Every claim, traced to its source.
Claims are checked against their sources, and when one cannot be confirmed, ktAi says so.
An AI platform for cross-border tax practice — opinion-grade, with every citation checked against authority.
General-purpose AI does not meet the standard of tax practice.
Generalist models fabricate citations. Legal search platforms surface documents but leave interpretation to the reader. Tax practice has no room for either failure. A stale rate or a missed regulation creates real exposure for the client and the firm. It needs a tool that meets the standard of the work.
ktAi is built that way. Citations that can't be verified are marked unverified. Positions the authority can't support are flagged.
What ktAi does.
It does four things.
-
Tax opinions and memoranda
It drafts opinion-grade reports with the structure a tax attorney already uses: statement of facts, legal question, framework, reasoning, conclusion. Each section is built from real source authority, and every citation is checked against the authority it cites. Any that cannot be confirmed is marked before the draft reaches you.
-
Cross-border analysis
The first corridor is US-Israel: both domestic regimes, the bilateral treaty, and the OECD frameworks around them. Most tools lose accuracy at the borders. ktAi was made for that.
-
Firm workflow
It fits the way firms already work: client portfolios with role-based access, senior review of junior drafts section by section, conflict-of-interest controls, and a full audit trail. The work product belongs to the firm. Every step in producing it is recorded.
-
Institutional memory
The firm's work compounds. Every opinion is indexed in the firm's private knowledge base, so past analysis can be found and built on. The next matter begins from what the firm already knows, not a blank page. Access stays within the firm.
Structured to work like a firm.
Tax practice is collaborative work. ktAi is structured the same way, inside the system and across the firm.
Inside the system
Analysis runs across specialized capabilities, each tuned to a domain or a jurisdiction. A synthesis step pulls the findings into a single position. Counter-arguments are set out alongside the position. Verification steps the model can't skip then check the result, and every citation is checked against its source before the work goes out, and any that can't be resolved is marked.
Across the firm
The platform mirrors the review hierarchy a firm already uses. A junior pulls a first draft. A senior reviewer marks it up section by section, and the author revises. A partner signs off. ktAi sits inside that workflow, so the work product carries attribution at every step.
Roy Kariv.
Roy Kariv has practiced cross-border tax for over a decade. He started as a prosecutor at the Israeli Tax Authority, attacking positions firms had taken. He left to start KLF Law Office in Tel Aviv, where he now builds and defends them. He built ktAi for himself first. He used it on live US-Israel matters at KLF, reviewing every output personally. It cut meaningful time off his opinion drafting. Then he opened it up to other firms.
The product carries that background. It frames questions and weighs authority the way Roy does. When it isn't sure, it says so. ktAi serves attorneys, advisors, accountants, tax directors, and in-house teams.
Recent essays.
Minimal Human Involvement
The UN's draft protocol on cross-border services decides where a service's income arises partly by asking how much human involvement it takes to deliver. The reasoning that has kept professional advice out of the digital category assumes advice cannot be tailored to a new client without more people, and that is the assumption AI is removing.
Work with us.
ktAi is open to inquiries from tax attorneys, advisors, accountants, tax directors, in-house teams, and firms thinking about AI for their practice.