Where does AI create value, and where does the risk land? A framework for deciding how to lead each AI initiative, in four moves: Amplify, Automate, Assure, Authorize.
Left to right: who carries the decision. On the left, an expert’s judgment determines the outcome and a person stands behind each output. On the right, the system produces outputs in volume, faster than people can review them one by one.
Top to bottom: what is at stake. At the top, the main question is how much an initiative gains. At the bottom, it is what happens when it is wrong, and who pays.
The judgment line. Moving work from left to right, from human judgment to machine scale, is the central AI decision. Crossing the line changes what you must measure and who must own it, so make that move deliberately rather than by drift.
The Matrix Vector
Our own measure of where AI pressure is moving. Every development in the living index is scored for impact, and together they add up to one arrow on the matrix: its direction shows where the pressure points, its length how concentrated it is.
This periodPrevious periodEach development, sized by impact
The Matrix Vector over time
The trend appears once six monthly editions are published. The first monthly edition covers November 2026, so the six-month view arrives in May 2027.
The compass trail: where the vector pointed each month, oldest to newest.The two components by month: human judgment to machine scale and risk carried to value created. Each is scaled from −100% to +100% of that month’s total impact.
How the Matrix Vector is calculated
Each development in the index is placed in one zone and given an impact score:
1, notable: changes decisions for some organizations.
2, significant: changes decisions across a sector or region.
3, major: changes decisions across sectors, or sets a landmark.
Each zone points toward its corner of the matrix. A development pulls the vector toward its zone by its impact score, and the pulls are added together. The direction of the result shows where pressure is heading. The concentration is the length of the result divided by the total impact: 100% means every development pulled the same way, and a low figure means pressure was spread across zones and largely cancelled out.
The vector is published monthly, covering the developments added that month. After six monthly editions, the page also shows the vector over time: the path its tip has taken across the matrix, and its two components month by month. Edition 1.0 opens with a baseline calculated from the developments that shaped the framework. Scores reflect editorial judgment, and every score is listed with its entry in the index and the CSV download.
Place your AI initiative
Answer six questions about one specific initiative to see where it sits on the matrix and what that means. It takes about a minute, and nothing you choose leaves your browser.
The placement is a starting point for discussion, not a scored assessment. Two people answering for the same initiative may disagree, and that disagreement is usually worth having.
Use the matrix on your own AI portfolio.
For boards, CEO offices and department heads: place every AI initiative, see where the portfolio leans against external pressure, and print a board pack.
Developments worldwide that move the matrix: new rules, rulings, prices, and evidence, each placed in a zone with a note on why it matters to leaders. Every entry records its jurisdiction, and coverage by region is reported each year. Edition 1.0 starts with the developments that shaped the framework; new entries are added monthly.
The Vector & Value Matrix is developed by Santosh Kumar at Vector & Value AI and reviewed every six months. If a zone or a question changes, the change is listed below rather than made quietly.
Limitations
The matrix is a framework for structured judgment, drawn from our consulting, product, and research work. It is not an experimentally validated model, and the placement questions are a guide rather than a measurement. Index entries reflect editorial judgment about which developments matter and where they sit. Monitoring covers every major region, but the Edition 1.0 baseline is concentrated in developments from the United States and India, together with international research.
How to cite
Kumar, S. (2026). The Vector & Value Matrix (Edition 1.0). Vector & Value AI. https://www.vectorvalueai.com/matrix
You may read and use them for your own reference, cite them, and quote short extracts with attribution to Santosh Kumar, Vector & Value AI, and a link to this page. Academic researchers may use the index for non-commercial research with the same attribution.
You need our written permission to use the framework, the index, or their graphics commercially, including in consulting, training, workshops, software, or reports provided to clients or sold; to reproduce them in substantial part; or to adapt them into a derivative framework or product. To request permission or a licence, contact us.
Trademarks. Vector & Value AI™, The Vector & Value Matrix™, and the Vector & Value logo are trademarks of Vector & Value AI and may not be used without permission.
Related resources
The MIT AI Governance Map tracks how AI is governed by laws and standards, and the Stanford AI Index tracks AI’s progress. The matrix asks a different question: what AI means for value, judgment, and responsibility inside an organization.
Feedback and corrections
Spotted an error, or have a development the index should include? Tell us. Corrections are noted in the revision history.
Revision history
1.0, October 2026. First edition: two axes, four zones, six placement questions, the first index entries, and the baseline Matrix Vector.
