I build operational infrastructure that allows teams to scale.

The Operational Challenge

Our org managed over 30,000 regulatory information requests per year coming in via highly fragmented channels (PDFs, paper mail, emails, screenshots). Triaging was entirely manual, taking critical SMEs bandwidth away from high-judgment issues. This manual drag led to severe operational bottleneck risks in overall speed and accuracy.

Phased Roadmap & Execution Strategy

Key Strategic Decisions & Tradeoffs

• Middleware Procurement: Selected Workato over a solution that was more tailored for legal to minimize maintenance, technical debt, and budget impact, allowing our lean team to focus on our core work.
• Safe Sequence: Implementing stable and auditable routing paths before deploying data extraction modules. This allowed operational partners to build full trust in the automated framework sequentially.

The Long-term Impact

Over 1,600 hours saved in production to date, with a verified roadmap driving over 5,750 hours of annual staff recovery across 8 separate stakeholder business units.

The Problem

The incumbent spend management system failed to deliver on basic processes, such as invoice review and accruals intake. This friction caused mounting vendor maintenance costs, and severely limited our spend-remediation workflows, creating a direct multi-million-dollar impact.

Execution Strategy

I authored the business case recommending the early contract termination of the legacy system and immediate migration to Brightflag. To mitigate extreme schedule risk, I built and negotiated a contractually-enforceable 100-day implementation schedule detailing penalties and bonuses aligned to technical milestones.

I managed a matrix of 10+ internal departments and co-led integration sprints for Okta, NetSuite, and Snowflake data syncs.

Direct Outcomes

• 60% Timeline Compression: Successfully built and shipped the replacement platform in exactly 100 days (vs. the 250+ days required to build the original legacy system).
• $5M Target to be Realized: Positioned the platform to capture and enforce automatic AI invoice guidelines, pacing to recoup $5M in annual vendor spend leakage.

The Challenge

Executing high-volume customer records requests required internal analysts to navigate 14+ siloed administrative search tools and legacy databases. Search pipelines averaged 15 minutes per query, suffered from strict case-sensitivity errors, and lacked consolidated user data compliance controls.

Approach & Strategic Decisions

I partnered directly with the data engineering team to construct the search engine and interface on top of their existing robust compliance platform.

I authored the requirements and strategy documents to drive alignment across Compliance, Security, Data, and Legal. The platform normalized multiple irregular inputs (dates, partial identifiers, transaction hashes) into an access-controlled, ranked-result user interface.

The Impact

Consolidated 14+ disparate administrative databases into a single clean interface. Saved over 1,000 internal team hours within the first 12 months of deployment.

The Capacity Problem

Our organization faced an ambitious 70% expected growth in contracts to review with a strictly capped, flat internal team headcount. High-judgment commercial agreements required intensive mechanical, line-by-line comparison reviews. Prior attempts using general in-house tools failed because models continually replaced whole paragraphs rather than executing tracked-change, word-level surgical edits.

The Solution

I structured this as an operational resource and tool vetting challenge. I wrote the detailed job specifications, hired, and now direct an AI Legal Ops Contractor to serve as my primary technical resource on the build.

To bypass marketing demos, I engineered a **"Gold Standard Test"** benchmark: I anonymized a live contract and playbooks, ran 10 counterparty drafts against 6 vendor models, and blindly graded their output against our legal experts' actual negotiation edits. We killed our initial pilot due to the discovered limits, and selected 2 vendor finalists for parallel in-production sandboxes.

Projected Impact

Targeting 1000+ professional staff hours saved annually. Also secured $61K in immediate operational hiring savings by sourcing a highly qualified specialist resource under budgeted scope.

The Operational Landscape

Despite managing $$$ (millions) in annual external services across multiple operational channels, legal leadership had no consolidated data on invoice cycle times, operational bottlenecks, vendor concentrations, or general spend metrics for their teams.

Designing the Stack

I defined and led the data requirements, partnering with Data Engineering to land the spend platform data directly inside Snowflake. Data Eng mapped and modeled the schema joins, then ** I personally designed and configured the production dashboards in Looker**. These included:

• Platform Health Dashboard: Tracked operational metrics, invoicing errors, and automated routing health.
• CLO Dashboard: Surfaced high-level spend analysis and vendor concentration metrics.
• Team Specific Suite: Provided department heads with their matter level spend in real time.

The Outcome

Successfully replaced opinion-driven decision-making with objective, clean data pipelines. This dataset became the foundational baseline that proved the $5.0M business case for the subsequent Brightflag platform migration.

Operational Context

ThemisOS is my personally-branded suite of LibreChat agents for Legal — Themis (Greek goddess of law and justice) + OS (operating system). I branded this as "your personal legal ops team"

Agent Set Up

Five agents live under the ThemisOS banner today, each architected as a specialist with handoffs to other agents and scoped MCP tool access (some with online access, some without). This was a purposeful design choice that improves accuracy and reduces token usage versus a monolithic generalist agent.

Live with the Legal Ops team today, expanding to the full Legal organization this week — timed deliberately to support end-user onboarding to Brightflag

Five Agents

• Head Agent routes request to relevant agent
• Ironclad Agent answers questions about ironclad and evisort.
• Outside Counsel Spend Agent retrieves invoices, budget, and matter info.
• Spend & Vendor Management Process Agent helps user naviagte through matter and vendor management processes.
• Spend & Vendor Jira Agent assists legal ops with faster Jira support ticket completion.