The End of the Chatbot: How Proactive AI is Rewiring Litigation

SAN FRANCISCO – September 25, 2026 – For the past three years, the legal industry’s adoption of artificial intelligence has been defined by the blinking cursor. The promise of generative AI was delivered through conversational copilots—chatbots that required lawyers to painstakingly engineer prompts, isolate context, and cross-examine the machine’s output. But the honeymoon phase has abruptly ended, replaced by a widespread operational friction known as "prompt fatigue."

Now, a San Francisco-based legal technology startup is betting that the future of legal AI isn't conversational at all; it's autonomous. Advocacy, a company founded by former M&A and securities litigator Téo Doremus, today unveiled Promptless. The platform represents a fundamental architectural shift from reactive chatbots to proactive, event-driven background agents, quietly executing tasks without ever being asked.

The release signals a critical inflection point in the mechanics of legal power. By transforming AI from a passive research tool into an active participant in case strategy, the startup is not just changing how litigators work—it is fundamentally challenging the economic and ethical frameworks that govern the business of law.

Beyond the Prompt: The Rise of the Autonomous Agent

Since the generative AI boom began, platforms have relied on session-isolated chat boxes. An attorney inputs a query, the model processes it, and an ephemeral answer is returned. If the lawyer doesn't ask the right question, the machine offers no insight.

Promptless dismantles this dynamic. Built on a proprietary centralized case memory system known as the Axiom context engine, the software operates asynchronously. Instead of waiting for a user, it monitors external triggers: a newly filed court docket, an incoming email from opposing counsel, or a massive document production. When a trigger occurs, the platform automatically cross-references the new data against the firm's living case theory. It reads documents on arrival, surfaces factual contradictions, and prepares multi-step research memoranda or first-draft outlines—all before the lead partner has even poured their morning coffee.

"Lawyers don’t write briefs by having conversations with a chatbot. Promptless lets humans work the way they want to," said Téo Doremus, CEO and co-founder of Advocacy. "Promptless comes to tasks with the team’s best intelligence already behind it, and it lets the lawyer run the case, not the AI."

This model-agnostic architecture—which allows firms to plug in their preferred foundation models—creates a significant technical moat. By maintaining a dynamic graph of the litigation, the system moves beyond semantic search and into strategic orchestration. But granting a machine persistent read-and-write access to a law firm's most sensitive communication channels introduces profound questions about control and confidentiality.

The Rule 11 Paradox: Shielding Against Hallucinations

In the high-stakes arena of complex litigation, delegating strategy to an algorithm is a perilous gambit. Federal Rule of Civil Procedure 11 mandates that attorneys independently verify the evidentiary support of their filings. The legal industry has already witnessed the catastrophic professional fallout of AI hallucinations, where lawyers faced severe sanctions for submitting fabricated case citations.

To neutralize this malpractice exposure, Advocacy has engineered a structural defense it calls "Authored Intelligence." Every action the proactive agent takes—whether flagging a contradictory deposition transcript or drafting a summary judgment motion—is isolated in an "unconfirmed" quarantine state. No generated insight becomes part of the operative case record until a licensed attorney affirmatively accepts it.

This mandatory human gatekeeping generates a timestamped, immutable audit trail of both human and algorithmic contributions. On the surface, this meticulous attribution serves as a shield against court sanctions, proving that counsel maintained supervisory control in compliance with recent American Bar Association (ABA) Formal Opinion 512 directives on generative AI.

However, risk management experts and malpractice insurance underwriters note that this audit trail is a double-edged sword. To eliminate platform friction, the software features "login-less approvals," allowing senior partners to approve research or delegate next steps directly from encrypted email threads on their mobile devices. If a litigator relies on a machine-generated analysis without conducting a diligent review, the system's own audit log could provide documentary evidence in a sanctions hearing that a partner spent a mere fourteen seconds reviewing a complex filing before tapping "approve" on their smartphone. Furthermore, transmitting sensitive case theories via email to facilitate these approvals tests the boundaries of client confidentiality under ABA Rule 1.6.

Pay-Per-Accepted-Thought: Disrupting the Billable Hour

Perhaps the most disruptive element of the Promptless launch is not its code, but its capitalization strategy. The company is abandoning the standard per-seat Software-as-a-Service (SaaS) licensing model that has dominated enterprise tech for a decade. Instead, firms pay a nominal fee to initialize a case workspace and are then charged à la carte only when an attorney affirmatively incorporates an AI-generated work product into the matter.

This "pay-for-accepted-work" model aligns vendor fees directly with delivered utility, but it strikes at the very heart of the law firm profitability puzzle: the billable hour.

Under long-standing ethical frameworks, such as ABA Formal Opinion 93-379, a lawyer billing hourly cannot charge a client for time not actually expended. If an autonomous agent drafts a motion in seconds that previously required fifteen associate hours, the firm can only bill for the partner's actual time spent reviewing the draft. The efficiency savings are passed entirely to the client, while the firm's revenue evaporates.

In theory, firms could pass the software's transaction cost to the client as a disbursement. Recent guidance, such as Texas Ethics Opinion 705 and Florida Bar Opinion 24-1, permits this only if billed at actual cost without markup and with explicit client consent. However, corporate general counsel and the Corporate Legal Operations Consortium (CLOC) are increasingly hostile to new vendor fee structures. Many corporate billing guidelines explicitly forbid law firms from passing AI platform charges through as client-reimbursable expenses, viewing them instead as routine firm overhead.

This creates a perverse economic incentive. Associates, pressured to maintain high billable hours, might review a machine-generated draft, manually retype its core arguments to justify their time, and then reject the output in the platform to spare the matter the software fee. Conversely, partners might over-accept algorithmic drafts without proper vetting if they treat the cost as a nominal pass-through while billing inflated review time. The friction between law firm cost-recovery structures and client billing compliance is set to become the defining battleground of legal operations.

The Strategic Rationale and Market Positioning

The quiet moves behind Advocacy’s ascent reveal a calculated strategy to corner the high-end litigation market. Emerging from stealth earlier this year with a multimillion-dollar seed round led by venture firm Relentless, the startup secured strategic participation from Silicon Valley law firm Fenwick & West LLP and Rel Labs, the venture arm of e-discovery giant Relativity. This alliance points toward a future where proactive agents are seamlessly integrated into the massive evidentiary databases that define modern antitrust and mass tort litigation.

Furthermore, the company has already secured an international design partnership with Scott+Scott Attorneys at Law LLP, a premier plaintiffs and commercial litigation powerhouse. This provides Advocacy with immediate frontline validation in document-intensive, high-stakes environments where traditional chatbots have historically stumbled.

The transition from reactive chatbots to proactive, event-driven agents is not merely a software upgrade; it is a redistribution of leverage. By embedding autonomous intelligence into the background of a case, the technology forces the legal profession to confront uncomfortable truths about the value of human judgment, the ethics of automated billing, and the true cost of efficiency. As machines begin to read the docket and anticipate the next move, the most successful litigators of the next decade will not be those who write the best prompts, but those who know exactly when to reject the machine's advice.