ProSearch Pulse
Julia HasenzahlJulia Hasenzahl
Join our conversation!
ProSearch Pulse

The ProSearch Pulse

This week we flagged the following:

    • Gregg Wirth at the Thomson Reuters Institute flags the “First Draft Trap,” where an AI memo becomes a cognitive anchor described by Kahneman and Tversky’s 1974 research, quietly narrowing strategy before real analysis begins.
    • Dan Hauck of NetDocuments says 2025 shifted legal AI from trying to using, and 2026 will decide sustained impact as “intelligent assistants” move into daily tools, demanding governed metadata and traceable collaboration.
    • Harris Tilevitz and Eric Friedman describe Skadden partners watching client sentiment flip from “do not use GenAI” to “we expect it,” with proposed three-year rate horizons and discounts that ratchet up as AI efficiency rises.

    In This Issue:

    • Clients price AI gains
    • Legal AI, operationalized
    • AI’s first-draft bias
    • Curation before ingestion

    Law firms under AI

    Skadden alumni warn AI could reshape legal pricing

    Story Image

    Two longtime Skadden leaders say clients now expect AI use, and some are pushing multi-year rate deals tied to efficiency gains.

    Two longtime Skadden technology leaders say the generative AI conversation has shifted fast. In a Legaltech News Q&A, former Skadden AI and technology chief Harris Tilevitz and former CEO Eric Friedman described a move from bans to expectations, with pricing and governance now central to law-firm tech strategy.

    Key insights

    • Tilevitz, who left Skadden after 38 years and now advises legal tech startup Harvey, said technology planning should start with what lawyers actually do, peer benchmarking, and partner demands, rather than chasing shiny objects.

    Some clients are proposing three-year rate deals where discounts "ratchet up" as AI efficiency increases, adding pressure on hours and rates.

    • Tilevitz and Friedman pointed to a familiar adoption pattern. Partners initially resist new systems, including databases, vendors, and the cloud, then normalize them over time. AI appears to be following the same arc, with one twist. Some partners now require AI under specific restrictions.
    • Friedman, now advising Ironclad and Litera, flagged a potential shift toward three-year rate horizons. As AI-driven efficiency improves, discounts could ratchet up, increasing pressure on hours and potentially on hourly rates.
    • Both leaders framed AI governance as a security-first discipline. Law firms generate confidential, partner-owned files every day, so protocols need to be rigorous before pilots or broader rollouts. Policies also need to reflect different partner tolerances for uploading documents and prompting.

    As partners bake AI efficiency expectations into outside counsel guidelines, the next legal-tech battleground moves from adoption to accountability. Firms will need to show tools are secure, outputs are reliable, and savings are measurable. That makes governance and measurement a strategic priority, because AI shapes how work gets done and how it gets priced.

    For ProSearch, this aligns with what we see across eDiscovery and legal data programs. Early planning, clear governance, and an AI + HI model help legal teams turn legal data into strategic insight while keeping security, defensibility, and cost discipline in focus.

    via www.law.com

    Read Now →

    Legal AI gets practical

    Forbes: Legal AI shifts from usage to impact in 2026

    Story Image

    NetDocuments CPO Dan Hauck says the next phase hinges on workflow integration, governed collaboration, and stronger metadata foundations across legal systems.

    Legal AI adoption is entering a more accountable phase. Writing in Forbes, NetDocuments CPO Dan Hauck argues that 2025 brought broad usage, while 2026 will test whether legal teams can turn that usage into sustained impact through deeper workflow integration, stronger knowledge foundations, and governance that holds up under real legal work.

    Key insights

    • Intelligent assistants are moving into the tools lawyers already use, supporting more ambient drafting and next-step guidance. Hauck flags a practical risk: suggestions that ignore role, matter context, or the task at hand can distract attorneys and create downstream issues.
    • Automation is shifting from rigid, rules-based workflows to intent-aware systems that blend probabilistic extraction with deterministic document assembly. The result is more nuance and better exception handling, because legal processes rarely fit a clean template.
    • Collaboration is getting harder to govern as work spreads across documents, messaging, research, and knowledge platforms, with lawyers, partners, and AI systems all contributing. The operational priority is security, traceability, and clear accountability.
    • Better models alone are not enough. Hauck emphasizes that search becomes understanding only when unstructured content is paired with structured metadata and contextual signals, supported by integrations that carry context across the stack.

    For legal leaders, the 2026 question is operational: which workflows will be redesigned so AI reduces friction while preserving defensibility? Teams that align governance, permissions, metadata, and integration early are better positioned to measure impact in time saved, fewer handoffs, and more consistent knowledge reuse across matters.

    From a ProSearch lens, this maps cleanly to an AI + HI model, where systems accelerate pattern recognition and recall, and people provide the judgment, guardrails, and matter-specific context that keeps the work defensible. The teams that treat integration and knowledge foundations as first-class design work will be the ones that can turn legal data into strategic insight.

    via www.forbes.com

    Read Now →

    Chris Dahl

    Chris Dahl

    2026 feels like the inflection point when “AI in legal” moves from pilots to a real operating model. Teams seeing measurable lifts align early on the fundamentals, co-designing workflows, and carrying metadata across systems. Then they hold the program accountable to outcomes like reduced cycle time, fewer handoffs across teams, and clear, defensible audit trails.

    AI and Legal Judgment

    Lawyers face AI’s “first draft” anchoring trap

    Story Image

    A Thomson Reuters Institute column warns that AI-generated drafts can quietly steer legal strategy before lawyers fully explore alternatives.

    AI drafting tools erase the blank page by producing polished legal prose in seconds. A new Thomson Reuters Institute column argues that speed carries a quieter trade-off: the First Draft Trap. An AI-generated starting point can lock in a theory, framing, and direction before real legal judgment has a chance to lead.

    Key insights

    • The column ties AI-first drafting to anchoring, drawing on Kahneman and Tversky’s 1974 research showing that early exposure can skew later judgment, even when the starting reference point is arbitrary.

    An AI draft can feel like analysis, but it often locks in a framing first—turning legal judgment into editing.

    • An AI draft becomes a seductive anchor because it reads like competent legal writing. Once a plausible narrative sits on the page, alternative theories, structures, and issue spot checks get harder to surface.
    • The author calls the workflow anti-Socratic. It compresses ambiguity too early. Legal training builds comfort with holding multiple possibilities in tension before committing to a position.
    • Senior lawyers’ peripheral vision, the ability to notice missing issues, alternative framings, and audience effects, can atrophy when teams default to “generate a draft” rather than mapping the problem space first.

    A procedural fix that keeps judgment in the driver’s seat

    The adjustment is simple: ask for the map, not the draft. Use AI to generate three materially different strategic framings, and for each one, capture both the strongest argument and the most serious vulnerability. Add a clear instruction: Do not write a draft yet.

    That small process change keeps options open long enough for human judgment to do its highest-value work, especially when legal teams and their partners are trying to turn legal data into strategic insight without letting the first plausible narrative set the destination.

    via www.thomsonreuters.com

    Read Now →