When Brokers Become the Marketplace
Using AI to automate transactions allows brokers to benefit from network effects and economies of scale, creating new marketplaces
AI brokers is one of the most under-discussed opportunities in startups today. AI brokers can create a new set of marketplaces. This is because broker markets consolidate when transactions can be fully automated. Think of how Google became the world’s largest ad broker or how online stockbrokers like E*Trade and Robinhood took over retail trading.
In this blog, I walk through 1) the reasons brokers exist, 2) how AI will transform them into marketplaces, and 3) how startups can capitalize on this new opportunity.
What is a broker? Why do they exist?
Brokers are third parties that manage a transaction between a buyer and seller. They help the buyer and seller by solving these problems:
Market Access: Connecting parties in fragmented or opaque markets. For example, insurance carriers would rather rely on brokers than staff a massive salesforce across the country.
Decision Support: Educating clients and helping them through high complexity or high risk transactions. For example, real estate agents manage a multi-step process and help clients mitigate legal or financial risks such as mortgage terms and title issues.
Price Discovery: Brokers help buyers and sellers agree on prices. They price in different ways depending on incentives. Some brokers make money on a percentage of sale price, so they bias toward sellers. Sometimes, they make money on spread.
Until now, only humans could solve these problems, which has kept these markets fragmented and hard to disrupt. Tech solutions for industries like loans, art, and real estate often end up working just like traditional brokers. They have not scaled enough to take over their respective markets.
Where are brokers most common?
Broker-mediated markets are characterized by low frequency, high complexity transactions in fragmented markets. While 2x2 above has plenty of exceptions, it is a helpful framework.
Marketplaces tend to be better business models than brokers: lower friction on both ends drives up repeat behavior and they benefit from network effects and economies of scale. Simple marketplaces have low transaction complexity and high frequency, which results in retentive customers and efficient unit economics. For complex transactions, split marketplaces may occupy two distinct roles: e.g., Robinhood aggregates consumer supply while Citadel Securities prices trades and provides liquidity.
Case Study: How Robinhood and Citadel replaced stock brokers
The old way of buying stocks was slow and manual. You had to call a human stockbroker, who would then yell orders on the trading floor or call other traders. Everything changed when computers entered the picture. Startups like E*Trade and Robinhood let people buy stocks online directly. They provided data on stocks so users could make informed choices and could offer faster trades than brokers due to the automation. As these platforms grew, they became defensible due to brand and liquidity. Today, 99% of retail trades happen online.
Simultaneously, quant funds like Knight Trading and Citadel Securities could price trades faster and better than humans. They paid the online brokers to send them trades, which pushed human traders out of the market. The quant funds colocated servers with stock exchanges, built private relationships with banks, and splurged on tech talent to solidify their market leadership. Today, just two companies - Citadel Securities and Virtu handle 60% of US retail stock trades.
In short, online brokerages and quant funds:
Simplified market access for consumers via electronic trading
Used online applications for education and decision support with up-to-date information
Applied quant algorithms and big data for improved price discovery and liquidity
The market consolidated from hundreds of thousands of individual stockbrokers to a few online brokerages and market makers.
How brokers can use LLMs to win and scale
LLMs have the capacity to improve market access, decision support, and price discovery for brokers. Applying LLMs to these problems means brokerages are no longer as limited by the need to hire and train personnel, allowing AI brokers to scale up far larger than predecessors.
Identify and engage leads faster: Processing unstructured data to target prospects efficiently, then using AI tools to qualify those leads (e.g., Goliath spotting potential home sellers before they are on-market).
Automating decision support: AI agents can flexibly handle complex queries and coach customers through complex sales, as shown by customer support agent companies (e.g., Decagon)
Ingesting more data for pricing: LLMs excel at parsing unstructured data, which can then be input to traditional statistical models. Those models help brokers price better and win more business from competitors. (e.g., Garner processes medical claims and uses quant algorithms to find high quality, low price physicians).
As brokers scale and digitally automate the buying process, they begin to resemble marketplaces, which benefit from economies of scale, network effects, and brand.
What tech-enabled brokers need to watch out for?
“Okay Jared, this is great, I’m ready to start building so I can disrupt those old brokers.”
Woah there, hold your horses! If it was that easy, there wouldn’t be any brokers left. Tech enabled brokers have to overcome a lack of trust, defensibility, and relationships.
Lack of Trust: People use brokers because they trust them, especially for big, risky decisions. It's easier to let an expert with more experience handle these choices. For AI brokers to replace humans, they need to accomplish the following:
Simplify: Make transaction terms and trade-offs short & sweet so that they're easy to understand.
Human backup: Provide human backup when needed so customers feel like they’re in safe hands.
Make it convenient: People will follow the path of least resistance, so being first, minimizing the number of steps in a transaction, and being the lowest price are all huge advantages that can overcome lack of brand.
Lack of Defensibility. Defensibility is an initial challenge, as LLM tools become commoditized. Tech-enabled brokers will face competition from other brokers and potentially sellers. They can secure their position by leveraging unique advantages that are hard to replicate:
Exclusive Partnerships: Access key data streams to enhance pricing and lead acquisition.
Trusted Brands: Build or acquire a strong brand and scale it using LLMs.
Pricing Power: Leverage economies of scale to offer lower prices and capture more business.
Local Expertise: It’s hard to build local insights into a product that scales nationwide, but using local insights to create a better product experience makes it harder for copies to follow.
Lack of Relationships. This is the trickiest problem for tech-enabled brokers. For some industries, brokers provide key relationships. Investment banking and talent agents are examples. Tech-enabled brokers will need to be creative to find a way in:
Start with lead gen: Generate fresh leads and send them to incumbent brokers to spark competition. Once you have quality leads, build relationships to eventually bypass incumbents.
Target an underserved niche: Focus on overlooked segments to build a strong brand and customer loyalty.
Acquihire a traditional broker: Acquire an experienced broker to quickly establish relationships. When doing this, be sure to diversify quickly.
Concluding thoughts
In the past, broker markets were spread out and messy, making it hard for tech companies to reach scale. But now AI and LLMs can handle many broker tasks automatically. Just as marketplaces made travel and rideshare and stock purchasing a breeze, tech-enabled brokers will bring simplicity to purchasing homes, insurance, and more. The only question is which markets will transform first, and who will lead that change. Maybe you?
A few asides!
Shout outs
Big thank you to Athena Kan, Max Yuan, David Song, Nihar Bobba, and many others for their helpful feedback.
Cool markets to potentially build in
Wholesale trade brokers:
Market Size: $775B market
Companies: 90K businesses averaging $8.6M in annual revenue
Fragmentation: top-2 have only ~7% market share.
Segments: food products, automotive, electrical equipment, industrial equipment.
Investment banking & securities intermediation:
Market Size: $393B market
Companies: 31K businesses averaging $13M in annual revenue
Fragmentation: top-3 have ~28% market share
Segments: trading, derivatives, debt, advisory, equity
Real estate brokers:
Market Size: $235B market
Companies: 1M businesses averaging $235K in annual revenue
Fragmentation: largest has only ~8% market share.
Segments: residential, commercial
Insurance brokers:
Market Size: $216B market
Companies: 422K businesses averaging $500K in annual revenue
Fragmentation: top-3 have only ~13% market share.
Segments: commercial P&C, consumer P&C, health
Freight brokers:
Market Size: $117B market
Companies: 77K businesses averaging $1.5M in annual revenue
Fragmentation: top-3 have only ~7% market share
Segments: domestic, international
Media brokers:
Market Size: $37B market
Companies: 7K businesses averaging $6M in annual revenue
Fragmentation: top-2 have only ~2% market share
Segments: radio, TV, magazines, newspapers
Loan brokers:
Market Size: $27B market
Companies: 21K businesses averaging $1.3M in annual revenue
Fragmentation: largest has only ~4% market share
Segments: residential mortgages, commercial mortgages
Commodity contracts intermediation
Market Size: $22B market
Companies: 5K businesses averaging $4.5M in annual revenue
Fragmentation: largest is not specified
Talent agents:
Market Size: $15B market
Companies: 38K businesses averaging $400K in annual revenue
Fragmentation: top-2 have only 16% market share
Art dealers:
Market Size: $14B market
Companies: 22K businesses averaging $600K in annual revenue
Fragmentation: the largest has only ~12% market share