Inside the Hub-and-Spoke

Inside the Hub-and-Spoke: How Attorneys at SLW Build Their Own Tools

Most law firm AI initiatives are central teams shipping tools to the rest of the firm. SLW Labs is structured the opposite way, and the structure is the point.

There is a familiar pattern in how large organizations approach internal innovation. A central team forms, with a budget and a mandate. The team builds tools and then tries to roll them out across the rest of the organization. Adoption is uneven. The team becomes a bottleneck. The most distinctive use cases, the ones that would justify the investment, are precisely the ones the central team does not have the domain expertise to address.

SLW Labs is not structured that way. We chose a hub-and-spoke model from the start, and the choice has shaped what we have been able to build.

The Hub

The hub is a small core team. Its job is not to ship products. Its job is to provide a development environment that is safe and fast for everyone else to ship products. That means three things in practice.

First, SLW Sandbox, the shared infrastructure. Sandbox is what the core team maintains, and it is the foundation every Labs product depends on. Authentication, audit logging, matter isolation, retrieval, the security primitives, the integration libraries (USPTO and EPO feeds, scientific OCR pipelines, the firm’s docket structure, case law lookup), and reusable components for citation handling, and prosecution-record retrieval, all of it lives in Sandbox. None of the infrastructure needs to be rebuilt for each product. New products can inherit the infrastructure on day one.

Sandbox is the structured program through which attorneys participate in Labs. A tiered model, Explorer, Builder, Deployer, provides graduated access to the platform. Explorer-tier attorneys begin with Replit or Microsoft Copilot Studio (already available to every SLW attorney through the firm’s O365 subscription), Claude.ai Enterprise, and the Power Platform, no terminal access required, no additional licensing. Builder-tier attorneys add Claude Code or Cursor along with per-project LLM API keys scoped to specific projects for cost visibility. Deployer-tier projects gain access to the controlled Tier 2 sandbox for matters that need real client data and a path to production deployment.

Second, standards. Every Labs product passes the same review gates before it ships: security review, governance review, attorney-experience review. The standards do not flex by product.

“The standards are the most important thing the hub provides. A spoke can move quickly precisely because it does not have to re-litigate authentication, audit logging, matter isolation, or governance for every new tool. Those are settled. What the spoke focuses on is the workflow that only patent attorneys understand.”
— Nathan Elder, SLW Labs

Third, SLW’s institutional scaffold. What makes the program work is the institutional scaffold around the tooling: a starter template repository in the firm’s GitHub organization, secrets management for per-project API keys, a curated prompt library, a project registry, and a monthly Demo Day where attorneys share what they have built. Without Sandbox, every spoke would have to assemble its own foundation. With it, spokes start with the platform already in place and focus on the workflow that only patent attorneys can design.

The Spokes

The spokes are the attorneys and technologists across the firm who build domain-specific workflows on top of the platform. ARTY’s drafting and validation pipeline. SimProf’s training scenarios. The client portfolio analyses that have run for some of our key clients. None of these came from a central innovation team. They came from people closer to the work.

This matters for two reasons. First, the people closest to the work see the right opportunities. ARTY’s antecedent basis checker is the kind of feature you build when you have personally reviewed thousands of claim sets and noticed where they fail. Second, the people closest to the work are the people clients trust to ship something on their matters. ARTY’s adoption inside the firm is real because attorneys recognize their own standards in how it operates.

The External Development Partner

SLW Labs works with an external development partner under firm direction, particularly on integration with the firm’s existing systems and platforms. Building a platform like ARTY requires deep software engineering, and the partner extends our capacity to ship cleanly without forcing the firm to staff a full internal engineering organization. The partner operates within the same security, governance, and review gates as everything else in Labs. They do not own the products; they help us build them.

What This Looks Like In Practice

A new product idea inside SLW Labs typically follows a path: an attorney or technologist identifies a problem, writes up a short proposal, gets feedback from the core team and from the security and IT functions, builds a prototype on the shared platform, and runs it on real matters with attorney supervision. If it works, it stays. If it does not, it gets pulled. Tools can be replaced weekly if the workflow improves.

This is why the SLW Labs product list moves. Tools graduate from planning to development to production. Some prototypes get folded into other products. The portfolio is alive, not fixed.

Why This Is The Right Model For A Patent Firm

Patent prosecution is a craft. The expert judgments that make a strong patent are made by attorneys and patent agents who have done the work for years. A central innovation team, however well-intentioned, cannot encode that craft on its own. The hub-and-spoke model is how we let craft and engineering meet on the same platform.

It is also how we keep moving. The firms that will lead the next decade are the ones that have built the muscle to ship internal tooling continuously. SLW Labs is the muscle.

In the next article in this series, we will turn to ARTY specifically, and to the quality validation architecture that runs alongside drafting. That is where the choice to build, rather than borrow, becomes most visible.