The Checkout Wall Facing Autonomous Software
Building an AI agent that can actually accomplish useful work across the web sounds straightforward until you hit the payment layer. Every time an autonomous script needs to buy data, spin up a service, or complete a transaction, it slams into a wall of login screens, credit card entry forms, and separate subscription paywalls.
Mitchell Jones learned this reality the hard way. After stepping away from Lendtable, his earlier Y Combinator-backed fintech venture (YC S20), he began experimenting heavily with artificial intelligence. He wanted to build what seemed like a simple form-filling agent. Before he realized what was happening, he had burned through more than $400 trying to build and test a basic prototype.
“I kept running into the same issue,” Jones explained in interviews regarding the genesis of Lava Payments. “I was using the same underlying models and tools again and again, but through different wrappers or platforms.” Each time he wanted to test a minor variation or access a utility, he was forced to start a new subscription, re-authenticate, and pay separately—even though he was already paying for access to the core model.
“That felt fundamentally broken,” he continued. “I didn’t want to keep rebuying access to the same thing under a different wrapper. What I wanted was a single wallet, one set of credits, and the ability to move between tools and providers without starting over every time so I could pay for what I was using.”
This frustration catalyzed a new vision: Lava Payments, a startup designed to tackle the structural friction of machine-to-service transactions in the emerging agent-native economy.
The Micro-Transaction Nightmare for AI Agents
As autonomous agents transition from experimental chat interfaces to active workers capable of executing multi-step workflows, they face a financial architecture built entirely for human beings. Human commerce relies on saved credit cards, two-factor authentication prompts, modal pop-ups, and email receipts. None of these mechanisms are compatible with an unsupervised background script executing fifty micro-tasks per minute across dozens of disparate web applications.
Without a dedicated payment abstraction layer, agents are effectively paralyzed. An agent tasked with gathering competitive pricing, purchasing API tokens, booking travel, or licensing datasets cannot pause its execution loop to ask a human user for a credit card number every time a $0.05 fee is encountered. Yet, establishing individual accounts and pre-funding balances across hundreds of micro-SaaS platforms is economically and operationally prohibitive.
The current landscape of AI tooling exacerbates this problem. Developers bounce between numerous wrappers and proprietary platforms, each demanding its own billing relationship. This creates immense overhead not just for individual developers like Jones, but for enterprise teams trying to deploy autonomous workflows at scale. Wrapper fatigue is matched by token exhaustion and friction at every external API boundary.
How Lava Payments Works: Prepaid Wallets and Usage Credits
Lava Payments approaches this bottleneck by introducing a universal digital wallet system powered by prepaid usage credits. Rather than managing fragmented subscriptions across dozens of discrete vendors, developers and end-users load funds into a centralized Lava wallet.
The mechanism is designed for seamless interoperability across the web:
- Unified Credits: Users purchase a single pool of usage credits that can be drawn down dynamically as agents interact with various participating merchants and services.
- Cross-Merchant Utility: Any merchant that integrates the Lava wallet can accept these credits instantly, removing the need for repeated credit card entry or per-service invoicing.
- Model Agnostic: The system integrates cleanly with foundational models such as OpenAI’s GPT series and Anthropic’s Claude, ensuring that pay-as-you-go transactions flow naturally alongside model inference costs.
- Autonomous Authorization: By operating on pre-allocated credits, agents can authorize micro-transactions independently without triggering constant human-in-the-loop verification prompts.
Jones draws a helpful analogy to traditional infrastructure: “Without Lava, agents can’t move smoothly through the internet because they constantly get blocked when it comes time to pay.” He compares it to telecommunications—when a person opens Google Maps on their smartphone, they do not pay Google a separate fee for every map tile loaded; they have already paid their internet service provider (such as Verizon or AT&T) for access. Lava aims to establish a similar baseline billing layer for software agents traversing the digital landscape.
Founder Background and the Path to Lava Payments
The journey toward founding Lava Payments is rooted in Mitchell Jones’s diverse experiences across elite financial institutions, tech giants, and startup accelerators. Born to a working-class family in Dayton, Ohio, Jones was taught early on that the traditional recipe for success was hard work, disciplined saving, and a rigorous education.
He followed that roadmap to Yale University, after which he built a career spanning high-profile roles at Goldman Sachs and Meta. However, the entrepreneurial bug proved stronger than corporate stability. Jones went on to found multiple fintech ventures, including Parable and Lendtable—the latter of which was part of Y Combinator’s Summer 2020 cohort (YC S20).
His background in fintech and lending gave him a deep appreciation for the plumbing of financial transactions, regulatory compliance, and capital allocation. When he encountered the severe payment bottlenecks plaguing modern AI development, his fintech expertise positioned him uniquely to solve them.
Securing Seed Funding in the Agent-Native Economy
To turn the vision of agent-centric financial infrastructure into reality, Lava Payments announced a $5.8 million seed funding round in August 2025. The round was led by prominent venture capital firm Lerer Hippeau, with participation from Harlem Capital, Streamlined Ventures, and Westbound.
The connection with Lerer Hippeau carries a personal dimension: Jones went to high school with Will McKelvey, now an investor at Lerer Hippeau. McKelvey had tracked Jones’s entrepreneurial trajectory for years, maintaining a mutual desire to collaborate on a future venture when the right opportunity presented itself. Lava Payments proved to be that defining moment.
The freshly secured capital is earmarked for strategic growth initiatives:
- Team Expansion: Hiring top-tier engineering, product, and compliance talent to scale the platform architecture.
- Product Development: Enhancing the core wallet infrastructure, expanding merchant integrations, and refining developer toolkits.
- Go-to-Market Strategy: Accelerating merchant adoption and establishing partnerships across the AI ecosystem.
Building the Invisible Infrastructure for the AI Web
As artificial intelligence shifts from passive assistant models to proactive, autonomous agents capable of conducting commerce, the underlying commercial rails must evolve in tandem. Lava Payments is positioning itself as the invisible, friction-free financial layer powering the agent-native economy.
Competitors and adjacent players—such as usage-based billing platform Metronome—are also addressing parts of the modern billing puzzle, but Lava’s focus remains sharply tailored to agent-to-service transactions and unified digital wallets.
Ultimately, Jones envisions a web where autonomous software operates with fluid efficiency, unhindered by legacy human payment gates. By solving the micro-transaction barrier, Lava Payments hopes to ensure that artificial intelligence tools remain accessible, extensible, and genuinely useful for builders everywhere—from Silicon Valley labs to hometown innovators in Dayton, Ohio.
Related reading
- Why the Linux Foundation Is Giving AI Agents Their Own Payment Standard With x402
- x402: The Linux Foundation's Bet on Machine-Native Payments
