Data law, practiced
one billable hour at a time
like software.
A boutique data-law firm run on an AI-native operating model, putting legal advice into working software since 2015, a decade before the market had a name for it. The firm didn’t pivot into this work; it started there.
- 25 years in data law
- Legal advice in software since 2015
- AI-first, human-validated
You can complain about robots practicing law, or you can build the robots. We're building the robots.
Jeffrey C. Sharer — 2017
That line is nine years old. That’s the point.
Ten years of this, before it had a name.
AI-native conversation is two to three years old. This practice has been answering “can machines deliver legal judgment, defensibly?” since before generative AI existed as a category.
First, a TRS-80
A searchable baseball-card database, built as a junior-high project. He was building this before the law had a word for it.
Nearly two decades at an AmLaw 10 firm
Partner at Sidley Austin; E-Discovery Task Force; working at the earliest edge of predictive coding and TAR. That was machine classification at a time when the profession viewed it with skepticism, to put it mildly.
The Akerman Data Law Center
Founder and co-chair of the Akerman Data Law Practice. The Data Law Center launched in December 2015 with Thomson Reuters and Neota Logic, encoding the breach-notification law of 50+ U.S. jurisdictions so the same analysis reached every one of them, and returned an answer in minutes. The press called it “TurboTax for law.” He called it collaborative disaggregation, a term later analyzed in legal academia.
The press caught up
Bloomberg Law placed it among the earliest expert-system deployments in law that actually shipped. The American Lawyer reported that no prospect had ever declined a demo. The Financial Times ranked Akerman #20 in 2016, with the Data Law Center among the top four in the FT’s first “Innovation in Collaboration” category. Fastcase 50 followed in 2017.
A law firm and a software company on one operating system
Sharer Law practices alongside partner firm LexShift, which builds the retention and classification tooling the practice runs on. One operating system, and one standard: if it can’t be defended to a regulator, it doesn’t ship.
Agentic practice, in daily use
Agents, model councils and persistent memory are in daily use here, described plainly, with real engagements and real accuracy numbers, including the places it fell short. “Order from Chaos,” co-presented at ARMA InfoNext 2026, walked three live AI-classification engagements end to end.
We spend more time in GitHub than in a legal DMS.
The center of gravity has moved from a traditional firm’s document management system to a version-controlled, systems-oriented engineering environment. It’s still a law practice, run by a 25-year data lawyer. That combination is the whole point of the firm.
Every engagement opens with the same question: what should the machine do here, and what should it not? Human judgment stays on what needs it: governance decisions, edge cases and the final read.
Every machine placement carries a confidence score, and that score sets how much lawyer review it gets, never whether it gets any.
Nothing ships on vibes; everything ships with a number attached.
Where the models disagree, you get a lawyer’s judgment rather than a majority vote. Several frontier models argue the question first. Agreement narrows a question; it doesn’t settle it.
Every change to your deliverable is attributable, reviewable and reversible, down to the line and the reason.
A lawyer’s redline and an engineer’s diff are the same instrument.
Context carries between sessions as durable state, so an engagement resumes where it stopped. Nobody reconstructs your environment from memory each time, and you’re not charged to re-explain it.
AI doesn’t fail — poorly scoped problems do.
Most of the work is deciding what to hand the machine and what to keep. Scoped well, common workflows return 2–10× the output for the same effort against our own prior baseline, which buys coverage rather than haste: the whole population reviewed instead of a sample. Scoped badly, they produce confident nonsense a lawyer then has to unpick.
What clients bring us.
Data Law & Privacy
The full data lifecycle: privacy compliance, cross-border transfer, breach-notification analysis. The domain where this practice first encoded legal advice into software.
Information Governance
Policy and the data challenge treated as one problem: frameworks, defensible disposition, and retention programs built to hold up when someone pushes back.
Global Records Retention
Retention schedules that hold up across 330+ jurisdictions, maintained continuously, because a retention schedule goes stale the quarter after it’s written.
AI Governance & Defensibility
“Great, it’s fast — now defend it to a regulator.” Human-in-the-loop architectures, confidence gating, accuracy validation and traceability.
E-Discovery & Litigation Readiness
Predictive coding and TAR since their earliest days; litigation preparedness and defensible collection. This practice first put machine classification in front of a court.
Jeffrey C. Sharer
Founder & PrincipalTwenty-five years at the nexus of law, technology and data. Roughly two decades at Sidley Austin, as a partner. Founder and co-chair of the Data Law Practice at Akerman, where he built one of the profession’s first expert systems for legal advice in 2015 and coined the term collaborative disaggregation.
Today he’s Founder and Principal of Sharer Law and CEO of LexShift. He builds the tooling, uses it daily, and can answer the defensibility question without translating, because he’s been answering it for a decade.
Representative matters
- Planned and managed through successful pilot a defensible disposition initiative for a Big Four professional services firm across a petabyte-scale email repository, covering both the legal advice and the technology solution used to execute it.
- Built from scratch a global records retention schedule, its delivery platform and related policies for a Fortune 200 multinational investment company, with the information-governance and privacy advice underneath it.
- Assessed the information-governance program of a Fortune 500 multinational manufacturer against multiple industry maturity models, and made the recommendations that followed.
- Advised a global online learning company on regulatory tracking and built the bespoke solution supporting it for the whole 40-person legal department.
- The American Lawyer
- Bloomberg Law
- ABA Journal
- Financial Times
- Fastcase 50
Bring us the problem the billable hour couldn’t solve.
Usually that’s a scope problem: more jurisdictions, more systems and more documents than review by hand can reach. We didn’t move. The market did.