Human-agent trust for legal services
Attorney-client privilege was hard to maintain when one person did the work. It is harder when the work is split between a lawyer and three AI agents. Here is the legal-services HATI playbook.
What is at stake
The privilege protecting attorney-client communications, the unauthorized practice of law restrictions, and the duty to supervise non-lawyer assistants — three foundational doctrines now collide with AI agent workflows. The doctrines did not anticipate non-human assistants. State bars are catching up; vendors that build with the structural answer first will win the next decade.
The four flows where HATI matters
Document drafting. AI drafting assistants generate first drafts of contracts, briefs, memoranda. Without attestation, the firm cannot prove which work product was authored vs. supervised vs. directed. Insurance carriers and bar regulators increasingly want this distinction.
Discovery and document review. Predictive coding has been judicially accepted in the US for over a decade, but agent-driven review at scale changes the supervision question. Each document classification needs a chain back to a supervising attorney.
Client intake. Conflict checks, KYC, retainer agreements signed via agent-mediated flows. Privilege attaches at intake; identity continuity from client to lawyer to agent must be cryptographic.
Billing. AI agents that compose timesheets and bill descriptions. State-bar rules forbid billing for work not performed by a billable resource; the audit log must distinguish authored from directed.
BIPA and biometric AI
Illinois Biometric Information Privacy Act (and its successors in other states) requires written consent for biometric collection. Manav's Layer 1 is biometric; deployment in BIPA jurisdictions requires consent flows handled per the statute. The architecture supports this — the consent itself becomes part of the delegation chain — but the flow needs explicit design.
The deployment
- Layer 1: every attorney verified, federated with bar-card or firm IDP.
- Layer 2: agents in document drafting, review, and billing under per-attorney delegations.
- Layer 3: every output stamped author/supervisor/director.
- Audit: tamper-evident, exportable for malpractice insurance and bar review.
The malpractice-insurance angle
Malpractice carriers today are quietly differentiating premiums for firms with documented HITL on AI workflows. The argument is straightforward: provable supervision is provably underwritable. Firms that integrate HATI early get access to better terms before the market normalizes them.
Common objections
The two pushbacks we hear from this vertical: integration risk — addressed by phased rollout starting with the audit trail (lowest risk, highest evidence-to-effort ratio), and internal politics — addressed by anchoring the project to a regulator deadline or a security-questionnaire deal-blocker, where the political question answers itself.
Frequently asked questions
What is the first integration to ship? The signed audit trail. It costs least, satisfies the most regulators, and produces the evidence everything else builds on. Every vertical we have integrated started here.
How does this affect end-customer experience? Invisibly, by design. The customer sees the same UI; the difference is in the audit log behind it. The latency added is single-digit milliseconds. The trust gain is structural.
What's the buying motion — security, compliance, or the line? Compliance writes the check; security signs off; the line of business sets the timeline. The strongest deals start with a regulator deadline; the next-strongest start with a deal-blocking security questionnaire.
Where to start
The first integration we recommend in this vertical: hipaa ai agents, then agent identity finance. Both are deployable inside a quarter; both produce regulator-grade evidence; both unblock procurement conversations the rest of the stack depends on.
Adjacent reading
For the regulatory ground truth in this vertical, see the Article 14 playbook. For the integration shape, see audit-trail design. For the buying motion, see the CISO compliance stack. Most successful pilots in this vertical follow that order: regulation first, integration second, procurement third.
What changes inside a litigation-track law firm
Three operational shifts, each with a hard date in the next 18 months. Privilege protection. Agent-assisted research must carry a delegation that names the supervising attorney; without it, the privilege claim is harder to defend on cross. Conflict checks at the action layer. Each agent action is checked against the firm's conflicts database before it runs, not after; the audit trail proves the check was performed. Engagement letters. New engagements include a clause specifying which agent classes are permitted, what scopes they receive, and how the audit trail is retained.
Most firms we have spoken to underestimate the engagement-letter change. Once the language is in one client's letter, it propagates through the firm's templates and becomes the new baseline within a quarter.
What the bar association will say about this
Bar associations across the major US jurisdictions are circulating draft ethics opinions on AI-assisted legal work. The drafts we have read converge on three requirements that the substrate addresses by default. First, the lawyer of record must be identifiable on every consequential AI-augmented work product. Second, the augmentation level must be declared per filing, not per practice. Third, the audit trail must be preservable across the lawyer's career, including across firm changes. The first two requirements are direct outputs of substrate-grade delegation. The third — career portability — is the property the substrate was specifically designed for. Lawyers building careers under the new ethics opinions need infrastructure that produces all three artifacts. Firms that adopt the substrate now produce the artifacts as a matter of course; firms that wait for the opinions to formalize will be retrofitting under an ethics-rule deadline, which is the worst class of timeline to ship infrastructure under. The bar associations will eventually name the substrate. The strategic move is to be the substrate they name.
Privilege survives the agent age. Only if the supervision is cryptographic.