User guide.
Everything you need to start using gōster — from uploading your first template to mastering voice dictation, custom rules and the clinical dictionary.
Welcome.
gōster is your radiology ghostwriter: dictate your findings naturally and get a professional, structured report written in your style, in seconds.
What gōster does
gōster turns your clinical dictation into a finished report. This isn't classic dictation that transcribes word by word: it interprets your intent, applies your templates, expands your short phrases with the dictionary you taught it, and respects the format you defined.
- Speak naturally. Discuss the case as with a colleague.
- Learns from you. Every correction you save becomes a rule for the future.
- Absolute fidelity. If you didn't dictate it, it doesn't appear. No hallucinations, no fabrications.
- Works out of the box. Even without templates or a dictionary, gōster applies standard radiology knowledge — personalize later.
- Web and mobile. Runs on any modern computer; your phone can be the microphone.
Before you start
This guide assumes you already have an active account. If you don't yet, start your free trial — 15 reports included and up to 3 active templates, no credit card.
Getting started.
Set up gōster once. After that you just dictate. Here are the three steps to your first report — takes less than ten minutes.
Upload your templates
In the sidebar, go to Templates → New. Upload one normal report per study type you read. gōster automatically detects the modality (CT, MRI, US, X-ray…) and the anatomic region.
Build your dictionary
Go to Dictionary → Import and upload previous pathological reports. gōster extracts each finding as a trigger, associates the full description from the body, and anonymizes patient data. Measurements are stripped; the description stays reusable.
Dictate and generate
Open a new study, activate the template by voice ("shoulder MRI") and dictate your findings as if discussing the case with a colleague. Press Generate report and review the output.
You only do it once
The initial setup is the only part that takes time. From there, gōster learns with every report — you won't touch templates or dictionary again unless you want to.
Your workspace.
One screen. Three columns. No hidden menus. This is the only view you need to master.
1 · Library
Your history, templates and dictionary. Filter by modality, region or date. Every item is one click away.
2 · Dictation
The central record button uses your accent color. While recording it pulses to your voice with expanding rings. Switch between keyboard and dictation with the top selector.
3 · Report
Editor with full formatting: bold, lists, tables, alignment and export. Ready to sign and copy.
Top bar
Active template, input mode (keyboard / dictation), profile and settings. Switch template in one click.
Dictation states
- Idle — the button shows the microphone icon in your accent color, resting with a soft breathing pulse.
- Listening — gōster is capturing your voice; the button pulses to your voice with expanding rings in your accent color.
- Paused — recording is temporarily halted with the button dimmer. Press again to resume.
- Processing — gōster interprets and writes; typically 10 to 25 seconds for a standard report.
- Ready — the report appears in the right column; you can keep dictating to add or correct.
Dictating a report.
The secret to gōster is that you don't need to dictate in a structured way. Talk as if you were describing the case to a colleague — gōster arranges the findings inside the template structure.
How to start
- Select or activate a template (by voice or from the library).
- Switch to Dictation mode and press the central microphone button.
- Speak naturally. No need to dictate headings like "findings" or "conclusion" — the template provides them.
- Press Generate report to produce the report.
What to dictate and what not to
- Yes: findings in any order, laterality, levels, measurements, classifications.
- Yes: your short phrases (triggers) that the dictionary expands automatically.
- Not needed: dictating headings like "findings" or "conclusion" — the template provides them.
- If you want explicit punctuation: you can dictate "period" or "comma" to force specific punctuation; gōster respects your intent.
Dictation is continuous
You can pause and resume as many times as you want. Take your time between phrases — gōster waits. Hit the microphone again to pause.
Templates.
A template is your normal report for a study type. gōster uses that structure — headings, sections, default phrases — for everything you dictate within that study.
Creating a template
- Go to
Library → Templates → New. - Upload a normal report in DOCX, DOC or PDF.
- Give it a clear name: for example, "Knee MRI — normal" or "Chest CT — COVID protocol".
- Enable or disable the automatic link to the dictionary.
Activating a template by voice
When you start a new study, you can activate the template just by saying the study name: "shoulder MRI", "unenhanced chest CT", "bilateral mammography". gōster identifies the template and loads the associated dictionary.
What if you don't have a template?
gōster still works. With no template active, it applies its standard radiology knowledge and structures the report with conventional headings (technique, findings, impression). Useful for infrequent studies, urgent cases, or day one before uploading templates.
When to have multiple templates for the same study
- When you read under different protocols (e.g. sports vs. degenerative knee MRI).
- When different institutions require different formats.
- When one variant uses classifications or tables the other doesn't.
Keep templates clean
Upload a normal report — not a pathological one. Pathological findings belong in the dictionary, not in the template.
Clinical dictionary.
The dictionary is where your style lives. Each entry pairs a trigger (what you dictate) with a full clinical description (what gōster writes).
How it works
When you upload past reports, gōster analyzes them like this:
- It identifies each finding in the conclusion and saves it as a trigger.
- It finds the full description of that finding in the body of the report.
- It strips measurements and specifics — the description stays generic.
- It anonymizes names, dates and patient identifiers.
- It merges similar entries without duplicating.
Editing the dictionary
- Go to
Library → Dictionary. Filter by modality or region. - Edit the trigger, the description or both.
- Add variants — multiple descriptions for the same trigger; gōster rotates them to avoid repetitive reports.
- Delete entries you no longer use.
The dictionary grows on its own
Every correction you make on a generated report is evaluated: if gōster detects a pattern, it offers to turn it into a new entry or update an existing one. Your clinical vocabulary expands effortlessly through daily use.
Tables by voice.
If your template includes tables — measurements, angles, side-by-side classifications — you dictate the values and gōster places them in the matching cells. Without touching the keyboard.
How to use voice tables
- Your template must contain the table with clear headers (e.g. Structure · Right · Left).
- Dictate each value with its context: "right Q angle fourteen degrees, left Q angle twelve degrees".
- gōster identifies the row, the column and formats the number with the correct unit.
Insall-Salvati · 1.08 · 1.04
TT-TG · 11.2 mm · 10.8 mm
Voice corrections.
You slip mid-dictation and correct yourself naturally. gōster discards the previous value and keeps your final decision — no keyboard, no going back.
How voice corrections work
gōster interprets corrections phrased naturally — things like "no, it's…", "sorry, I meant…", "drop the part about…", or "change X to Y". The engine combines literal, semantic and phonetic matching against your most recent dictation; you don't need magic phrases, just correct yourself as you would speaking with a colleague.
Instructions & rules.
Define how gōster writes for you. Your style, your conventions, your terminology — always applied, without you having to remember.
Global instructions
They apply to every report. Configure them once in Profile → Global instructions.
- "Avoid the phrases is observed and is identified."
- "All numerical values must be expressed in millimeters."
- "Always use the first-person plural ('we observe', 'we identify')."
Per-template instructions
They apply only inside that template. Configure them per template, under Template → Special instructions.
- "If joint effusion is mentioned, always include the quantification."
- "For this template, express angles with one decimal place."
- "Always report the collateral ligaments even if not mentioned."
Conflicting rules
If a per-template rule contradicts a global one, the per-template rule wins. This is by design: the global is your general style, the specific one is the exception for that study type.
Edit & export.
Once the report is generated, you edit it with keyboard or voice, and send it to whatever system you use to sign it.
Editing the report
- Click anywhere to edit just like in a word processor.
- Use the format bar for bold, italics, lists and alignment.
- If you dictate another sentence, gōster inserts it where your cursor is.
- Save changes so they become future rules.
Export formats
- Copy with formatting — to clipboard with bold, lists and tables, ready to paste into Word or your RIS.
- Copy as plain text — unformatted, for RIS / PACS systems that accept only plain text.
Universal paste
The Copy with formatting option preserves bold, lists and tables in most modern systems. If your RIS accepts only plain text, use the dedicated option — gōster applies the correct line breaks.
Phone as microphone.
Better audio quality? Wireless mobility? Pair your phone and use it as a wireless microphone. No app, no installation.
Pairing your phone
- In the dictation bar, tap the QR icon.
- Scan the code with your phone's camera.
- Allow microphone access in the phone's browser.
- Done — the indicator in the web app changes to "phone connected".
Tips
- Plug headphones with a mic into the phone for better quality and less ambient noise.
- When you're done, close the tab on the phone or disconnect from the web app.
- Phone and computer can be on different networks — the connection is web-based.
Appearance.
Your workspace, your way. Change it whenever from the settings panel in the sidebar.
Light / dark theme
Switch with one click. The preference is saved per device.
Tone control
Use the brightness slider to fine-tune the background warmth. Useful for reducing visual fatigue on long shifts.
24 accent colors
Pick the color for cursors, buttons, dictation waves and highlights. Four rows: signature, warm neutrals, vivid warm and vivid cool. Only chroma changes — contrast stays readable.
Language
Switch between Spanish and English from the language menu. The preference is saved in your account and applies to the UI, emails and prompts.
Keyboard shortcuts.
Shortcuts available today in gōster.
| Shortcut | Action |
|---|---|
| ⌘ + K | Quick search (templates, dictionary, history) |
On Windows / Linux, replace ⌘ with Ctrl.
Privacy & data.
Your data is yours. This section explains how we handle it.
Transport and storage
- All data travels encrypted with TLS.
- Servers run on certified cloud infrastructure.
- Dictation audio is not stored: it exists only as a temporary file during the processing of the active study. It never lands in history, permanent database, logs or any associated files.
Automatic anonymization
When you upload reports to build the dictionary, gōster automatically anonymizes:
- Full patient name and family members.
- Identifiable dates.
- ID numbers, medical records, phone and address.
- Referring physician and institutional facility names.
Use for training
We do not use your information to train public models. The models that learn from you apply only to your account. If you need institution-level isolation, contact us about the Enterprise plan.
Delete your account
If you want to stop using gōster, email us at hello@goster.ai and we'll permanently delete your account and all associated data.
Troubleshooting.
The microphone won't activate
- Check that the browser has microphone permission (padlock icon in the URL bar).
- Make sure no other tab or app is using the microphone.
- On macOS, go to Settings → Privacy → Microphone and authorize your browser.
- Restart the browser if the indicator stays gray.
Generation takes longer than usual
- Check your internet connection (≥ 5 Mbps recommended).
- Very long dictations (> 5 min) take proportionally longer.
- If the problem persists, email us at hello@goster.ai.
The dictionary expands something I don't want
- Open the entry in
Dictionaryand edit the description, or add an alternate variant. - If two triggers overlap, rename them so they're more specific.
- Dictate a correction such as "drop the part about X" on the generated report and save changes — gōster learns.
The phone QR won't connect
- Check that the phone has internet access.
- If the code expired, close it and reopen the QR from the dictation bar.
- If you're on a corporate network, ask IT to allow outbound WebSocket connections.
Frequently asked questions.
Contact.
We're here to help. Email the team directly or visit the site for more info.
Support & sales
hello@goster.ai — single team inbox; we reply as soon as possible.
Website
goster.ai — landing, plans and announcements.
Ready for your first report?
Open gōster and dictate. If you get stuck, come back to this guide.
© 2026 gōster · Made with care by radiologists, for radiologists.