Tax research is a retrieval-and-citation problem, which is exactly what large language models grounded in a curated corpus do well. The established platforms have all shipped it — Thomson Reuters' Checkpoint Edge with CoCounsel, Wolters Kluwer's Expert AI over CCH AnswerConnect, Bloomberg Tax Answers — each returning short answers backed by links to primary authority.
Blue J took a different route, starting from machine-learning prediction of case outcomes at the University of Toronto before adding generative research across US, Canadian, and UK law. It raised $122M in 2025 and grew from about 200 client organisations in 2021 to over 2,500. The discipline that separates these from a general chatbot is citation: an answer a practitioner cannot verify is worthless.
Frequently asked
- Can AI replace a tax research subscription?
- Not for professional work. The value of these platforms is the licensed, maintained corpus the AI reads from — primary law, regulations, and editorial analysis. A general chatbot without that grounding cannot cite authority reliably enough to rely on.
- Do AI tax research tools cite their sources?
- The credible ones do, and this is the main quality test. Checkpoint Edge, CCH AnswerConnect, Bloomberg Tax Answers, and Blue J all return citations to primary authority alongside the answer.
- What is AI tax planning software?
- It analyses a completed return or client situation against a library of strategies, estimates the savings from each, and generates a client-ready plan. TaxPlanIQ and Instead both work this way, aimed at firms selling advisory work rather than at taxpayers.