A16z: From Hourly Billing to Outcome-Based Revenue Sharing, a $100 Billion Relay Race

TechFlow Summary: This analysis from a16z points directly to the most critical turning point in AI commercialization: shifting from hourly cost savings to revenue-sharing based on outcomes. If you're a SaaS practitioner or investor, this article explains why the next wave of AI unicorns won't sell "seats," but will instead take a cut of your client's revenue growth.

Traditionally, calling your ISP to report an outage usually goes like this: a robotic voice first addresses you as a "valued customer," then puts you through several minutes of phone menu navigation: "Press 3 for technical support, 2 for broadband, 4 for other issues." You're then left on hold with waiting music, feeling forgotten, until a representative finally answers and asks for your name, account number, and problem description. If restarting the router doesn't fix it, you wait again, only to have to retell the whole story to another rep before finally scheduling an on-site visit.

AI is already transforming these customer service scenarios. Instead of struggling with robotic voices and multiple representatives, an AI agent can now answer immediately. It can access your account, check if others in your area are affected, resolve simple issues directly, or schedule a technician appointment. No more "all representatives are currently busy," no more holding music. Until recently, this was essentially the technological ceiling.

But now, new capabilities are enabling AI agents to track leads and execute tasks over longer time horizons. We believe this will allow AI companies to move beyond single customer interactions and rethink how they handle complex customer journeys spanning weeks or even months. AI will guide insurance clients from their first call to final settlement, mortgage borrowers from pre-approval to closing, and enterprise sales leads from first demo to signed contract. This not only delivers better outcomes for customers, but also allows AI vendors to get paid based on revenue growth, rather than just charging for cost savings.

Value Only Materializes at the Finish Line of the Relay

There are many types of businesses where creating customer value requires numerous steps across weeks or even months. Take the traditional plaintiff law firm case preparation process as an example. Potential cases typically come in as inbound calls, and intake specialists gather key details before handing them to attorneys to assess whether to take the client. Once the firm signs the case, investigation and discovery begin: paralegals pull police reports, try to locate witnesses, send preservation notices to defendants to secure evidence, and continuously update the client on progress. Then the firm waits for the client to conclude medical treatment, drafts a medical timeline, and hands the materials to attorneys to finalize the demand letter.

This process is what we used to call a "relay": a long journey where each handoff carries the risk of drop-off or value leakage. Prospective clients typically call multiple personal injury firms, usually contracting with whichever responds first. Thus, if a caller reaches only voicemail, or qualification moves too slowly, the opportunity leaks from the pipeline. Once signed, discovery spawns parallel workflows; firms may lack the patience to place a fifth call after four go unanswered. Additionally, staffing limitations can prevent meticulous review of every exhibit, particularly in routine or low-value matters. Carefully parsing physician notes could uncover signs of brain injury, substantially increasing the settlement payout, yet the firm might just accept the first offer from the defense. Lastly, cases can linger for months involving numerous parties, with information scattered across files, personnel, and software platforms. This renders it nearly impossible for the firm to maintain a comprehensive view of each client. As a result, critical invoices may never make it into the demand letter, untreated periods remain unaddressed, and settlement totals inevitably suffer.

The Opportunity for Long-Haul Agents

Until recently, AI agents mostly handled bounded tasks: "send this follow-up email," "summarize that call," "create this document," alongside narrow-scope interactions. Customer service became an ideal early deployment scenario precisely because the problem space is well-defined, enabling high reliability while allowing customization based on company-specific policies. Interactions are short, outcomes are verifiable, and if an AI agent encounters an edge case it can't handle, a human team is typically still available to escalate.

Today, long-haul agents can pursue goals across dozens of steps, multiple modalities, and extended timeframes, enabling them to complete the entire relay from start to finish. In the plaintiff firm example, a long-haul agent can be assigned a case starting from the initial intake call, with the objective of maximizing the client's recovery amount. The agent saves and retrieves case context as needed. It auto-wakes based on events or deadlines, proactively dispatches sub-agents to contact witnesses, update clients, and communicate with lawyers via phone, email, or SMS. It then synthesizes discoveries, plans next steps, drafts documents for review, and pushes toward the best possible resolution. Across clients, it continuously learns; data generated from each case enables it to handle the next one even better than before.

Agents solve the primary value-leakage points observed in today's relays. At the journey's start, prospects no longer churn due to slow qualification, since agents are online regardless of time or day of the week and can assess cases almost instantly. Discovery is no longer bottlenecked by manpower, as the case agent can spin up a sub-agent to make a fifth, tenth, or twentieth call at any moment. Because agents scale nearly infinitely, even seemingly routine matters receive thorough handling, ensuring every exhibit is carefully reviewed and potentially overlooked brain injury clues are surfaced. Finally, agents maintain a complete case record—from witness call recordings to police reports to medical documentation—ensuring zero information loss.

Paying for Outcomes

Beyond plugging these leaks, owning the full customer journey transforms the vendor's business model. When agents deliver results with higher quality and speed, they drive revenue growth for clients, not just cost reduction. Crucially, because impacts are observable and attributable to the agent, vendors can share in the incremental value they create by pricing based on work products or outcomes. Using the plaintiff firm as an example, an individual case is a natural unit of work, and the final settlement value is clearly identifiable. Moreover, as agents capture an increasingly larger share of the final deliverables, tying outcomes back to the agent becomes straightforward. Vendors can therefore share in revenue growth, commanding contract values previously unattainable through cost-cutting alone.

Admittedly, shifting to outcome-based pricing isn't frictionless. In the plaintiff legal industry, most states prohibit non-lawyers from sharing legal fees, meaning pricing must align with settlement amounts without direct revenue splitting. In other verticals, vendors may face procurement pushback, as there's little precedent for outcome-based contracts. Even as agents approach full ownership of the outcome, attribution disputes will initially arise. Therefore, successfully deploying long-running agents presents both a commercial and a technical challenge.

However, outcome-based pricing in high-value customer journeys is hardly a novel concept.

The Case for an AI-Native Red Ventures

Red Ventures (RV) serves as a strong historical precedent. Founded by Ric Elias and Dan Feldstein, RV pioneered digital, end-to-end outsourced sales operations for long-term consumer contracts. Taking insurance as an example, RV captures prospect information when users search for these products (via owned properties like homeinsurance.com), initiates outbound outreach, guides them through coverage options, and ultimately closes the policy. Because it fully owns this journey, RV earns a fee from insurers for every closed policy. This model was highly effective: according to S&P Global data, RV generated approximately $370 million in EBITDA on roughly $1.4 billion in revenue in 2022.

Today, this model faces structural headwinds. At the top of the funnel, how consumers discover products is shifting. They're no longer reliant on Google searches or comparison sites to explore and evaluate offerings, but instead ask personal agents like Meta's Muse or Instinct directly for recommendations, eroding the value of historically acquiring discovery-focused traffic. Further downstream, hundreds or thousands of humans operating linear funnels are being replaced by long-running agents that can navigate vastly more complex journeys. They operate across products, communication channels, and in parallel, support any language, scale massively, and possess infinite persistence. RV's EBITDA is projected to fall to around $160 million in 2025 due to client churn and softening ad spend, signaling the model's fragility, with AI set to further disrupt the entire landscape.

Three Go-to-Market Approaches

There are distinct ways to bring long-running agents to market:

(1) Selling the platform: Under this model, vendors sell long-running agents as software to augment existing enterprise operations. Integrating agents into current teams and workflows, as seen in plaintiff firms, is a prime example. Platform selling offers upside potential, as pricing isn't anchored by comparable software or outsourcing contracts. However, it elongates sales cycles due to the need for new budget line items, and it becomes difficult to cleanly attribute results when an agent interacts across multiple departments.

(2) Selling services: Here, vendors take over an already-outsourced function to deliver expected outcomes more efficiently, similar to how Red Ventures owns the entire acquisition journey for insurers. This approach shortens sales cycles because buyer ROI calculations are straightforward and BPO budget lines already exist. Nevertheless, legacy pre-AI pricing models may fail to capture the true value long-running agents can deliver, and since many high-value journeys are currently operated in-house, this model's total addressable market remains constrained.

(3) Full-stack play: Going full-stack means vertical integration to become an AI-native operator. For instance, instead of selling long-running agents to a debt collection agency, a company could acquire debt portfolios and collect independently. This allows you to capture the entirety of AI-driven upside and sidestep attribution debates. However, it also requires absorbing servicing functions like licensing and compliance that sit outside the core customer journey, introducing regulatory overhead and operational complexity.

It's too early to declare which approach will dominate, but we are confident that long-running agents will win at the scale of the relay race, expanding the AI-addressable market from a portion of labor budgets to a portion of revenue. Plaintiff case preparation is merely one example; we expect agents to drive outcomes across other high-value journeys, including mortgage origination, insurance claims processing, complex enterprise sales, and countless other scenarios. If you're building to capture this opportunity, we'd love to hear from you!

Join the official Coincamps community:

X: https://x.com/coincamps

Telegram: https://t.me/coin_camps