Why Frontier Models?
ExChek is a plugin for the smartest AI in the world — the same Claude, ChatGPT, or Cursor your team is probably already using. This page explains, in plain English, what those tools can actually do when you give them export-compliance superpowers.
Plain English, no jargon
A frontier model is the smartest, most capable AI in the world — right now. Today that means Claude (Anthropic), ChatGPT (OpenAI), and the model behind Cursor. They get measurably smarter every few months. You’re probably paying for at least one of them already.
ExChek runs inside these models. We don’t replace your AI. We make the AI you already have actually useful for export compliance.
The three things that matter
Three words you’ll see thrown around. Here’s what each one does, with a real ExChek example.
01
A regular chatbot answers questions. An agent does the job. It reads your part spec, looks up regulations, checks denied-parties lists, and writes you a memo — from one sentence you typed.
In ExChekYou ask: “Can I ship this headlight housing to Germany?” The agent reads the datasheet, classifies it as EAR99, screens the buyer, checks Germany’s status, and hands you a 1-page memo.
02
Tools are the specific things an agent can reach for mid-task. Pull live eCFR regulation text. Hit the Consolidated Screening List. Generate a Word document. Sign it with a tamper-evident audit log. Each is a tool the agent calls when it needs one.
In ExChekA single classification request can fire 6–10 tool calls: read 15 CFR 774, check the CCL, lookup country list, run CSL screening, build the memo, write the audit-chain entry.
03
Connectors plug the agent into systems you already use — your file folders, Google Drive, your ERP, your email, your part catalog. Once it’s connected, the agent works on your real data, not a copy-pasted snippet.
In ExChekConnect your part catalog once. Say “classify everything we shipped last quarter.” Walk away. Come back to a spreadsheet of ECCNs and a folder of memos.
Other things frontier models can do
These are the features that make Claude (and the others) feel less like “AI” and more like a colleague who happens to read everything ever published, every time you ask.
Memory
Your part numbers. The countries you ship to most. Your house rules (“always escalate anything over $10K to Sarah”). You teach it once.
Skills
A skill is a focused, expert workflow you can drop into Claude. ExChek ships 16 of them — classification, screening, licensing, jurisdiction, and more.
Subagents
When a job is big — classify a 400-part catalog — the lead agent splits the work across helper agents, runs them in parallel, and stitches the answer back together.
Shared workspace
Cowork lets your shipping coordinator, ops lead, and outside counsel all use the same Claude with the same compliance brain installed. Same answers, same audit trail.
Why this matters for export compliance
Export compliance is regulated, citation-heavy, repetitive, and high-stakes. Those are the four things agents are best at. Agents read every word of every regulation, every time. They cite their work. They don’t skip steps because it’s Friday at 4:45.
You don’t need to learn AI. You ask in plain English. You get an audit-ready memo with citations you can hand to your lawyer, your customer, or BIS.
Want to dig deeper?
Security & Safety
Local-first MCP, no telemetry, HMAC-chained audit log, OS keychain. The full security architecture.
Read the pageAccuracy
Live eCFR data, citation-backed determinations, human-in-the-loop on every output. How we keep it honest.
Read the pageUnderstanding costs
Free engine, your existing AI subscription, and the paid plans. The real numbers.
Read the pageDownload the plugin. Upload it to Cowork. Ask your first question. Five minutes, no credit card.