A tested library of AI prompts for regulated work.
Loddle is a reference of profession-specific prompts for lawyers, architects, healthcare administrators, and contractors. Every prompt is tested on real models and structured to a published schema — so you can read the methodology before you trust it with client work.
Four professions. Prompts written like a colleague would write them.
Each library is built around profession-specific constraints — prompts that know your vocabulary and the things you cannot say.
Lawyers
Billing narratives, client communications, discovery review, deposition prep — AI assistance for the unbillable side of practice.
Architects
Construction documents, owner correspondence, code research, narrative specifications — built with licensed-practice constraints baked in.
Healthcare Admin
Patient communication, regulatory documentation, claims appeals, HIPAA-aware operations — built with compliance-aware constraints baked in.
Contractors
Change orders, subcontractor RFIs, owner status reports, scope clarifications — written in the language of construction.
Four quality gates before a prompt ships.
Generic prompt sites publish whatever sounds plausible. Loddle ships nothing until it has cleared every gate below — and we publish the audit trail.
Tested on real models
Every prompt has a sample output generated by Claude working through a realistic, synthetic professional scenario — not real client data — using a dedicated generation script. The full output is published on the prompt page before you copy anything.
Structured to a published schema
Each prompt declares its role, task, structure, and constraints in the same schema. You can read the contract before you trust it with client work.
Profession-scoped
Constraints written for a specific profession. The lawyer's billing prompt knows what "contingency" means. The contractor's RFI prompt knows what a CCD is. Every prompt declares its role, constraints, an uncertainty instruction, and a professional disclaimer as data — not bolted on after the fact.
Confidence tiers and honest disclaimers
Confidence tiers — Basic, Professional, Advanced — reflect the stakes of the task and are displayed on every prompt. Every prompt carries a profession-appropriate disclaimer; the language is consistent across the library to avoid drift.
Start here — selected prompts.
FEATURED · 10 SELECTED ACROSS 4 PROFESSIONS
Client Matter Status Update Drafter
Draft professional client status update letters and emails that clearly communicate case developments, next steps, and what the client needs to do — without legal jargon that confuses or alarm that panics. This prompt produces clear, organized updates that reinforce your responsiveness and manage client expectations.
Legal Motion Draft Assistant
Generate a structured first draft of any legal motion — from a motion to dismiss to a motion for summary judgment. This prompt produces a complete motion framework with proper caption, statement of facts, legal standard, argument sections, and conclusion, ready for attorney review and citation population.
Case Law Research Brief Generator
Transform a legal question into a structured research brief that surfaces relevant case law, identifies controlling precedents, and outlines the state of the law in your jurisdiction. Use this prompt to accelerate legal research and produce a first-draft brief that organizes arguments around the strongest available authorities.
Architectural Project Brief Generator
Generate a comprehensive architectural project brief from client inputs and initial consultation notes. This prompt guides an AI to structure a professional project brief covering program requirements, design goals, site constraints, budget parameters, and schedule expectations — establishing a clear foundation for the design process.
Design Concept Presentation Builder
Structure a compelling design concept presentation for a client meeting, design review board, or public presentation. This prompt helps architects organize their design narrative, key visuals, and talking points into a presentation structure that communicates the design intent clearly and builds client confidence in the direction.
RFI Response Drafter for Architects
Draft a professional, clear, and contractually appropriate response to a Request for Information (RFI) from a contractor. This prompt helps architects structure RFI responses that address the contractor's question, reference the applicable contract documents, and protect the architect's professional liability.
HIPAA Policy Document Drafter
Draft a HIPAA-compliant policy document for a specific privacy or security requirement. This prompt helps healthcare administrators write comprehensive, plain-language HIPAA policies that address regulatory requirements, establish clear staff expectations, and demonstrate good-faith compliance efforts during an audit or breach investigation.
Prior Authorization Letter Drafter
Draft a comprehensive prior authorization request letter for medical procedures, medications, or services. This prompt helps healthcare administrators write clear, medically justified authorization requests that address common payer denial criteria and increase the likelihood of first-pass approval.
Construction Change Order Drafter
Draft a professional, well-documented change order request for additional work or cost due to a contract change. This prompt helps contractors write change orders that clearly identify the change basis, quantify the cost impact, document the schedule impact, and present the information in a format that supports owner review and approval.
Construction RFI Drafter
Draft a clear, professional Request for Information (RFI) to the architect or engineer of record. This prompt helps contractors write RFIs that precisely describe the field condition, identify the specific contract document ambiguity, and request a timely response — protecting the contractor's schedule and avoiding unauthorized field decisions.
An honest breakdown of where each model helps — and where both fall short.
For each niche we break down how Claude and ChatGPT differ on professional work, so you can choose deliberately. We publish verified results only — never benchmarks we have not run.
Neither Claude nor ChatGPT can verify citations against live legal databases — both can produce plausible-sounding but incorrect references. Every Loddle prompt embeds an uncertainty instruction and a professional disclaimer for exactly this reason.
We ran a controlled, blind-judged head-to-head of Claude and ChatGPT across representative prompts in every niche — the verified, dimension-by-dimension results are on each per-niche comparison page.
Stop guessing whether the prompt is safe to use.
Read the methodology. Open any prompt. Trust the work — or don't. Start free today.
- Every prompt validated automatically on every build, against a published schema.
- Schema-defined output you can actually inspect before running.
- Profession-appropriate disclaimers and confidence tiers on every prompt.
- Constraints built for the profession — not generic prompt templates.