// ProblemSection — Holdpoint homepage problem section // No imports, no build step. Load React + Babel standalone, then this file. // Requires Space Grotesk and Inter (Google Fonts) to be loaded by the page. // Replaces Problem.jsx and ProblemGraph.jsx. // Sized to fit one screen; on shorter screens site.css scales it a step // (see "Problem section" there). function ProblemSection() { return (

Safe or capable. Most firms are told to pick one.

Regulated firms are under pressure to use AI, but the obvious routes fail either the compliance test or the practical one. Keep AI out, and staff do the work by hand or turn to public tools on their own, creating data risk concerns. Connect a model straight to your systems, and client data leaves the firm with outputs nobody can trace.

HOW FIRMS APPROACH AI TODAY
{/* On narrow screens the chart keeps a legible size and scrolls sideways inside the card rather than shrinking its labels to nothing. */}
SAFE AND CAPABLE Where regulated firms need to be SAFE, NOT CAPABLE NEITHER CAPABLE, NOT SAFE How safe it is with client data What AI can do for the firm AI banned outright Pilots stuck in a sandbox Staff using public AI tools Model wired straight into client data

Data scattered across systems

Client information sits across many systems with no shared structure, so AI has nothing coherent to work from.

Client data leaves the firm

Most AI tools work by sending information to an outside model, beyond the firm’s control.

No one can account for the output

AI answers rarely show their source, who reviewed them, or a record of what happened.

); } window.ProblemSection = ProblemSection;