Claude Artifacts: Build a Room Capacity Calculator You Can Verify

An arts center wants a small interactive calculator to help staff plan chairs for workshops. The input is a room’s approved maximum occupancy, a proposed attendee count, and the number of facilitators. Claude Artifacts can create an interactive tool, but a visually polished calculator is not a building-code interpretation or a certified capacity assessment. This hypothetical walkthrough treats the authorized occupancy limit and accessibility requirements as external rules that staff supply, not facts invented by AI. The goal is a testable decision aid with clear assumptions, explicit failure states, and appropriate sharing controls.

Get signed-off constraints before generating code

Find the facility’s current occupancy authorization, accessible seating plan, aisle clearance requirements, and policy for counting facilitators. Do not ask Claude to invent a legal occupancy number from a photograph or room dimensions. Create a tiny test brief: maximum permitted occupants 40, eight chairs reserved in a particular layout, two facilitators who count toward the total, and an example event with 31 attendees. The exact numbers here are illustrative only and not universal regulations. Record who owns the real constraints, the date of the floor plan, and which values the user may change. Inputs need labels and units, not guesses.

Specify the calculation in plain language

Ask for a self-contained calculator with clearly labeled inputs for approved maximum, participants, staff, and optional reserved spaces. Define what each term means. For this scenario, total people equals participants plus staff, and a warning appears whenever the total exceeds the manually supplied occupancy maximum. Reserved accessible spaces belong in the layout review and must not silently reduce or increase the legal limit without an approved rule. Require validation for negative numbers, decimals, missing values, and very large entries. Ask for a visible explanation of the math so a nontechnical coordinator can check the result without reading source code.

Create a Claude Artifact rather than a hidden answer

Claude’s current documentation describes Artifacts as self-contained documents, dashboards, designs, and interactive tools accessible beside a conversation or through the Artifacts experience. Ask Claude to create an interactive calculator and review the preview alongside any editable code or structure. The feature requires the relevant create-and-edit capability enabled in account or organization settings, and certain templates and app connections depend on plan. Avoid instructions that assume every workspace offers the same dashboards or storage features. A useful artifact should display its formula, current inputs, warning logic, and a note that a staff-approved floor plan remains the controlling source.

Run a small truth table independently

Test the calculator with known pairs before using it for planning. With a maximum of 40 and two staff counted, 38 attendees should reach the limit exactly; 39 attendees should exceed it by one. Zero attendees is valid if staff still count; a negative staff entry is invalid. Test a blank maximum, text in numeric fields, and a changed value after the first calculation. Write expected outputs before clicking each example, then compare the displayed calculation with a simple manual addition. If the app fails any boundary case, stop and repair the logic rather than trusting how professional the interface looks.

Make accessibility requirements visible but separate

Headcount arithmetic alone cannot confirm an accessible layout. Add a conspicuous field or note linking to the signed-off seating plan, plus a check for keeping marked paths and reserved positions clear. Do not let the app automatically state ‘accessible’ because the sum is under 40. People need a real layout review including turning space, sightlines, exits, and local rules. If an event has mixed seating arrangements, provide a way to select a previously approved configuration by name, not an AI-generated assertion of compliance. Warnings should say what to verify and who has the authority to approve it.

Check labels, feedback, and keyboard behavior

Test the tool without a mouse: move through controls with Tab, use the keyboard to change numbers, and confirm focus remains visible. Ensure every error appears in text rather than color alone. A red number with no explanation is insufficient for a volunteer who has difficulty seeing the difference. Check that labels remain attached to inputs and that updates do not silently reset a value. Keep the interface usable at mobile widths and browser zoom. An Artifact is a prototype unless the organization completes its own accessibility and reliability testing; a model-generated component is not a certification.

Inspect the code or reasoning behind warnings

Ask Claude to expose the calculation in a short code snippet or plain-language formula and compare it with your original brief. Look for off-by-one errors, incorrect defaults, rounding of fractional inputs, or a hidden assumption that staff do not count. The app should never increase the supplied maximum when a user enters more attendees. If storage or connected apps are involved, investigate where values persist and whether users can see one another’s information. For a simple planning calculator, local, nonpersonal inputs are preferable. No live occupancy records or participant identities are needed to test the logic.

Decide how and with whom to share

Claude’s help distinguishes private Artifacts from sharing with account holders or organization members, with differing capabilities by plan and artifact type. Check the Share control and organizational policy before distributing a link. Some connected or AI-powered artifacts cannot be shared through the same public-link option, and saved information may be visible to other people with access. Remove event rosters, personal details, and private architectural records from a demo. If the tool is used by volunteers, give them an explicit version, an owner, and a link to the official capacity procedure. Shared links are not a substitute for ongoing maintenance.

Record sign-off and plan for updates

The facilities lead should validate the calculator against its authorized occupancy document and approve the precise formula and warnings. Keep a change log for any revision to limits, room usage, or accessibility notes; retest every boundary example after an edit. A simple artifact can save time, but it cannot inspect a room, interpret a fire code, or establish that chairs actually fit. This scenario is instructional and was not run in a live Claude account. Treat the final tool as a supplementary interface to verified rules, never as a source that creates those rules merely by generating a convincing answer.

How We Research

Our research for this tutorial consisted of reading the official vendor documentation: Claude Help — What are artifacts and how do I use them?; Claude Help — Share artifacts. We identified documented feature names, user controls, access and plan restrictions, and operations requiring independent human approval. We then wrote the illustrative scenario and its review checklist; we did not conduct live product tests, upload any confidential data, measure AI accuracy, or seek real user consent. The artwork is original conceptual instructional design, not product screenshots or measured model output. Documentation and subscription terms can change, so confirm current controls in your own account. Physical safety, published event details, permissions, and factual approval are always the responsibility of the relevant human owner.

Frequently Asked Questions