Blue J

What it is

Blue J is a generative AI research tool built only for tax. You ask a tax question in plain language and get a written answer with inline citations to the Internal Revenue Code, Treasury regulations, IRS guidance, court decisions, and state and local authority. You can ask follow-ups, upload client documents for analysis, and export the result as a memo in PDF or DOCX. The US product covers federal and state and local tax (SALT). Separate regional editions cover Canada and the UK.

The content library is the differentiator. Blue J holds an exclusive US license to Tax Notes news and analysis from Tax Analysts. Since a September 2025 partnership, it also embeds IBFD content covering 220+ jurisdictions for cross-border questions such as permanent establishment risk, global minimum tax, and transfer pricing. General legal research assistants do not carry either source.

Benjamin Alarie, a tax law professor at the University of Toronto, founded Blue J in Toronto. Its first product, launched in 2015, predicted court outcomes with supervised machine learning. In August 2023 the company replaced it with the current LLM-based research product. On August 4, 2025, Blue J announced a $122M Series D led by Oak HC/FT and Sapphire Ventures at a valuation above $300M, after more than doubling revenue and customers in the first half of 2025. The vendor now says 6,000+ firms use it, including RSM, Crowe, and KPMG UK. CPA.com, the AICPA’s technology subsidiary, names Blue J its preferred tax research partner.

Why legal ops teams look at it

The question it answers is “what do we give the tax practice when our general legal AI does not cover the Code?” Tools like CoCounsel and Harvey are licensed to litigators and transactional lawyers. The tax group then either keeps a separate research library or pastes Code sections into a general chatbot with no citations to check. Blue J is a tax-only seat that sits beside the legal stack and is priced low enough to give to every preparer and associate.

Specific use case: a 40-person accounting firm’s tax department gets about 30 client questions a week, such as whether a client’s remote employees create nexus in a new state, or how a Section 1031 exchange interacts with a related-party sale. Associates ask Blue J first, check the cited authorities, and export a draft memo for a manager to review. Managers review the authority, not a blank page. The vendor reports 3 hours saved per user per week. Treat that as a vendor claim, not a benchmark.

Pricing reality

Blue J publishes its price, which is rare in this category. The Individual plan is $1,498 per user per year (about $125 a month), billed annually. It includes unlimited questions, drafting, and file uploads, full US federal and SALT coverage, and a 7-day free trial with no credit card. The Team plan adds SSO, access controls, firm-wide usage insights, and a dedicated customer success manager, at a custom price. Canadian and UK editions are priced on their own regional pages.

Discounts are common. The NJCPA lists 20% off for members, which brings a seat to $1,198. The Ohio Society of CPAs added Blue J as a member benefit in February 2026. For comparison, Thomson Reuters lists CoCounsel Tax Essentials at $1,300 per user per year for firms of up to 10 users, capped at 20 citation-backed responses a week, and without access to the Checkpoint Edge research library. Checkpoint, CCH AnswerConnect, and Bloomberg Tax are quote-based and usually cost more per seat for multi-user licenses.

Best for

Tax partners, tax managers, and legal ops leaders at accounting firms, law firm tax groups, and corporate tax departments that need citation-backed answers on US federal, state, Canadian, UK, or cross-border tax questions, and want a per-seat price they can roll out to the whole practice without a procurement cycle.

Alternatives and when to pick them

• Thomson Reuters Checkpoint with CoCounsel Tax: one of the two incumbent tax research libraries most US firms already license. Pick it when the firm already pays for Checkpoint and runs Thomson Reuters tax software, so the AI answers come from the editorial content your staff already trusts.
• Wolters Kluwer CCH AnswerConnect: the other incumbent, now with Expert AI on top of human-reviewed CCH analysis. Pick it when the firm runs CCH Axcess or ProSystem fx and wants research inside the same vendor relationship.
• Bloomberg Tax: Bloomberg Tax Answers and AI Assistant are included with a Bloomberg Tax subscription at no extra charge. Pick it when the corporate tax team already relies on Bloomberg Tax Portfolios.
• Harvey: the fastest-growing legal AI platform, which Big Four firms including PwC use for tax work. Pick it when you want one AI workspace across legal and tax and can pay enterprise pricing for it.
• CoCounsel (legal edition): pick it when the need is litigation and transactional research and tax is an occasional side question.

Watch-outs

• Independent reviews describe the old product. Blue J’s Capterra and G2 pages show fewer than 10 reviews, and most of them describe the pre-2023 outcome-prediction tool, not today’s research assistant. Guard: run the 7-day trial on 20 questions your seniors already answered, and score every citation for accuracy and relevance before you roll it out.
• AI answers are a starting point, not a position. A cited answer can still misread an authority or miss a recent change. Guard: make citation checking part of the review step, and require a manager to open at least the primary authorities before any memo goes to a client.
• US-only data residency. Blue J stores and processes all data in AWS us-east-1 and offers no other region or on-premise option. Guard: for Canadian, UK, or EU client files, confirm with your privacy counsel that US processing is acceptable, or research on anonymized facts. Uploaded files are deleted after 24 hours, and customer data is not used to train models.
• No API or integrations. Blue J is a standalone web app with no public API and no MCP server. Guard: if you need answers inside Word, a document management system, or your tax software, budget for copy-and-paste in the pilot or choose a vendor that integrates with your stack.

Background: Legal AI grounding vs hallucination.