Translating the Tools & Trends That Shape the Legal Profession
    Rising Above the Noise

    The Playbook

    Laura Kennedy & Chelsey Lambert 5 min read
    The Playbook

    Monday, July 6, 2026

    Kirkland & Ellis signed a deal with Syllo on Monday to formalize its litigation methodology into an encoded system. The firm named the initiative Secret Sauce Litigation. The naming is the story. Twenty-four hours later, Spellbook opened early access to Autonomous Contract Management, an agent that runs the negotiation cycle against a firm’s own history of accepted redlines. The same Tuesday, Colorado’s landmark AI accountability law was supposed to take effect. It didn’t. Governor Jared Polis had already signed a replacement statute in May that stripped out the mandatory bias audits, the risk management program, and the duty of care. Three moves in three days, pointed at the same thing.

    The codification of practice — how a firm actually works, encoded so it can run at machine speed — is now the frontier where vendors, firms, and regulators are all making their moves. The vendors are building autonomous agents that run against the firm’s own history. The firms are encoding proprietary methodology into systems no vendor can sell them. The regulators just decided not to require any of it to happen in a particular way. What the industry sold as “AI adoption” for the last two years is turning into something more structural. The playbook — the specific, encoded, executable version of how a firm actually practices — is the asset now.

    The Vendor Move: Spellbook Launches Autonomous Contract Management

    Spellbook opened early access to Autonomous Contract Management on Tuesday, June 30, and framed it as the biggest expansion of its product since the company launched in 2022. General availability is planned for the coming months. What the system does is run the contract lifecycle end to end without a lawyer clicking through each step. ACM pulls contracts and drafts from the systems where legal work already happens — Outlook, Slack, connected document repositories — reviews and redlines them, and does so against the user’s own contractual standards rather than a public template.

    Scott Stevenson, Spellbook’s co-founder and CEO, named the design choice directly. “The more information Spellbook has about your history of negotiations, the better we can review your future agreements,” Stevenson told Law.com. “If we know you make these changes to a document that you usually get accepted, then we can propose those changes the next time you do a similar deal. People want, one: this feedback loop of your historical agreements feeding into how you review your next agreements. Two: a lot of customers didn’t get what they wanted out of CLM in the first place because it took too much configuration.”

    The architectural decision is what separates ACM from an ordinary contract review assistant. The training set is the firm’s own history of redlines, positions, and accepted terms. The output is the firm’s own playbook, executed at machine speed against the next incoming contract. The lawyer’s role shifts from clicking through review to setting the standards the agent runs against, and then verifying the results.

    Stevenson also named a second design choice. Spellbook is expanding its integrations across finance systems, not just legal ones. “It’s very hard to get people to use any new piece of software, and it’s very hard to go to your sales team and be like, ‘Okay, now you have to log into this legal software,’” Stevenson said. “So the main way we do it is go to where all those teams already are. We’re looking at finance systems. Finance is always trying to reconcile the actual financial transactions with the contracts to figure out what’s been paid and what hasn’t.” The pattern from Issue 5 — vendors going to the surfaces the users already work in, rather than making users come to a new surface — is now the default assumption.

    The Firm Move: Kirkland’s Secret Sauce and Reed Smith’s Cornell Program

    Kirkland & Ellis inked its deal with Syllo on Monday, June 29, and named the initiative Secret Sauce Litigation. The framing is deliberate. Kirkland is not adopting a vendor’s litigation platform. It is encoding its own methodology, its own accumulated pattern of how the firm actually litigates, into a proprietary system built with Syllo as the technical partner. The move sits alongside Kirkland’s $500 million Palantir bet from May, which Issue 3 covered, and the firm’s ongoing legal AI hiring push. Read together, the three moves describe a coherent posture. The infrastructure is being built. The methodology is being encoded. The people who understand both are being recruited.

    Herbert Smith Freehills Kramer made an architecturally identical move a few days earlier. On Friday, June 26, HSF Kramer announced a major deal to build a proprietary system of intelligence leveraging its own data and client knowledge. The vendor partner is different. The strategic decision is the same. Both firms have decided that the moat is not the platform they choose. The moat is the methodology they encode into whichever platform they build on.

    Reed Smith made the third firm-side move of the week. On Tuesday, June 30, the firm launched an AI Leadership Program at Cornell University, described as equipping partners to support clients navigating AI transitions. The training angle continues the thread from Clio’s Legal AI Accelerator, the Florida Bar Preferred Bar Program, and BARBRI’s Lega acquisition, all covered in Issues 5 and 6. The distinguishing choice is who Reed Smith is training. Partners specifically, not associates or staff. The firm has decided that the people who make the codification decisions — which methodology gets encoded, what gets prioritized, what standards get set — need to be trained at the partnership level. Industry commentary this week questioned whether partner-directed AI training programs actually change behavior. The question is fair. The move itself is still meaningful because it names the level at which firms think the codification work has to be led.

    The Regulatory Move: The Colorado AI Act Reset

    The largest structural story of the week is the one that didn’t happen the way most people expected it to. Colorado’s SB 24-205, the Consumer Protections for Artificial Intelligence Act, was scheduled to take effect Tuesday, June 30. It was the most comprehensive statelevel AI accountability law in the country, modeled in part on the EU AI Act, and it applied to any high-risk AI system used to make consequential decisions in employment, housing, lending, health care, insurance, education, and — the category that matters for this newsletter — legal services.

    The law never took effect. Governor Jared Polis signed SB 26-189, the Automated DecisionMaking Technology Act, on May 14, 2026. That statute repealed and replaced SB 24-205 before the June 30 deadline arrived. The new effective date is January 1, 2027. The replacement is a materially different law. The duty of care, requiring deployers to use reasonable care to protect consumers from known or foreseeable algorithmic discrimination, is gone. The mandatory annual algorithmic bias impact assessments are gone. The risk management program requirement, which had been the piece of the law that pushed developers and deployers toward frameworks like the NIST AI Risk Management Framework and ISO 42001, is gone. What remains is a narrower disclosure requirement. When automated decision-making technology contributes to an adverse outcome, the deployer has to tell the affected individual within 30 days.

    For law firms and legal AI vendors that had been building compliance programs against the SB 24-205 framework, this is a full reset. For firms that had not yet started, the takeaway is subtler and more important. State-level AI regulation is a moving target that can be moved backward, not just forward. The compliance infrastructure that many assumed would eventually standardize how firms encode their AI systems is not going to be delivered by Colorado, at least not on the timeline the industry planned around. And the states that were watching Colorado as the bellwether now have a different model to weigh.

    The interaction with the firm-side codification moves is what makes this consequential. Kirkland is encoding its litigation methodology. HSF Kramer is encoding its client and matter data. Spellbook is training its agents on firm-specific negotiation history. All three moves accelerate a category of AI system that the original Colorado law would have brought inside a duty-of-care framework, at least where it touched legal services. The replacement law does not. Firms building their own systems now sit in a regulatory landscape that will not require them to conform their methodology encoding to a common standard. The competitive advantage of a well-encoded playbook just got larger, and the regulatory floor got lower.

    In Other News This Week

    Descrybe launched verified legal research inside ChatGPT on Thursday, July 2, extending its earlier consumer-focused integration into a workflow product. The company also announced that its paid subscriber base grew by more than 700 in the six weeks before the launch. The move is consistent with the Big Tech vertical pattern from Issue 6. Perplexity built Computer for Counsel to sit above Westlaw and LexisNexis. Descrybe is building a legal research layer inside OpenAI’s consumer surface. Different Big Tech partner, same architectural choice.

    Litera released a report on Thursday, July 2, titled In-House Legal Is Winning on Risk

    Instinct, Losing on Governance Infrastructure. The report sits alongside Thomson Reuters’ Future of Professionals Report 2026 from Issue 6 as a matched pair. Two of the largest legal tech companies in the world published reports two weeks apart pointing at the same weakness. In-house teams have the instinct to see the risk. They do not yet have the governance infrastructure to act on it.

    Deloitte Legal published a market analysis on Tuesday, June 30, projecting that AI agents could handle 30% of the work within corporate legal departments in the next three to five years. Read the number as a directional estimate rather than a prediction. The framing carries the signal. Deloitte is telling its corporate clients to plan for a 30% shift in what legal work actually looks like, which will pull directly on both in-house team structures and outside counsel relationships.

    Scan Logic announced major enhancements to BillSync on Thursday, July 2, aimed at helping firms reduce revenue leakage and get paid faster. The billing side of legal AI is quiet compared to the platform-level moves but is compounding steadily. Juristic joined Board Intelligence on Wednesday, July 1, folding a legal task management and timeline product into a board technology platform. The category boundaries between legal, board, and finance software keep softening, which is the same trend Spellbook named.

    The Big Tech backdrop is worth flagging briefly. Anthropic released Claude Fable 5 on Wednesday, July 1, after a multi-week safety review and coordination with the Trump administration. Mythos 5, Anthropic’s enterprise model, had been released conditionally first, and now Fable 5 has cleared the same coordinated pattern. OpenAI’s GPT-5.6 remains under White House-directed delay for national security review, with mid-July as the current target for general availability. For legal AI platforms rebuilt on foundation models — Thomson Reuters on the Claude Agent SDK, Clio routing across multiple models — the pattern is a new dependency worth naming. The underlying model layer can now be gated by an interagency review the legal platform vendor does not fully control.

    The CTO Gap Zone™ of the Week: The Encoded Practice

    Last week’s Gap Zone named the rental relationship every firm now sits inside. The model is rented from a foundation company. The content is rented from a legal data provider. The firm context, the matter data, the playbooks, the clause libraries, the institutional memory of how the firm actually works, is the one layer the firm owns. This week extends the argument. Owning the layer is not the same as being ready to encode it.

    Most firms have never inventoried their playbooks in a form that can be handed to a system. Litigation strategy lives in the heads of the partners who have run the cases. Negotiation standards live in the marked-up drafts the partners have accepted or rejected. Client intake patterns live in the emails and matter notes the intake team has written for years. All of that content exists inside the firm. Very little of it is captured, structured, or annotated in a way that lets an agent read it, learn from it, and run against it. The Gap Zone is the distance between owning the raw material of the playbook and having the encoded version that can actually be operated.

    Kirkland’s Secret Sauce Litigation initiative is a public bet that the distance is closable and worth closing. HSF Kramer’s proprietary system of intelligence is the same bet in different words. Spellbook’s Autonomous Contract Management is the vendor version of the same bet, aimed at firms that will not build their own but still want the agent to run against their history rather than a generic template. Reed Smith’s Cornell program is the training version of the bet, aimed at making sure the partners who lead the codification work know what the work is.

    The practical question for a firm leader is which version of the bet the firm is running. There are three postures worth naming. The firm can build its own system with a technical partner, in the Kirkland or HSF Kramer mold, which requires a serious capital commitment and internal owner. The firm can adopt a vendor system that runs against firm history, in the Spellbook mold, which requires the firm to have its history organized enough to be usable. Or the firm can defer the decision and continue running on generic tools, which is the option that will look most reasonable in the short term and produce the least differentiation over time. All three are valid choices in the sense that all three are real strategies. Only the first two produce an encoded playbook. The third produces continued exposure to whichever platform’s default settings happen to be running that quarter.

    The Gap Zone is not the absence of tools. Tools are everywhere. The Gap Zone is the absence of the encoded practice that makes the tools produce the firm’s work rather than someone else’s average.

    What This Means For Your Firm

    Three implications for any firm running an AI program through the back half of 2026.

    First, the vendor autonomy move is going to spread quickly, and the terms on which vendors offer autonomy will matter more than the fact of autonomy itself. Spellbook is training its agents on your firm’s negotiation history. The next generation of similar products from other vendors will do the same for other practice areas. The specific question worth asking about each is where the training data lives, who owns the encoded output, and what happens to your firm’s methodology if you leave the platform. Autonomy is coming. The commercial architecture of autonomy is what to negotiate.

    Second, the firm-side codification move is the direction of travel for firms with the scale and appetite to build their own systems. Not every firm needs to be Kirkland. Every firm needs to have inventoried its playbooks in a form that could, in principle, be encoded. The inventory itself is the strategic asset. It is what a vendor partner will build against, what a lateral hire will contribute to, and what a client will find credible when the firm claims a differentiated approach to a matter. Firms that have not started the inventory will find themselves competing on tools while their peers compete on encoded practice.

    Third, the regulatory backdrop will not force any of this. The Colorado reset is a signal that the compliance floor is lower and less standardized than the industry expected six months ago. That signal cuts two ways. It reduces the immediate compliance burden. It also removes the external pressure that would have pushed firms toward a common standard for how encoded practice gets documented, audited, and governed. Firms that codify well will build durable advantage. Firms that do not codify will fall behind quietly, matter by matter, without a regulator ever showing up to name the gap.

    The playbook is the asset. The encoding is the work. The firms that treat both as strategic priorities for the rest of 2026, and invest with the discipline the work requires, will be the ones with something to show for it by the end of the year.

    Laura Kennedy and Chelsey Lambert are Co-CEOs and Co-Founders of Legal Tech Decoded. Rising Above the Noise lands every Monday at legaltechdecoded.com.

    The Daily Brief

    Every morning, the most important legal tech developments — curated and context-rich — delivered before 8 AM.

    Free. No spam. Unsubscribe any time. Read by 28,000+ legal professionals.