Guide · Published November 14, 2026 · Reviewed November 14, 2026 · Ingenious AI team · Not legal advice
The Free ADMT Notice Generator: How the /try Flow Builds Your First Notice
The short answer: The free notice generator produces a complete pre-use notice — business, tool, decision kind, correction contact, review contact — from the same document builder (app/documents.py) that the account side uses, in the same shape the guides in this series dissect. The Free plan behind it covers Colorado, one decision type, and the core record log. This walkthrough covers what the flow asks for, what the generated document contains line by line, and the checks to run on your first output before anything like it goes to a real person.
Check your duties and dates — free Build your first notice — free
What is the free generator? It is the kit's working front door: a no-account flow that builds one real notice from your inputs, so the document can be evaluated as a document — read as your applicant will read it — rather than from feature lists. The drafting standard it implements is the five-part structure in How to Write an ADMT Pre-Use Notice.
What the flow asks for, and why each field exists
The inputs mirror the notice's working parts. Business profile facts: your business name, the tool's name, the review contact (name and email), and the correction contact — the fields every generated document draws on, so no notice ships with a placeholder. Case facts: the person's name, the decision type (employment, housing, lending, education, or other consequential), and the state, which selects the law line and rights wording the builder composes.
Nothing decorative is asked, because nothing decorative is printed. If a field feels burdensome — finding the tool's exact trade name, say — that burden is the duty's, not the form's: Connecticut's written notice requires the trade name expressly, and Colorado's notice names the tool. The generator simply refuses to let the gap pass silently, which is the point of generating from a profile instead of editing a sample.
What comes out: the document, section by section
The output is the template the Colorado guide dissects: the heading (notice about an automated decision tool), the from/to/date block, the opening sentence carrying business, tool, and decision, the 'what this means' lines stating the tool's role and the consequence, the rights section naming the state law with the correction and review paths and their contacts, the questions line, and the standard footer — generated with the kit, not legal advice, records what was said and when, does not judge fairness.
Read the generated document twice. First as the business: is the role sentence true of our workflow as it actually runs? Second as the person: after one read, can I say what tool, what decision, what I can do, and whom I contact? The second reading is the compliance reading — the duties are owed to that reader. Any sentence that fails either reading is edited in the profile or the workflow, not in the exported document.
Worked example: Juniper Row Rentals
Example — fictional business, real duty structure. Juniper Row, the fictional landlord from the housing guide, evaluated the kit starting from the free generator rather than from a sales call. Its first session:
- Profile built in minutes: business name, screening tool name, review contact (the property manager), correction contact (the office inbox) — the four facts every document needs, entered once.
- First notice generated: for a test applicant, housing decision, Colorado — the output read back to the team twice: once for workflow truth (the role sentence adjusted to describe the screener's actual weight), once as the applicant.
- Gap found immediately: the correction contact inbox was unmonitored on weekends — a profile fact the generator exposed before any real notice depended on it; a rota is set.
- Tier decision made from use: Colorado-only, one decision type (tenant screening) fits the Free plan's stated shape; the firm starts there, with Pro noted as the step if its Connecticut portfolio activates the second state.
- Record started: the first real notices issued the following week are generated per applicant and logged — the test notice's chain discipline, now on live cases.
What the free tier is, and where its edges are
The Free plan's stated shape — Colorado, one decision type, the core record log — is a real operating configuration for a single-state, single-decision operator (the small landlord of the deadlines guide is the archetype). Its edges are equally stated: a second state or the full builder/queue/export set is the Pro tier ($49/month), and multi-seat review is Team ($99/month). The pricing page carries the authoritative version of all three.
The generator itself is the evaluation instrument for those edges: build notices for your actual decision type, read them as your people will, and the question of which tier your volume needs usually answers itself. The plan comparison is worked through in ADMT Pricing Plans Compared.
From first notice to operating programme
A generated notice is the seed of the record discipline: generated per person, dated, and logged, it becomes the first entry in the decision's chain — the chain that later carries the explanation, any correction and re-run, and any review. Businesses that start by generating real notices for real cases (rather than approving a template in the abstract) discover their profile gaps in the first week, when they are free to fix.
The natural next steps in order: run the free duties checker for your state and decision type to confirm the duty set the notice belongs to; complete the inventory row for the tool; set the explanation workflow in place before the first adverse outcome arrives, because Colorado's 30 days counts from the decision, not from your readiness.
Accuracy notes: what the generator will not do
The generator builds the document from your facts; it does not verify them. If the tool's role in your workflow is effectively decisive, the builder's balanced role sentence is yours to correct — the honesty standard belongs to the deployer, per the token-review analysis. It does not decide whether your use is covered (that is the checker's analysis, and ultimately advice on your facts), and it does not send anything anywhere: generation and sending are separate acts, and the person-facing send remains yours.
Those limits are features of a templates-and-records tool, stated on the documents themselves. Within them, the generator does the one thing this series keeps asking for: it produces the real artefact, in your facts, readable cold — which is more evaluation value than any feature matrix provides.
- Build the profile completely before generating: business, tool trade name, review contact, correction contact — no placeholders survive generation
- Generate for your real decision type and state, then read the output twice — as the business for truth, as the person for comprehension
- Correct failures in the profile or the workflow, never by hand-editing the exported document
- Confirm the duty set behind the notice with the free checker, and inventory the tool the same week
- Choose the tier from your actual shape — states, decision types, seats — using the pricing page as the authority
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.
Related guides
- How to Write an ADMT Pre-Use Notice People Actually Understand
- ADMT Notice Kit Pricing: Free, Pro ($49), and Team ($99) Compared
- Colorado AI Act Automated Decision Notices: What to Send
- Adverse-Outcome Explanations in Plain Language
See all ADMT guides for the full set.
Frequently asked questions
What does the free generator produce?
A complete pre-use notice from the same builder as the account side: business, tool, decision kind, the tool's role, the state-law rights section with correction and review contacts, and the standard not-legal-advice footer.
What is included in the Free plan?
Colorado coverage, one decision type, and the core record log, per the pricing page. Pro ($49/month) adds all three states, the explanation builder, review queue, and exports; Team ($99/month) adds multi-seat review.
Do I need an account to try it?
The /try flow is the free front door for building a first notice. The account side is where the record, tracker, and queue operate.
Will the generator check whether my use is covered?
No — that analysis belongs to the free duties checker (jurisdiction, decision type, the tool's role) and, on your facts, to advice. The generator builds documents from the facts you enter; it does not verify coverage or your workflow's truth.
Can I edit the generated notice?
Fix the inputs, not the output: profile facts and workflow truths are corrected at the source so every future document inherits the fix, and the generated document in the record stays the document that was actually produced.