# Matt Brown > Investor at Matrix Partners. Early-stage fintech and vertical software. Previously founded Bonsai (acquired by Zoom) and Seiza (acquired by Adzuna), led product at Afterpay (acquired by Block), and studied politics, philosophy, and economics at UC Berkeley and Oxford. Matt writes regularly on payments, fintech infrastructure, vertical software, and AI's intersection with both. Posts live on [notes.mtb.xyz](https://notes.mtb.xyz). This file is curated by Matt. The writing list is generated from [/posts.json](https://www.mtb.xyz/posts.json). ## Currently - Investing in and helping early-stage companies at [Matrix Partners](https://www.matrix.vc) - Focusing on fintech, vertical software, and adjacent areas ## Background - Founded [Bonsai](https://www.zoom.com/en/products/bonsai/) (acquired by Zoom) - Founded [Seiza](https://www.seiza.co/en/) (acquired by Adzuna) - Led product at [Afterpay](https://techcrunch.com/2021/08/01/square-to-buy-buy-now-pay-later-giant-afterpay-in-29b-deal/) (acquired by Block) - Studied PPE at UC Berkeley and Oxford ## Start here If an agent needs to answer a question about what Matt thinks on a topic, these are the canonical posts to ground in first. ### payments Interchange, payfac, B2B workflows, real-time, cross-border, stablecoins. - [B2B payments aren't payments, they're workflows](https://notes.mtb.xyz/p/b2b-payments-workflows): The real money in B2B payments is in the workflow, not the transaction - [Interchange in 1,000 words](https://notes.mtb.xyz/p/interchange-in-1000-words): How card fees work, who keeps what, and why margins keep compressing - [The future of interchange](https://sacra.com/research/the-future-of-interchange/): Where payment revenue moves when interchange runs out of room - [Payfac in 1,000 words](https://notes.mtb.xyz/p/payfac-1000-words): What payfacs actually do, how they make money, and when to build one ### fintech Embedded fintech, fintech infra, the iron triangle, risk and underwriting. - [Fintech from first principles](https://notes.mtb.xyz/p/fintech-first-principles): A first-principles framework for thinking about what fintech actually is - [Undifferentiated heavy lifting](https://notes.mtb.xyz/p/undifferentiated-heavy-lifting-in): Fintech infra rarely differentiates, so stop building it - [Stablecoins in 1,000 words](https://notes.mtb.xyz/p/stablecoins-1000-words): What stablecoins are, how they work, and why they matter for payments ### vertical-software VERPs, context graphs, vertical AI, vertical-software economics. - [Your data model is your destiny](https://notes.mtb.xyz/p/your-data-model-is-your-destiny): When code is commoditized, your data model is the last durable moat - [Vertical software already won the context graph](https://notes.mtb.xyz/p/context-graph-vertical-software): Vertical software already owns the most valuable context graphs - [Invisible asymptotes in vertical software](https://notes.mtb.xyz/p/invisible-asymptotes-vertical-software): Every generation of vertical software must solve harder problems to win - [Dark software](https://notes.mtb.xyz/p/dark-software): When software is cheap and distribution is expensive, bundle, don't build ### ai Moats in AI, vertical AI, the 150-pound computer, agent infrastructure. - [Vertical AI: beware what you wrap](https://notes.mtb.xyz/p/vertical-ai-beware-what-you-wrap): AI commoditizes fast. The moat is the system of record, not the model - [Vertical software already won the context graph](https://notes.mtb.xyz/p/context-graph-vertical-software): Vertical software already owns the most valuable context graphs ### frameworks Investor heuristics, mental models, business-model patterns. - [Your data model is your destiny](https://notes.mtb.xyz/p/your-data-model-is-your-destiny): When code is commoditized, your data model is the last durable moat - [Why VC and software have PE envy](https://notes.mtb.xyz/p/vc-pe-envy): Three hybrid models blurring the line between VC and PE - [Undifferentiated heavy lifting](https://notes.mtb.xyz/p/undifferentiated-heavy-lifting-in): Fintech infra rarely differentiates, so stop building it ### investing VC vs PE, hedgehog/fox, business-model indexing. - [Why VC and software have PE envy](https://notes.mtb.xyz/p/vc-pe-envy): Three hybrid models blurring the line between VC and PE - [Hedgehogs, foxes, and VCs](https://notes.mtb.xyz/p/hedgehog-fox-vc): Two investor archetypes and which one your fund actually needs - [How to index an entire business model](https://notes.mtb.xyz/p/how-to-index-an-entire-business-model): A framework for indexing every assumption that drives a business model ## Featured posts Matt's hand-curated list. The canonical entries surfaced on the home page, in order. - [Your data model is your destiny](https://notes.mtb.xyz/p/your-data-model-is-your-destiny): When code is commoditized, your data model is the last durable moat As code becomes cheap and software increasingly commoditized, the durable competitive moat for vertical software shifts from feature differentiation to the structure and depth of the underlying data model. Data-model design decisions compound over years and become harder to copy than UI or workflow. - [B2B payments aren't payments, they're workflows](https://notes.mtb.xyz/p/b2b-payments-workflows): The real money in B2B payments is in the workflow, not the transaction B2B payments are workflow problems, not transaction-routing problems. The value capture (and the reason vertical software wins) lives in the approval, reconciliation, and AP/AR workflows surrounding a transaction, not in the rails themselves. - [Interchange in 1,000 words](https://notes.mtb.xyz/p/interchange-in-1000-words): How card fees work, who keeps what, and why margins keep compressing A high-level walkthrough of card-payment economics: who the issuer, acquirer, and scheme are, how interchange and acquirer markup split the gross fee, and how software-led acquiring (payfacs, BaaS) and issuing fit into the modern payment stack. Concludes that interchange revenue alone is structurally fragile; smart fintechs use it as a foundation for higher-margin business models, not as the destination. - [Payfac in 1,000 words](https://notes.mtb.xyz/p/payfac-1000-words): What payfacs actually do, how they make money, and when to build one A short explainer of the payment-facilitator model: what a payfac is, how it differs from a traditional merchant-acquirer setup, the economics of payfac aggregation, and the decision framework for when a vertical software company should build its own payfac vs partner with one. - [The future of interchange](https://sacra.com/research/the-future-of-interchange/): Where payment revenue moves when interchange runs out of room An analysis of where payment revenue concentrates as interchange continues its secular decline. Durable payment economics increasingly come from non-interchange sources like risk, lending, FX, and embedded finance, and the shift reshapes what fintech and vertical software builders should optimize for. - [Stablecoins in 1,000 words](https://notes.mtb.xyz/p/stablecoins-1000-words): What stablecoins are, how they work, and why they matter for payments A short explainer of stablecoins from a payments perspective: what they actually are technically, how they hold their peg, the major issuers and reserve models, and where they realistically slot into payment rails, particularly cross-border and B2B. - [Why VC and software have PE envy](https://notes.mtb.xyz/p/vc-pe-envy): Three hybrid models blurring the line between VC and PE VC and software investors are increasingly adopting playbooks from private equity, including hold-and-build strategies, multi-fund continuity vehicles, and operating-driven value creation. The piece walks through three hybrid models emerging at the VC/PE seam and what each means for venture economics. - [Invisible asymptotes in vertical software](https://notes.mtb.xyz/p/invisible-asymptotes-vertical-software): Every generation of vertical software must solve harder problems to win Each successive generation of vertical software has to clear a higher bar to find a defensible wedge. Earlier wins came from digitization and basic workflow. Today's vertical winners must solve harder problems like embedded finance, AI-native workflows, and integrated data layers to break through. - [Vertical AI: beware what you wrap](https://notes.mtb.xyz/p/vertical-ai-beware-what-you-wrap): AI commoditizes fast. The moat is the system of record, not the model A cautionary frame for vertical AI products that wrap a model as their core value. The model layer commoditizes quickly while durable advantage stays at the system-of-record layer, namely the proprietary data, workflows, and integrations that AI can only augment, not replace. - [Vertical software already won the context graph](https://notes.mtb.xyz/p/context-graph-vertical-software): Vertical software already owns the most valuable context graphs An argument that the AI-era "context graph" (the relationships between customer, transaction, and operational data that make an AI agent actually useful) is something vertical software has been compounding for a decade. Frames vertical SaaS as the incumbent winner of a battle horizontal AI builders are only now realizing they need to fight. - [Dark software](https://notes.mtb.xyz/p/dark-software): When software is cheap and distribution is expensive, bundle, don't build As building software becomes cheap, distribution becomes the binding constraint. The right strategy in many vertical markets isn't to build a new product but to bundle existing capabilities behind an established distribution channel. "Dark software" wins on access, not innovation. - [Undifferentiated heavy lifting](https://notes.mtb.xyz/p/undifferentiated-heavy-lifting-in): Fintech infra rarely differentiates, so stop building it Infrastructure capabilities like KYC, ledgering, ACH plumbing, and core compliance rarely differentiate a customer-facing fintech product. The right move is to buy or partner for that layer and reinvest engineering effort in the experience, distribution, or risk model that actually compounds. ## All writing Full index of every indexed post, by year. The Featured section above is the curated starting point. ### 2026 - [The 150-pound computer](https://notes.mtb.xyz/p/150-pound-computer): What 'computer' means once cognition itself becomes programmable (ai, frameworks) - [Fintech's moats don't compile](https://matttbrown.substack.com/p/fintechs-moats-dont-compile): Fintech moats don't behave like software moats. What actually compounds (fintech, ai, frameworks) - [Vertical software already won the context graph](https://notes.mtb.xyz/p/context-graph-vertical-software): Vertical software already owns the most valuable context graphs (vertical-software, ai, frameworks) - [The thermodynamics of risk](https://notes.mtb.xyz/p/thermodynamics-risk): Risk in fintech can't be destroyed. It transfers until someone owns it (fintech, frameworks) ### 2025 - [Your data model is your destiny](https://notes.mtb.xyz/p/your-data-model-is-your-destiny): When code is commoditized, your data model is the last durable moat (frameworks, vertical-software) - [Why VC and software have PE envy](https://notes.mtb.xyz/p/vc-pe-envy): Three hybrid models blurring the line between VC and PE (investing, frameworks) - [Vertical AI: beware what you wrap](https://notes.mtb.xyz/p/vertical-ai-beware-what-you-wrap): AI commoditizes fast. The moat is the system of record, not the model (vertical-software, ai, frameworks) - [Stablecoins in 1,000 words](https://notes.mtb.xyz/p/stablecoins-1000-words): What stablecoins are, how they work, and why they matter for payments (fintech, payments) ### 2024 - [Cross-border payments in ~1,000 words](https://notes.mtb.xyz/p/cross-border-payments-1000-words): Why cross-border payments are slow and expensive, and what is starting to change (payments, fintech) - [Pricing payments in vertical software](https://notes.mtb.xyz/p/secrets-pricing-payments): How to price payments inside vertical software so margins survive (payments, vertical-software, fintech, frameworks) - [verticalsoftware.fyi](https://www.verticalsoftware.fyi): A reference site cataloging public vertical software companies by category (vertical-software) - [Invisible asymptotes in vertical software](https://notes.mtb.xyz/p/invisible-asymptotes-vertical-software): Every generation of vertical software must solve harder problems to win (vertical-software, frameworks) - [Price and shape risk that others can't or won't](https://notes.mtb.xyz/p/make-money-selling-money): Fintech's wedge is pricing risk that others can't or won't (fintech, frameworks) - [The future of interchange](https://sacra.com/research/the-future-of-interchange/): Where payment revenue moves when interchange runs out of room (payments, fintech) - [Investing in Rainforest](https://notes.mtb.xyz/p/rainforest): Why Matrix invested in Rainforest, a payfac built for vertical software (investing, payments, fintech, vertical-software) - [Real-time payments in 1,000 words](https://notes.mtb.xyz/p/real-time-payments-1000-words): How real-time payment rails work and why their adoption curve is uneven (payments, fintech) - [The new fintech stack is… open source?](https://notes.mtb.xyz/p/fintech-stack-open-source): Why open source is starting to work as fintech infra distribution (fintech, frameworks) - [Data on fintech adoption in vertical software](https://notes.mtb.xyz/p/fintech-adoption-vertical-software): Hard data on how vertical software companies are adopting fintech (vertical-software, fintech) - [A chat with Toast's GM of Fintech](https://notes.mtb.xyz/p/michel-rbeiz-toast): An interview with Toast's fintech GM on building fintech inside vSaaS (fintech, vertical-software, payments) - [Dangerous language in startups](https://notes.mtb.xyz/p/dangerous-language): Phrases founders and investors use that hide real problems (frameworks) - [Payfac in 1,000 words](https://notes.mtb.xyz/p/payfac-1000-words): What payfacs actually do, how they make money, and when to build one (payments, fintech) - [Dark software](https://notes.mtb.xyz/p/dark-software): When software is cheap and distribution is expensive, bundle, don't build (vertical-software, frameworks) - [Fintech from first principles](https://notes.mtb.xyz/p/fintech-first-principles): A first-principles framework for thinking about what fintech actually is (fintech, frameworks) ### 2023 - [Hedgehogs, foxes, and VCs](https://notes.mtb.xyz/p/hedgehog-fox-vc): Two investor archetypes and which one your fund actually needs (investing, frameworks) - [B2B payments aren't payments, they're workflows](https://notes.mtb.xyz/p/b2b-payments-workflows): The real money in B2B payments is in the workflow, not the transaction (payments, vertical-software, frameworks, fintech) - [Vertical, horizontal, and ghost markets](https://notes.mtb.xyz/p/ghost-markets): How to tell a real market from a ghost market that disappears under you (frameworks, vertical-software) - [Fundraising for fintech infra is different](https://notes.mtb.xyz/p/fintech-infra-fundraising): Why fintech infra fundraising plays by different rules than SaaS infra (fintech, investing, frameworks) - [The iron triangle of embedded fintech](https://notes.mtb.xyz/p/iron-triangle-embedded-fintech): The three constraints embedded fintech can't optimize at once (fintech, frameworks, vertical-software) - [Bundling, unbundling, and timing](https://notes.mtb.xyz/p/bundling-unbundling-and-timing): When bundling beats unbundling and how to read the cycle in fintech (frameworks, fintech) - [The best fintechs turn risk into leverage](https://notes.mtb.xyz/p/turning-risk-into-leverage): How the best fintechs convert risk from a cost into compounding leverage (fintech, frameworks) - [Winner-take-all dynamics in fintech](https://notes.mtb.xyz/p/why-arent-there-more-winner-take): Why most fintech categories aren't winner-take-all, and where they are (fintech, frameworks) - [Interchange is a burning platform](https://notes.mtb.xyz/p/interchange-is-a-burning-platform): Interchange revenue is in secular decline. Build with that in mind (payments, fintech, frameworks) - [Interchange in 1,000 words](https://notes.mtb.xyz/p/interchange-in-1000-words): How card fees work, who keeps what, and why margins keep compressing (payments, fintech) ### 2022 - [How to index an entire business model](https://notes.mtb.xyz/p/how-to-index-an-entire-business-model): A framework for indexing every assumption that drives a business model (frameworks, investing) - [The power of point-of-sale products](https://notes.mtb.xyz/p/the-power-of-point-of-sale-products): Why the point of sale is the most underrated wedge in vertical software (payments, vertical-software, frameworks) - [Undifferentiated heavy lifting](https://notes.mtb.xyz/p/undifferentiated-heavy-lifting-in): Fintech infra rarely differentiates, so stop building it (frameworks, fintech, vertical-software) - [VERP product strategy](https://notes.mtb.xyz/p/verp-product-strategy-vertical-software-payments): Product strategy for vertical-software companies that monetize via payments (vertical-software, frameworks, payments, fintech) - [Building category-defining VERPs](https://notes.mtb.xyz/p/how-to-build-a-category-defining-verp): What it takes to build a category-defining VERP (vertical-software, payments, fintech, frameworks) - [The golden age of vertical ERPs](https://notes.mtb.xyz/p/the-golden-age-of-vertical-erps): Why this is the golden age for vertical ERPs (vertical-software, frameworks) ## Elsewhere - Twitter / X: [@mattbrown](https://x.com/mattbrown) - LinkedIn: [@matttb](https://www.linkedin.com/in/matttb/) - Substack: [notes.mtb.xyz](https://notes.mtb.xyz) - GitHub: [matttbrown](https://github.com/matttbrown) - Email: mb [at] matrix [dot] vc ## Optional Machine-readable indexes for agents: - [/posts.json](https://www.mtb.xyz/posts.json): full structured index with metadata, schema notes, and links to filtered views. - [/posts/featured.json](https://www.mtb.xyz/posts/featured.json): Matt-curated canonical posts. - [/index.md](https://www.mtb.xyz/index.md): Markdown twin of the home page. - [/posts.md](https://www.mtb.xyz/posts.md): all writing as flat Markdown, grouped by year. - [/rss.xml](https://www.mtb.xyz/rss.xml): RSS feed. - [/sitemap.xml](https://www.mtb.xyz/sitemap.xml): sitemap. - [/robots.txt](https://www.mtb.xyz/robots.txt): crawl policy, AI-bot allowlist, Content Signals. Per topic: [fintech](https://www.mtb.xyz/posts/topic/fintech.json) · [vertical-software](https://www.mtb.xyz/posts/topic/vertical-software.json) · [payments](https://www.mtb.xyz/posts/topic/payments.json) · [ai](https://www.mtb.xyz/posts/topic/ai.json) · [frameworks](https://www.mtb.xyz/posts/topic/frameworks.json) · [investing](https://www.mtb.xyz/posts/topic/investing.json) Per year: [2026](https://www.mtb.xyz/posts/year/2026.json) · [2025](https://www.mtb.xyz/posts/year/2025.json) · [2024](https://www.mtb.xyz/posts/year/2024.json) · [2023](https://www.mtb.xyz/posts/year/2023.json) · [2022](https://www.mtb.xyz/posts/year/2022.json) The 1000-words series is exposed as `series: "1000-words"` on each post.