and serve as a platform for partnerships between incumbents, Fintech companies, and everyone else. It could help manage some of the burden of running infrastructure, support actors struggling with technology, and help guide standards outside the technology. 

If it worked for Early Warning Services and SWIFT, it could also work for Fintech and Open Banking. 

The important thing would be ownership structure. All of its banks own SWIFT. Inevitably, the large banks have a large voice, but the long tail has a substantial influence too.

5. The Road Ahead

Assuming 1033’s proposed rule completes comments before the end of the year. 

a) Regulation needs to pass before the next election. The Director of the CFPB is a political appointee and is widely regarded as having a great deal of discretion. Director Rohit Chopra announced almost a year ago, at Money 2020, a desire to pass Dodd-Frank 1033.

If that political appointee changes, will we see a watering down of rules or a scrapping? A lot can happen between now and the end of 2024.

b) Expect a scramble for standards. The front runner here is FDX, who would come into sharp focus. Given the CFPB's posture, currently pro-competition and pro-Fintech, they can become the de facto standard and secure wide buy-in. 

The FDX standard as of 5.0 includes much of what the CFPB calls for (two-way data sharing, consumer dashboards to revoke access, etc). 

But what about the economics?

c) Expect howls of pain and lobbying. The consumer banking lobby has already started tweeting. I have a smidge of sympathy. If large incumbents are expected to bear the cost of these APIs with zero fee upside, that scenario is worse than the status quo. Does that happen without pushback? Unlikely.

If the incumbents are not held to an economic limit on fees, the Fintech lobby will do everything it can too. 

But it doesn't have to be this way. 

6. What should you do?

a) It’s time for everyone to lean in. Regulation changes the chess board but not the pieces or the incentives. In any market, action trumps inaction.

• Bank, you gotta get your API infrastructure in place. I think it’s now inevitable this API access is required. The question is, how do you take advantage? Quit whining. Time to get on with the job. A mature incumbent market actor sees the win/wins and actively tries to develop them. 

• Aggregator. Time to lean into incentives. How do you help the banks? They can’t charge fees, but you can. The tables just turned to your advantage.

• Fintech companies. You have some important responsibilities on disclosure, security, and privacy, don’t f*ck it up.

b) Owning the consumer gateway becomes a battleground. We will have more reliable Open Banking data for verifying account ownership, fraud prevention, or payment initiation. Owning that “gateway” spot becomes prime real estate.

c) Get creative. Is there an “everyone but Fidelity” alternative to Akoya that the industry could build to bring brokerage to the table? Creating a captive Open Finance platform and shutting off 3rd parties, isn't it.

7. The US and Open Finance: Perfectly Imperfect

Regulation is coming, but it won't fix everything.

Unlocking the promise will take work on the incentive structures and finding win/wins.

The free market delivered more volume, use cases, and real-world impact than the regulated Open Banking we see elsewhere. But it also creaks at the seams, hampered by poor conversion and price wars.

Consumers lose in this market.

1033 needs to pass.

The CFPB Director will say things at Money 2020 when you read this. 

Like you, I’m waiting to see exactly what.

Have fun in Vegas if you're going.

And if you're not, drop me a reply and say hi 👋

ST.

4 Fintech Companies 💸

1. Ethos AI - Finance AI Data Governance Platform

Ethos helps regulated financial institutions ensure their AI, machine learning, and data comply with regulations. It helps data scientists and compliance teams identify risks, build policies, and create compliance documentation. This aims to remove the manual processes and build workflows for teams to manage data governance.

🤔 AI in consumer finance is hard. Data governance is manual. Financial services AI must be "explainable" and comply with regulations like fair lending practices. Today, this process is manual and involves creating 200-page documents to hand to compliance teams and regulators. This means just getting a model live can take months and is often out of reach for smaller companies. While "AI" tech is getting all of the hype, the plumbing we need to actually use the tech doesn't get discussed. Ethos competes with a strategy consultancy selling $billions for data governance "programs." We need this if we want nice things.

2. Verisoul - Detect fake users

Verisoul is a single API and SDK for detecting fake users and KYC. It bundles a collection of device, behavior, KYC, and telco/email data collection into a small line of embeddable code for its customers. Users can then build "no code" rules to block or screen users from onboarding. 

🤔 They're competing primarily with Captcha and Cloudflare bot detection, which aren't good enough. Bots and scams through social media and marketplaces are getting worse. That means friction for good users and losses for marketplaces. What I like about Verisoul is their homepage is incredibly well packaged. Captcha is so out of date we need better solutions for sure. 

3. Rainforest - Better embedded payments for platforms

Rainforest provides cards, ACH, RTP, and more payment types for e-commerce merchants, marketplaces, and platforms. The service has pre-built components to manage payments, reporting, and chargebacks. What makes Rainforest aim to bear the risk and compliance burden for its clients like merchant onboarding.

🤔 Atlanta payments nerds do embedded payments as a service. Rainforest differs from companies like Stripe, which focuses on "DIY" solutions for platforms. Rainforest aims to be "the payments department you hire as an API." It also gives data portability and aims to help with vertical-specific challenges like healthcare or logistics. Atlanta has a long history of payment companies, and the founders have experienced the bugbears of the new generation of PayFacs, whose "all in one" lock-in isn't for everyone. 

4. Oneslipstream - One API for Credit Bureau's (UK)

Historically, CRAs are a dumpster fire for integration. Oneslipstream aims to solve this with a single, developer-friendly API for all of the major CRAs in the United Kingdom. It provides real-time and event-driven data reports.

🤔 All risk problems are data problems, and the CRA data is, at best, not great. A report by the FCA found that CRAs often had inaccurate data, and the major CRAs might even disagree with each other. This is the foundation against which we manage risks like fraud, AML, and how we score consumers for lending. If Oneslipstream can at least aggregate that data, that's a positive step. My mind wonders about the possibility of them enriching that data, too, to make it easier to use and combining it with other sources, just like the ecosystem on top of Plaid and Open Banking Today.

Things to know 👀

1. Goldman profits by 33% led by consumer BNPL business write down

Goldman Sachs reports a 33% drop in profit and a 1% drop in revenue, as their CEO promises to stop doing high-profile DJ sets. Bond trading revenue fell by 6%, and the bank has suffered as high interest rates have discouraged companies from heading for an IPO. Unwinding retail operations like Greensky have caused losses, noting a $508m write-down tied to the Greensky business. 

🤔 The goal of going consumer was to diversify from their cyclical investment banking business. Trading and advisory accounts for 2/3rds of the group’s revenue. For Goldman, the objective of entering consumer checking, lending, BNPL, and partnerships like Apple Card was to diversify away from Investment Banking and trading. They went hard into nearly every consumer line of business building, buying and partnering and for a while, they were the case study of the attacker brand.

🤔 Going too hard too fast created losses when the downturn in investment banking happened. The partnerships have had rumors of regulatory issues; the BNPL acquisition proved expensive and loss-making. Goldman’s lending business is down, and the Greensky acquisition to sale has proven costly. Credit Card lending is still growing, but have a high charge off rates and faces regulatory hurdles. These are all problems that can be managed when the investment bank and group is doing well/. They get exposed when the cycle turns.

🤔 Contrast this with Morgan Stanley. MS has focussed primarily on private banking and wealth management over the past decade. This is a bright spot in an otherwise difficult banking market. Perhaps less is more for Goldman in the future.

🤔 Can they exit Apple card fast enough? There are now widely reported rumors Goldman wants to exit the Apple Card deal. As I wrote last week, Apple will be a bank in all but name but hasn’t enjoyed its partnership experience so far. The feeling is mutual. 

🤔 The IPO market is closed, the consumer business is hyper-competitive and Goldman needs a win. The greensky sale will show up in results soon which will be a temporary boost, but Goldman can’t wait for the wind to change. They do have a private banking and wealth practice and a burgeoning transaction banking practice. These could become bright spots in time. 

🤔 David Solomon no longer being a DJ makes sense. This was fun PR for Goldman and a great way to stand out from other less cool investment banks, but a string of terrible results makes his job look less secure. The bank needs to regain credibility with investors, and appearing focused will be important.

2. Synapse Public Spat with Evolve. + Lays off 40% of Staff 

This follows a tip from Jason Mikula that B2B Neobank Mercury would not renew with Baas Platform Synapse. Synapse claims it has issues with Evolve in a medium post, Evolve claims it had notified of its intent to terminate with Synapse in August of 2022 with a goal of “prioritizing direct relationships with Fintech companies. Synapse announced it has let 86 people go following an 18% reduction in June. (This number accounts for domestic and not international or contract workers).

🤔 Evolve is part of a trend for banks to go more direct. The OCC issued an enforcement action against Evolve in May of 2022. It followed this with guidance on “3rd party” risk management a year later. This has caused banks to reconsider the role BaaS providers can and should play in compliance. Ultimately banks are choosing to take this more in house. This means adding new processes, skills and providers to give them more direct control.

🤔 Synapse was a pioneer. Synapse was founded in 2014, long before many modern peers. It arguably caught the wave of Neobanks and non-banks who didn't want to integrate directly with a payment processor to offer debit (and now) credit cards. Since then, BaaS has become a category with multiple approaches and routes to market. Today, even the mighty Stripe and Adyen offer a similar service.

🤔 Its model of compliance integration was a double-edged sword. Synapse would take much of the compliance and risk management operations away from the Fintech company and offer those as a service. That reduced the time to market and complexity for its customers. However, clients in finance have unique risk profiles, and rarely does one compliance department fit all. 

🤔 The Crypto bull run also compounded this issue. Synapse didn't want to cede the Crypto opportunity to other platforms and found itself placed to offer this product. The risk surface area multiplied as companies began integrating debit cards with Crypto. Now, you had one provider with a concentration of risky clients.

🤔 Every company will be a Fintech company, but only if they get compliance right. Financial services revenue is a multiplier for any business, but it comes at the cost of risk management. No provider can ever fully manage that as an outsourcer, but the best providers can make it much more manageable. 

🤔 Horizontal integration is hard. To some extent, you see this issue with Stripe's struggles in enterprise, whose natural inclination is to bundle Payments, Fraud, Treasury, Tokenization, and Billing into one package. The small company gets everything they need and efficiencies. The big company lacks control and wants to unbundle (or at least have optionality). Being modular, partnering under the hood, and more of a "school of fish" than "build it all ourselves" might be the way out here (which we see in Treasury Prime, Unit, etc).

Good Reads 📚

1. Unbundling AI 

Are LLMs (Large Language Models) the beginning of a new platform shift to chat as the interface? We have not had a real shift in how humans interact with computers since Xerox PARC and the GUI. We had a false dawn with Siri and Alexa, but LLMs are good enough, mostly, but have some kinks to work out. It has two problems: 1) A science problem, it gives probable answers, not deterministic answers, and 2) The product problem, how do you present and package that uncertainty? It appears certain, but it's often not accurate. 

The LLM "prompt" box is also a blank canvas. Your output quality depends on the prompted input(s) quality. The interface feels like a game of battleships; you input a prompt and hope it does what you want. If not, try again. Just as every Excel function became a software company, you can use ChatGPT to do anything, but having a specific tool might be better.

🤔 Solving LLMs interface problems with "agents": What Meta is doing with AI personas is an interesting product approach. They recently demonstrated a personal assistant persona and a chef persona. They're not the first to try this; Siri and Alexa were an attempt, and pre-iPhone, there were many. But now the agents might actually be good enough. 

🤔 Every SaaS company could be an "agent." Imagine a "lawyer" or "accountant" AI persona rather than a prompt. There's also a world where you hire a "software engineer" agent and interact with them in a virtual 3D space. Open AI just hired Jonny Ive to solve the user interface problem. The reality is we just don't know yet.

🤔 The interface feels like early computers. It's command-line based, and now it's quickly learning about graphics or voice. But what is the optimal way to interact with this new type of computer? Focussed SaaS tools? Voice? Video? Metaverse shenanigans? 

🤔 Today's front-running models might not be dominant. While Antropic and OpenAI have a clear lead, cheaper open-source models punch above their weight massively. 

That's all, folks. 👋

Remember, if you're enjoying this content, please do tell all your fintech friends to check it out and hit the subscribe button :)

Disclosures: (1) All content and views expressed here are the authors' personal opinions and do not reflect the views of any of their employers or employees. (2) All companies or assets mentioned by the author in which the author has a personal and/or financial interest are denoted with a * (3) Any companies mentioned in Rants are top of mind and used for illustrative purposes only. (4) I'm not an expert at everything you read here. Some of it is me thinking out loud and learning as I go; please don't take it as gospel—strong opinions, weakly held.