tell me about your capabilities, informations and accuracy, soursces

Orris - Capabilities, Information & Accuracy

Here's a straightforward overview of what I am, what I can do, and where my limits are.

What I Am

I'm Orris, an AI assistant built to help across a wide range of domains - from general knowledge and research to specialized medical, scientific, and technical questions. I run on a large language model and have access to several tools that extend my capabilities well beyond a typical chatbot.

Capabilities

Knowledge & Research

General knowledge across science, history, math, technology, law, finance, culture, and more
Medical & clinical questions - I have access to a curated library of medical textbooks (pharmacology, anatomy, pathology, physiology, nursing, and more) that I consult before answering health-related questions
PubMed literature search - I can search peer-reviewed medical research, retrieve abstracts, find systematic reviews, meta-analyses, and RCTs, and even download full-text PDFs of papers
Live web search - I can search the web in real time for current events, recent news, up-to-date guidelines, and anything my training data may not cover
Web extraction - I can read full articles, documentation, and web pages from specific URLs

Document & File Generation

PowerPoint (.pptx) - Create full slide decks and presentations
Word (.docx) - Generate formatted Word documents, reports, memos, letters
Excel (.xlsx) - Build spreadsheets, analyze tabular data, clean datasets
PDF - Read, extract text, merge, split, rotate, watermark, and OCR PDFs
Images - Generate original images, diagrams, posters, flashcards, and visual content

Code & Computation

Code execution in a persistent Linux sandbox (Python, Node.js, shell scripts, and more)
Data analysis, file processing, automation scripts
Running builds, tests, and multi-step computational tasks

Specialized Tools

Medical library - Authoritative textbook content consulted for every clinical question
Image fetching - Retrieve and display images from URLs, including medical diagrams and clinical photos
Subagent delegation - For large, complex tasks I can spin up isolated sub-agents to work in parallel, then synthesize results

Information Sources

Source	What it covers
Training data	General knowledge up to a training cutoff (may be outdated for recent events)
Medical textbook library	Curated, peer-reviewed medical textbooks across specialties
PubMed	Peer-reviewed journal articles, clinical trials, systematic reviews, meta-analyses
Live web search	Real-time web results - news, guidelines, recent publications, general facts
Web extraction	Full content from specific URLs for deep reading
Uploaded files	PDFs, spreadsheets, documents you provide directly

Accuracy & Limitations

What I do to maximize accuracy

For medical questions, I always consult the textbook library before answering - I don't rely on training memory alone
I run web searches on nearly every turn to catch anything my training data may have missed
I cite sources inline so you can verify claims
I flag uncertainty explicitly - if I don't know something or my sources are thin, I say so

Known limitations

Training cutoff: My base knowledge has a cutoff date. For anything time-sensitive (drug approvals, recent guidelines, breaking news), I supplement with web search and PubMed
I can hallucinate: Like all language models, I can generate plausible-sounding but incorrect information, especially for very specific facts, statistics, dosages, or obscure topics. Always verify critical information with primary sources
Not a substitute for professionals: For medical, legal, or financial decisions, my answers are informational only - consult a licensed professional
Web search has a budget: I have a limited number of tool calls per session, so I prioritize the most relevant searches
File generation is sandbox-dependent: Generated files (PPTX, DOCX, etc.) are created in a sandboxed environment and linked for download

On medical accuracy specifically

I pull answers from actual textbook content, not just training memory
I cross-reference with recent PubMed evidence when relevant
I warn you if recent research may contradict older textbook guidance
I will never fabricate drug dosages, clinical guidelines, or study results

Current Date

Today is June 09, 2026. My live web search capability means I can access information up to today for topics covered by public web sources.

If you want to test any specific capability, just ask. I'll show you what I can do.

Can you create a presentation or document for me?

Asking for Preferences

Reading File

Finding Sources

Running Command

Finding Sources

Reading File

Finding Sources

Reading File

Running Command

Writing File

Running Command

Your credits are exhausted. Your daily credits reset tomorrow, or you can purchase a top-up for more.

word document

Your credits are exhausted. Your daily credits reset tomorrow, or you can purchase a top-up for more.

This is a shared conversation. Sign in to Orris to start your own chat.