← Back

Guide · Published November 14, 2026 · Reviewed November 14, 2026 · Ingenious AI team · Not legal advice

← All guides

Getting Started With the ADMT Notice Kit: A First-Week Plan That Ends With Real Notices

The short answer: A working ADMT programme can be stood up in a week if the order is right: Day 1 — run the free duties checker for every state-decision combination you operate and file the results. Day 2 — inventory the tools those combinations involve. Day 3 — build the business profile and generate the first real notices in the free generator. Day 4 — stand up the explanation, correction, and review workflows before the first adverse outcome starts a clock. Day 5 — set the record habits (exports, verification, the monthly audits) and the forward calendar (October 1, 2027; December 2, 2027). What follows is that week, day by day, with the deliverable that proves each day happened.

Check your duties and dates — free Build your first notice — free

What is this plan? It is the onboarding sequence assembled from this guide series' dependency order — scope before assessments, assessments before notices, notices before response workflows, record underneath — compressed into five working days for a business whose inventory is of ordinary size. Larger operations should run the same order over more days rather than re-ordering it; the sequence, not the speed, is the content.

Day 1: the checker sweep and the coverage file

List the combinations first: every state you operate in, every decision type you make about people, and — honestly, from workflow knowledge — each tool's role. Run each combination through the checker and file every result, including the 'no duties found' and 'check the list' verdicts, with the date. Where a verdict is conditional, the condition becomes a Day 1 research task, not a footnote: the statutory lists are printed on the result for exactly that work.

In parallel, open the coverage file per state: California's thresholds measured, Connecticut's triggers measured, Colorado's use test noted. Deliverable: a duty matrix draft — states down the side, decision types across, a checker-backed finding in every cell. Every later day refines that matrix; nothing later substitutes for it.

Day 2: the inventory, honestly scoped

Run the five-pass inventory for the tools inside the matrix's scope: procurement, workflows (backwards from each decision), data flows (forwards from each person's data), shadow tools under amnesty, and the register itself with one row per tool — name and trade name, decisions, states, weight, data categories and sources, notice status, record location, review date. Day 2's deliverable is the register with a duty set or a documented negative on every row.

Vendor questions go out on Day 2 as well, because their answers have lead time: the twelve questions, in writing, to each in-scope tool's vendor, with renewals noted as leverage. The answers will land across the following weeks and complete the documentation file the assessments need.

Worked example: Elm Creek Family Services

Example — fictional business, real duty structure. Elm Creek, the fictional services provider from the export guide, onboarded in one week. Its five deliverables, as filed:

Day 3: profile, first notices, placement test

Build the business profile — business name, tool names, review contact, correction contact, all current and monitored — and generate the first notices for each in-scope decision type and state. Read each one twice (as the business, then as the person), fix failures at the profile or workflow level, and place the notices at their points of interaction in the real flows.

Close the day with the applicant phone test on each flow: notice encountered pre-use, contacts working, opt-out completable where California requires it. Deliverable: live notices, generating per person and logging, with the placement results filed. If assessments are owed (California; Connecticut profiling), Day 3 also drafts them from the inventory's data pass — they must be complete before the flows they govern run under the new notices.

Day 4: the response machinery, before it is needed

Stand up the four response workflows with owners and rotas: explanations drafted from reasons captured at decision time (and the capture itself switched on at the source — today's decisions are the first whose reasons exist), corrections with the re-run step, the review queue with named reviewers holding real authority and a deputy, and access answering assembled from the record. Dry-run each workflow once on a test case.

Deliverable: four workflows that have each completed one test case end-to-end, with their logs showing it. Day 4 is the day most programmes skip and most audits wish they hadn't: the first real adverse outcome starts Colorado's 30 days without consulting your readiness, and the queue that exists only as an intention will be designed, badly, under that clock.

Day 5: record habits and the forward calendar

Set the standing habits: the monthly audits (opt-out reconciliation, review metrics, correction patterns), the quarterly chain verification and export sample with the stranger test, the register review cadence, and the re-test triggers for every negative finding and exception position. Run the one-hour test on a Day 3–4 case as the week's acceptance check — the full file, produced inside the hour, is the proof the system coheres.

Close with the calendar: January 1, 2027 readiness state recorded; Connecticut's October 1, 2027 notice pair diarised against deployment dates; the EU December 2, 2027 gate set if Annex III areas apply; Colorado's AG-rules watch noted. Deliverable: a one-page programme state — matrix, register, live notices, staffed workflows, habits, calendar — that next quarter's review can diff against reality. Then the programme runs itself at the cadence you set, which is what the plans are priced to support.

  1. Day 1 — checker sweep for every combination, coverage files opened, duty matrix drafted with a filed finding per cell
  2. Day 2 — five-pass inventory into a register with duty sets or documented negatives; twelve vendor questions sent in writing
  3. Day 3 — profile built, first real notices generated and placed, applicant phone test passed on every flow
  4. Day 4 — explanations, corrections, reviews, and access workflows staffed and dry-run before the first live clock starts
  5. Day 5 — audits, verification, and review cadences set; one-hour test passed; the 2027 forward calendar diarised with owners

Sources, as carried in the kit's own data: the duty summaries in app/rules.py and the checker data in app/checker_data.py, verified against primary or authoritative sources on October 28, 2026. Run the free duties checker for your jurisdiction, decision type, and the tool's role, and compare the kit's plans for the record, builder, and review queue. More from Ingenious AI is on the hub.

Build your first notice — free

Related guides

See all ADMT guides for the full set.

Frequently asked questions

How long does it take to get started with the kit?

The sequence in this guide runs in five working days for an ordinary-size inventory: checker, inventory, profile and notices, response workflows, record habits. Larger operations should stretch the days, not re-order them.

What should we do first — the checker or the inventory?

The checker sweep first, at combination level, because it defines the scope the inventory must cover and produces the duty matrix every later step refines. The inventory follows on Day 2, inside that scope.

When should the explanation workflow be built?

Before the first adverse outcome under the new programme — Day 4 of this plan. Colorado's 30-day clock counts from the decision, and reasons must be captured at decision time, so the workflow and the capture precede live volume.

What proves the first week worked?

Artefacts, not activity: filed checker results, a register with findings per row, live notices that passed the phone test, four dry-run workflows, and a one-hour test passed on a real first-week case.

What tier do we need to start?

Start on the tier your register supports — Free for Colorado-only, one decision type; Pro when a second state, the builder, the queue, or exports become real; Team when review becomes a multi-seat rota. The plan comparison is in the pricing guide.

Build your first notice — free