Insights

• The AI Guide for Legal Practice

  A five-part, practical series for building AI that keeps privilege and leaves a file note: skills, connections, plugins and agents, prompting, and the infrastructure underneath it all. Written for law firms and legaltech.

• The AI Field Guide for Regulated Finance

  A five-part, practical series for building AI you can put in front of a regulator: skills, connections, plugins and agents, prompting, and the infrastructure underneath it all. Written for fintech, wealthtech and fund operations.

• Everyone Has a Prompt. Nobody Has a Method.

  Three in four firms are using AI. Fewer than half fully understand what they are running. The gap is not the prompt, it is the operating model. Here are the seven decisions that turn visible AI activity into an owned, measurable capability.

• The Company Brain: The Layer That Stops Every AI Tool Plateauing

  Every AI tool hits the same ceiling: it does not know your business, and nothing you teach it survives the next tool. The firms pulling ahead are not running more AI. They run it on one layer of context that never resets.

• Who Owns the Wait? The Coordination Tax in Regulated Work

  80% of firms say AI improved personal productivity; only 37% see any EBIT impact. The value is between the steps, in the handoffs nobody owns. In regulated work, that coordination tax is really a proof tax.

• In Regulated Work, the Coordination Tax Is a Proof Tax

  In regulated finance and legal the coordination tax is a proof tax. Handoffs stall for lack of evidence, not judgement, so the agent must emit its record as it runs.

• Five Tests for Your First Company-Brain Use Case

  The first use case decides how hard the second one is. Five tests separate a good first build from an expensive false start. Sequencing is the risk, not the technology.

• Governed Workflows: The Eight Stages That Make a Method Repeatable

  Risk does not enter when someone opens a tool. It enters when an output moves to the next stage of work. Eight stages that turn a prompt into a governed workflow.

• Verify, Reconcile, Route: What an AI Agent Does at a Handoff

  An AI agent at a handoff does three things: verify, reconcile and route. Here is how each maps onto a regulated finance or legal workflow, and why it never routes straight from the model.

• Buy, Rent, or Own: The Three Decisions Behind a Company Brain

  Three forks decide what a company brain becomes. Most firms answer all three by default. Only the first is really about strategy. The other two set timing and staffing.

• The 80/37 Gap: Why AI Lifts Productivity but Not Enterprise Value

  80% of firms say AI raised personal productivity but only 37% see EBIT impact. The coordination tax sits between the steps, and cost programmes only cut the visible 7%.

• Where AI Earns Its Place, and Where a Rule Is Safer

  Task shape decides suitability and cost of error decides control. Five task types, five routes, and eight screening questions for placing AI where it belongs.

• Activity Is Not Capability: The Six Levels of AI Adoption

  Licences, active users and time saved measure use, not adoption. Six levels of maturity, why firms stall at level four, and what leaders actually need to understand.

• Why Every AI Tool Hits the Same Ceiling

  Every AI tool arrives capable in general and blank about you. The ceiling is the same because the context never carries across. The fix is a layer you own.

• The Frontier: Attestation, Knowledge Graphs and the Infrastructure Under Finance AI

  The frontier of finance AI is no longer the model. It is attestation, knowledge graphs and the harness underneath, plus the one problem nobody has solved yet.

• The Frontier: Attestation, Knowledge Graphs and the Infrastructure Under Legal AI

  The frontier of legal AI is not the model. It is attestation, knowledge graphs and the runtime infrastructure that lets a third party verify what an agent did, offline.

• Plugins and Agents: Packaging AI Your Finance Firm Can Deploy

  Packaging AI for finance means bundling skills, commands and connectors into one installable plugin with its permissions attached. Here is how to deploy it safely.

• Prompting for a Defensible Answer in Regulated Finance

  Prompting for regulated finance is about getting an answer whose defects are visible, not the most confident one. Six moves and a bench card to make a draft checkable.

• Plugins and Agents: Packaging a Practice's Playbook for Deployment

  Legal AI plugins and agents package a practice's skills, connectors and playbook into one installable, versioned unit, so a colleague can run it with its permissions attached.

• Prompting for a Checkable Answer in Legal Work

  Legal AI prompting is not about a clever answer. It is about a checkable one, whose sources are traceable and whose gaps are labelled, so a qualified lawyer can verify it fast.

• Connecting AI to the Systems You Already Govern: MCP for Finance

  MCP for finance wires Claude into governed systems so the answer comes from the record, not an export. Here is what a connector preserves and how to sequence it.

• AI Skills for Finance: Capturing a Process That Runs the Same Way Every Time

  AI skills for finance capture a process so it runs the same way every time and leaves an audit trail. Here is how to build one your compliance team can defend.

• Connecting AI to Your DMS and Docket Without Losing Privilege

  Connecting AI to your DMS, contract repository and docket keeps matter context, entitlements and privilege intact, where a downloaded export throws all of it away.

• AI Skills for Law Firms: A Playbook That Runs the Same Way on Every Matter

  AI skills for law firms turn a clever prompt into a captured playbook that runs the same way on every matter, protects privilege, and leaves a file note you could show a regulator.