Hottest NEW AI tools for Research: Must-Watch AI Apps

Academic research is getting reshaped by an AI “tool arms race,” with new systems moving beyond chat into drafting, citation-building, and even automated multi-step workflows. The standout theme is speed: researchers can now generate grant outlines, literature-review starting points, and structured academic text by pulling in references and synthesizing across multiple documents—often with only a short prompt.

Sight (spelled “sight” in the transcript) is presented as a one-prompt research assistant that can draft sections of essays and grant proposals while drawing on research articles. A live example shows a grant-writing prompt (“explore how … affects chromosome …”) producing a structured set of “specific aims” and then returning real references. The assistant appears to search and analyze documents in the background, assembling a grant proposal draft that includes cited sources. The presenter also notes that the generated literature reviews tend to be shorter than traditional ones, but they still reduce the “grunt work” of structuring and getting a first pass of relevant literature. The practical takeaway: better results come from being more specific about what the researcher wants—such as targeting a word count, subfield, and time window.

Jenny AI is positioned as a more hands-on writing partner. After logging in, it keeps generating text continuously, letting users accept or request the next chunk without restarting from scratch. It also supports adding citations directly into the draft, and it offers features for creating section and subheadings—turning a rough topic request into an organized academic structure. A key promise is expanding compatibility with reference managers, with the CEO mentioning upcoming support for Mendeley and Zotero, which would make it easier to move from AI drafting to standard academic workflows.

Hey GPT is introduced as a paid tool that connects chat-based AI to the internet and to uploaded files. The transcript emphasizes “chat with PDFs” as a major research advantage: users can upload papers, then ask for outlines and “important findings” that are pulled from the documents. It also supports chatting with websites, though the process can be a bit clunky and may cost tokens depending on the amount of web content crawled. The overall pitch is consolidation—one place to query multiple sources (PDFs, HTML pages, and Google-linked information) rather than switching tools.

Finally, the transcript shifts from single-assistant tools to agentic automation. Agents are described as AI systems that spawn other tasks or other agents to pursue a goal. Auto GPT is shown as an example that can run a “science agent” to collect information and write it into a text document, but it currently gets stuck in loops and may require an API (and therefore can cost money) if it continuously searches the web. Agent GPT is also mentioned as a browser-based beta alternative. The message is cautious optimism: agent workflows look promising for academia, but reliability and cost control remain early-stage issues.