7 Use Cases for Charlie

Charlie is your AI scientific research agent. It can help you with your entire scientific workflow through 7 main use cases.

Charlie can provide detailed answers to your scientific questions by accessing and synthesizing information from peer-reviewed journal articles while citing its sources. These sources can be viewed in the right panel of the interface.

Example questions:

• What are the latest advances in myocardial infarction?
• What is the role of autophagy in the resistance of tumor cells to chemotherapy treatments?
• What are the known mechanisms of protein X in cell signaling?

If you have specific research articles or documents, Charlie can summarize the key results, methodologies and conclusions for you.

Upload your PDFs to the Library and ask questions directly about the content of your documents.

Example questions:

• What are the conclusions of this article?
• Can you summarize the experimental method?
• What are the main results reported?
• Can you explain the "Discussion" section in simple terms?

Charlie can help you gather and review the literature on a specific topic, highlighting the most relevant studies and their contributions to the field.

Use the year filters to narrow your search and explore the literature progressively by refining your questions as you go.

Example questions:

• What are the latest advances regarding protein kinase B (AKT)?
• What strategies are used to deliver siRNA into tumor cells?
• Are there any approaches using nanoparticles?

Charlie can help you build your scientific hypotheses by testing them against the existing literature. It's a true thinking partner for testing and refining your ideas.

Example dialogue:

Drawing on the literature, Charlie can help you design experimental protocols to test your research hypothesis.

Example request:

If you have data from experiments or studies, Charlie can help you interpret the results and provide insights based on existing scientific knowledge.

Example scenario:

Charlie can help you prepare manuscripts for publication, responding in the format you need. Whether it's structuring an introduction, writing an abstract or rephrasing a section, Charlie adapts to your needs.

Example requests:

• Can you summarize this in a 200-word paragraph?
• Help me structure my introduction on AKT inhibitors
• Rephrase this section in more accessible terms
• Suggest an organization for my "Discussion" section

Enable the "Reflection" toggle in the input bar for questions that require a thorough analysis. Charlie triggers an orchestrator that runs a series of searches (scientific databases + the Internet) and drafts a well-reasoned response. The reasoning is displayed in real time.

Ideal for: cross-referencing publications from PubMed with the latest news on clinical trials online, comparing the current state of academic research with recent announcements from pharmaceutical companies, or producing a cross-source summary.

Ask Charlie, “Draw me a diagram of signaling path X” or “Create a timeline of posts in chronological order.” It generates a Mermaid diagram displayed directly in the conversation, along with citations to the sources that support it.

The diagram opens in an interactive viewer (zoom, export). You can also edit it to adjust the labels or layout.

Every article cited by Charlie can be shared via a permanent URL in the following format /article/es_xxxxxxxx. Send the link to a colleague: they will be taken directly to the article’s details page (title, abstract, external link, metadata).

You can also share an entire conversation thread with a colleague so they can pick up where you left off without having to explain the context.

The filter panel allows you to narrow your search to a subset of indexes: bioRxiv preprints only to scan for unreviewed content, HAL only for French-language research, or a custom combination. Citations indicate the original source; links (PubMed, DOI, HAL, bioRxiv) are clickable.

Automatic deduplication: An item that appears in multiple indexes (for example, PubMed and HAL) is displayed only once thanks to the canonical identifier es_id.