AI for Law Firms and Legal Departments
I help law firms and legal departments put AI to work in their legal operations, one workflow at a time. I map how the work really moves, then design and implement workflows your lawyers will trust.
The line that never moves: the agent proposes, the lawyer decides.
From scattered work to workflows that run
Assessment and Roadmap
Legal work is often scattered across apps, tracked by hand and interrupted all day. I start by finding where the time goes.
Includes:
- Legal operations assessment
- Inventory of tools, templates and playbooks
- The highest-volume workflows first
- A roadmap with clear priorities
- Baseline measures before any change
Workflow Design
Every matter runs through the same four steps. I design each workflow around them, with the lawyer at the decision.
Includes:
- Input: the record gathered into one cited summary
- Gap analysis against your templates and playbook
- Reconciliation: the lawyer decides
- Output: drafts as suggestions the lawyer owns
- Contract review, privacy requests, legal gap analysis
Controls Lawyers Can Trust
Four controls are built into every workflow, so supervision is part of the design and not an afterthought.
Includes:
- Never sends, signs or approves
- Never edits, only suggests
- Never asserts without a source
- Never reads across matters
- Review procedures and playbooks in writing
Implementation and Adoption
One workflow goes live for one team, is measured, and only then is extended to the next.
Includes:
- AI tool selection and vendor negotiation
- A pilot with one team, measured in your numbers
- Team training and adoption support
- Rollout workflow by workflow
- Ongoing regulatory watch and maintenance
Questions law firms and legal departments ask
In three of the four steps every matter goes through. AI can assemble the input (the record, the thread and the files in one cited summary), run the gap analysis (the draft compared with your templates, playbook and precedents), and prepare the output as suggestions. The third step, deciding what to accept, push back on or escalate, stays with the lawyer.
In my view, no. AI can take over much of the assembling and comparing that used to fill a lawyer's day, but the practice of law stays human. The useful question is where the human stays in the loop, and that is what I design for.
With four controls built into every workflow. The agent never sends, signs or approves. It never edits, it only suggests. It never asserts without a source. It never reads across matters. Each lawyer reviews and owns what leaves the team.
I work one workflow at a time so that you see a result before committing to more. A first workflow is mapped, piloted with one team and measured against how the work was done before. The timeline depends on your tools and on how much of your playbook is already written down, so I set it with you after the assessment.
Not necessarily. I start from the tools and licences you already have and from what your security team will approve. Where a new tool is justified, I help you select it and negotiate its terms, including how your data may be used.
