Most students pick up research tools the way they pick up citation styles: from whoever sat next to them in the first year. That works until a thesis gets big. A better way to choose is to look at the stages of the work and ask which stage is slow for you right now. Research breaks into finding papers, reading them, keeping track of sources, analyzing data, writing, and, for some fields, running a formal review.
Start with search. Google Scholar and Semantic Scholar cover a lot of ground for free, and subject indexes such as PubMed, Scopus or Web of Science matter when your field expects a documented, repeatable search. Question-led engines like Consensus and Elicit are useful for a quick first look at a topic, especially outside your own area. Citation maps such as Connected Papers and Litmaps help once you have one strong paper and want to find what sits around it.
Next, pick one reference manager and stay with it. Zotero is free and open source, Mendeley and EndNote are common through university licenses, and Paperpile suits people who write in Google Docs. The choice matters less than using it from the first paper you save. Moving a few hundred references with messy metadata a month before submission costs far more time than any feature saves.
For reading, the question is where your highlights and notes should live. Many reference managers now include a PDF reader, so a separate app only earns its place if it does something yours does not, such as linking excerpts across papers or answering questions about a long document. Treat AI answers about a paper as a guide to where to look, and check the page they point to.
Writing tools need the most care. Grammar and language editors such as Writefull, Paperpal and Trinka are generally accepted for polishing your own text, but rules on AI-generated or AI-paraphrased text differ by university, department and journal. Read your institution's policy and the author guidelines of the journal you are aiming for, and disclose AI use where they ask you to.
For data analysis, AI assistants can write and explain code in Python or R, which saves time on routine cleaning and plotting. They also make mistakes that look correct. Run the code yourself, read it line by line, and keep the scripts with your data so the analysis can be reproduced. For qualitative work, AI coding suggestions in MAXQDA, ATLAS.ti or NVivo are a starting point, and your own coding decisions are what the method rests on.
Whatever you use, check every citation before it goes into a draft. AI tools can produce references that look real and do not exist, or attach a real paper to a claim it never made. Open the source, confirm it says what you are citing it for, and keep your institution's AI policy at hand.