Paste your scribbled notes.
Whatever you'd write on the back of a drug chart. Acronyms, units, fragments — the prompts are built to handle clinic shorthand.
HR 78 reg (was 112)
echo LVEF 52% MR mild
plan: continue, f/u 6/12
No new software. No new login. ModusLab is a library of expertly-written instructions for an AI you already use.
Whatever you'd write on the back of a drug chart. Acronyms, units, fragments — the prompts are built to handle clinic shorthand.
Clinic letter, discharge, patient explainer, MDT summary, SBAR — 33 prompts across the three toolkits, each one engineered and tested.
You get back a draft built only from what you typed in. Skim it, fix the bits that matter, sign. The thinking stays with you.
Every toolkit is engineered around a single instruction the model must obey: only restate, structure, or paraphrase what you typed in. No retrieval. No filling in plausible detail. No inventing what the patient didn't say.
The model can only use what you give it. Empty fields stay empty — flagged for you to complete by hand.
Every prompt ships with an explicit list of what the model must decline — diagnoses, doses, anything outside your notes.
Each prompt opens with a reminder of the information-governance posture you should already be holding to. PID stays inside your sanctioned deployment.
Each toolkit is a curated set of prompts with worked examples, source-locked scaffolding, and notes on when not to use them.
Sixteen field-tested prompts for the documentation that fills your day.
Nine prompts for the paper analysis, synthesis, and academic writing that takes your evenings.
Eight prompts for the career work that decides the next five years.
All thirty-three prompts across clinical, academic, and professional work. The whole library, at the steepest discount of the launch.
Real-shape input. The output uses nothing that isn't in the notes above. Every clinical fact is traceable.
Admission: 14 March — chest pain. Diagnosed NSTEMI on the basis of a peak troponin I of 412.
Procedure: Coronary angiography on 16 March showed a culprit lesion in the left anterior descending artery, treated with one drug-eluting stent. No procedural complications. Subsequent transthoracic echocardiogram (17 March) demonstrated preserved left ventricular function (LVEF 55%) without regional wall motion abnormality.
Discharge medications: aspirin 75 mg lifelong; ticagrelor 90 mg twice daily for 12 months; atorvastatin 80 mg nocte; bisoprolol 2.5 mg once daily.
Follow-up: cardiology clinic in 6 weeks; GP review in 1 week. Safety-netting: any recurrent chest pain warrants emergency assessment via 999.
The one most doctors copy first. Paste your dictation, get a structured GP-ready letter in seconds. Read it now, no sign-up — or get our most-used prompts as a keepable PDF.
Email capture activates at launch. (Set the Buttondown username in src/lib/newsletter.ts.)
You already pay for an AI subscription — ChatGPT, Claude, or Gemini. The model itself is excellent. What it lacks for clinical work is structure and safety constraints. ModusLab gives you those: a tested set of instructions you paste into the AI you already use. No new tool to install, no second subscription, no new login. The product is the prompts.
No. ModusLab sells written instructions — text — that you paste into a general-purpose language model you've already licensed. It's not a clinical decision-support tool, not a SaMD, and not intended to influence diagnosis or treatment. The output is a drafting aid you read, edit, and sign.
The prompts are model-agnostic and designed against the current generation of frontier assistants (ChatGPT, Claude, Gemini). They work in any chat interface that accepts long instructions. If your trust has a sanctioned enterprise deployment of one of these models, the prompts paste in unchanged.
Each prompt opens with a reminder of the information-governance posture you should already be holding to. The toolkits assume you are using a sanctioned deployment for any input containing PID — and the worked examples use anonymised composites throughout.
After you pay through Stripe, you get an email with a download link to the PDF (or zip, for the bundle). It's yours immediately — open it, copy a prompt, paste it into the AI you already use. No account to set up, no app to install.
Because the product is a digital download — once the file is in your inbox, it can't be returned — purchases are non-refundable.
Open the AI you already pay for. Paste a prompt. Sign the draft. The full library by tonight.