WEAVER · THE RETURN ON INTELLIGENCE COMPANY
The measurable business value generated per dollar spent on AI.
The value created when an organization builds the capability to continuously convert insight into action, action into advantage, and advantage into sustained growth.
56% of CEOs have not seen either a cost saving or a revenue generation opportunity from generative AI. That is the PwC Global CEO Survey, Davos, January 2026, and it should make every operator uncomfortable. The capability is not the problem. Measurement is. Programs spin up, models get demoed, decks get circulated, and somewhere between the pilot and the P&L nobody can say what came back.
We can be specific about that, because we measure it. Among the companies benchmarked by our own AI ROI Index, the average score for AI strategy and ROI alignment is 64 out of 100. For workforce readiness it is 44. A 20 point gap between having an AI plan and having the people, skills and operating systems required to put that plan into practice.
Weaver is an AI-Native Systems Integrator, a joint venture between TQuila, with nearly three decades of enterprise execution, and You.com, one of the pioneers of frontier AI. We build production AI on a fixed fee, and we organize every engagement around one metric.
We call it Return on Intelligence: the measurable business value generated per dollar spent on AI.
This page is how we measure it. Six drivers, four levels, one scorecard, used the same way at the free first score, in the paid assessment, in the build, and on the dashboard your board reads two years later. The framework does not change. That is the point of it. It is how you see yourself move, and how you see the field move around you.
Used in production. Not in pilot.
“We've normalized spending millions on AI while flying completely blind. Imagine running finance without accounting or marketing without analytics. That's where many organizations are with AI today, which is why we've built a free ROI Index — because measuring AI value shouldn't be reserved for those companies able to afford six-figure consulting engagements.”
THE SCOREBOARD IS THE METHOD
Most firms sell a framework and measure something else. We do not. The AI ROI Index is the measurement system, and it is the same measurement system underneath every service we sell.
It is also scored against something no other firm holds. Most maturity models compare you to a survey panel: a few hundred people describing themselves once, in a questionnaire, at a moment in time. Your Index is scored against a live cohort of enterprises, built on two years of anonymized production AI usage data from You.com. Everyone else scores you against opinion. We score you against production.
Six drivers, each scored 1 to 5 against a fixed rubric and converted to a 0 to 100 driver score. Those average into one Index number, which places you in a band from Nascent through Emerging, Developing and Advanced to Leading, against the peer median, the middle half of your peer set, and your percentile inside a cohort of 1,500 or more enterprise peers.
Two rules keep it honest. Where a driver cannot be credibly established it is marked awaiting or unknown and left unscored, never scored as low. And the math is done by the tool, the same way every time: the same formula applies to every company, whatever its size, sector, or who answered the questions.
10 to 15 min
TO YOUR FIRST INDEX SCORE
6 drivers
ONE RUBRIC, EVERY ENGAGEMENT
1,500+ peers
A LIVE ENTERPRISE COHORT, SHARPER EVERY TIME A COMPANY JOINS
Free of charge
COMPANIES SIGN UP AT GOWEAVER.AI/ROI-INDEX
THE SIX DRIVERS
These are the only six things we measure, and they are the same six the Index reports back to you by name. Every score, every roadmap, every build, and every dashboard reports against them. Miss one and the program tends not to survive contact with the real world.
01
Owned by: Business and Value
Is every AI initiative tied to a business outcome with a number on it?
LowUnclear or non-existent ROI tracking. Budget graded on activity rather than outcome.
HighTotal-cost-of-ownership discipline delivering measurable gains, and at the top end AI driving new revenue streams rather than only cost takeout.
02
Owned by: Technology and Data
Can you prove what the system did, and why?
LowNo formal policies and high risk of data leakage.
HighRole-based access control and PII redaction as a floor, then a live control layer that monitors every AI system in one place, then automated governance-by-design.
03
Owned by: Technology and Data
Is your AI estate, meaning everything you have built or bought so far, engineered, or just accumulated?
LowUnsanctioned shadow AI and unstructured experimentation on personal laptops.
HighDeterministic blended solutions, algorithms plus machine learning plus generative AI, then scaled architecture with built-in regression protection so a change in the underlying model never becomes your problem, and finally autonomous multi-agent orchestration across platforms.
04
Owned by: Technology and Data
Does AI sit inside the systems the business already runs, or beside them?
LowFragmented, siloed data with poor hygiene.
HighIntegrations deeply embedded into existing ERP and CRM tools, and at the top end real-time autonomous data actions with continuous learning.
05
Owned by: Business and Value
How much of this is actually live?
What we countAI solutions running in production versus AI solutions still sitting in pilot.
Why it is hereThis is the driver most maturity models never measure, and it is the one that separates a program with momentum from a program with a pipeline of demos. The other five describe capability. This one says whether the capability landed. It is also the only driver a board can verify without taking anyone's word for it.
06
Owned by: People and Adoption
Do your people know what to do with it?
LowA widespread data and AI skills gap across knowledge workers.
HighPractice-wide AI literacy through hack-a-thons and standing office hours, and at the top end a workforce that has moved from executing tasks to orchestrating intelligence.
What the Index is already telling us
01 · Strategy and ROI
64 / 100
Cohort average, the highest of the six
88 percent of companies benchmarked have at least an active, owned AI strategy, and 63 percent have one that drives budget decisions or receives board-level review.
02 · Governance
25 / 100
Peer median, the weakest of the six
Most companies benchmarked cannot yet prove what their AI systems did, or why. Governance is the driver furthest behind the rest of the scoreboard.
05 · Production
12%
Could not count their live solutions
12 percent of companies benchmarked could not say how many AI solutions they currently have running in production. Not a low number. No number.
06 · People
44 / 100
Cohort average, the lowest of the six
60 percent say only specialists or a small group of power users can build AI-assisted workflows, and 59 percent report essentially no employee-built AI agents or skills.
Among companies benchmarked by the Weaver AI ROI Index. Strategy and workforce figures are cohort averages. Governance is a peer median. Weaver AI ROI Index, first benchmark findings, August 2026.
Six drivers is the whole framework. The same six are scored in the free Index, evidenced in the AI Assessment, written into build acceptance criteria, and reported on the dashboard afterward. One vocabulary, one scoreboard, from the first ten minutes to the second year. If a claim does not report against one of these six, it is not a Return on Intelligence claim. And because every company that completes the Index adds to the cohort, these are not our opinions about the market. They are what the market told us about itself, on the record, driver by driver.
HOW YOUR SCORE GETS MADE
The outside view is automatic. The inside view comes from three people, and your Index does not land until all three are in.
Step 1
Put in your company website and the Index goes and gets what is publicly available: financials, your public AI posture, announced initiatives, models in production, partnerships, named AI leadership, and what can be established about data and technology maturity. Every source is captured and kept, every citation is checked against what was actually gathered, and anything that cannot be evidenced is dropped rather than shown. That is your opening picture, in about two minutes, and it is the last thing here that happens without you.
Step 2
Six drivers, and very few people hold all six with real confidence. A CFO can tell you what AI is expected to pay back. He usually cannot tell you the true state of data integration, and he should not have to guess. So the assessment is split into three perspectives, and each one scores the drivers it owns:
Step 3
Each perspective can be sent to the colleague who owns it in one click, and you can see exactly where each one stands: not invited, awaiting, or submitted. If you hold the answers to all three, fill them out yourself. It is the same assessment either way.
Step 4
Until then you have a provisional read, and the page says so, against the count of perspectives received. Every driver nobody has answered yet shows as awaiting rather than being scored low or quietly averaged away. When the third perspective lands, you get your Index, your band, your peer median, and the drivers where the gap to your peers is widest.
We ask for three because the CFO is the one who has to defend the number, and a number built on one executive's best guess about someone else's function will not survive a board meeting. Ten to fifteen minutes each, from three people who already know their own answers, is a small price for a figure you can put in front of a board and a live position against 1,500 or more peers. Most of our clients find the three answers disagree somewhere. That disagreement is usually the most valuable thing in the report.
This is the difference between a grade and a diagnosis. A public-only score tells you how AI-forward your company looks from the outside. Only the internal answers, from the people who actually run the strategy, the estate, and the workforce, tell you what you are getting back. That is why the Index asks you to bring in your colleagues rather than pretending one executive can speak for all six drivers.
Both steps end in dollars. They are different kinds of dollars, and we say which is which.
The Index gives you a modeled figure: your own AI spend and production footprint, read against your size and sector and the peer coefficients in the cohort. That is a defensible range, in the same way a credit score or an actuarial reserve is defensible, and it is enough to know whether a gap is worth a meeting. It is not audited, and we never present it as though it were.
The AI Assessment gives you an evidenced figure: five separate evidence sources, a coverage matrix across roughly 230 use cases, meaning every candidate use case checked for coverage rather than a sample, and leadership claims triangulated against ground truth. Every assumption exposed, a defensible payback period, a number your CFO can take to the board and defend line by line.
The Index tells you where you stand. The Assessment proves it, and prices the climb.
THE FOUR LEVELS
The drivers tell you what we measure. The levels tell you where you sit on each one, and what the next rung actually looks like. Every engagement starts by placing you on this grid, driver by driver. Almost nobody sits at the same level across all six, and the gaps between them are usually where the money is.
Level 1
Experiment and Prepare
Level 2
Pilot and Capability Building
Level 3
Industrialize and Orchestrate
Level 4
The Agentic Enterprise
Strategy and ROI
Governance
Technology
Data
Production
People
Weaver service
Level 4 is the destination, not the finish line. The frontier keeps moving, which is why standing still at Level 4 is the new way to fall behind, and why the Index is built to be re-run rather than framed.
ONE LOOP, FIRST SIGNAL TO BANKED ROI
Most AI programs stall between the pilot and the P&L. Weaver runs one continuous loop: see the gap, price the gap, close the gap, and watch the return land, in dollars, live. Same six drivers at every stop.
01
Know where you actually stand.
Score your AI against 1,500 or more enterprise peers in 10 to 15 minutes. Free of charge, and you leave with your gaps modeled in dollars against your peer set, not handed back as a maturity label.
Drivers: All six, scored once the three perspectives are in.
02
A roadmap with a number on it.
A fixed-fee, fixed-scope, fixed-timeline deep dive, roughly two to three weeks, that puts dual-vector ROI, meaning hard cost reduction plus revenue acceleration, behind every initiative, with a defensible payback period and every assumption exposed. You get a maturity scorecard against the same six drivers, the ROI of climbing at least the next two levels, a risk register, and a sequenced roadmap of vetted Weaver and honest non-Weaver work. If the ROI is not there, we say so.
Drivers: All six, evidenced rather than self-reported, and costed line by line.
03
Governed agents, live in weeks.
Deterministic machine learning and generative AI engineered together, embedded into the existing ERP and CRM stack, with single sign-on, role-based access and audit logging by default, and proprietary regression protection so foundation-model drift, meaning a vendor changing the model underneath you, does not become your problem. The roadmap ships instead of sitting in committee.
Drivers: Governance, Technology, Data, and above all Production.
04
The return, on a dashboard.
Embedded training, Centers of Excellence, hack-a-thons and standing office hours, with AI spend, savings realized, and time to next agent tracked continuously. Your board sees the ROI you were promised.
Drivers: People, and the Strategy and ROI score that only a live dashboard can defend.
Then it starts again. Live deployments raise the Production count, the Index re-scores the six drivers, and the next initiative is priced against a number you have already banked once. The scoreboard never changes shape. Your position on it does, and so does the field around you. That is the next section.
10 to 15 min
TO YOUR FIRST INDEX SCORE
Weeks
FROM ROADMAP TO LIVE AGENTS
Fixed-fee
FIXED SCOPE, FIXED TIMELINE
Always-on
PROOF OF RETURN, NOT A SLIDE
THE SCORECARD MOVES
Most maturity assessments hand you a grade and a PDF. Twelve months later the PDF is worth nothing, because the grade was measured against a bar that has since moved and a peer set that has since improved.
The Index is built the other way. Three things change underneath your score, continuously.
Every company that completes the Index makes the comparison better for everyone in it. The peer median, the middle half, and your percentile are recalculated from a living cohort, not from a survey run once and cited for three years. This is the part we are most serious about keeping free: a benchmark only means anything at scale, and a benchmark nobody can afford is just a proposal in a costume.
What counted as Advanced on Technology in 2024 is table stakes now. Multi-agent orchestration was a research demo two years ago and is a Level 4 expectation today. Your score can fall while your capability improves, because the market moved faster than you did. That is not a flaw in the measurement. That is the measurement doing its job.
Every system that goes live raises your Production count and re-scores the drivers around it. The loop is not a metaphor. It is the same six numbers, measured again, against a peer set that has also moved.
This is why the Index is built to be re-run rather than framed. Return on Intelligence is not a grade you are awarded once. It is a position you hold, against a field that is moving, and the only honest way to know you still hold it is to measure again.
THE FRAMEWORK HOLDS STILL
An instrument that changes cannot measure change. A scale that recalibrates itself every year tells you nothing about whether you lost weight, and a maturity model that quietly rewrites its own criteria between versions destroys the only thing a benchmark is for.
So the Index is fixed on purpose. Six drivers, four levels, one rubric, this year and next. Three things follow from that.
The same six drivers, the same four levels, the same scoring rubric, every time anyone runs it. That is what makes your score next year comparable to your score today, what makes your number comparable to a peer's, and what lets your own delivery team report progress against the same criteria the free Index used in the first ten minutes. We are not redesigning it next quarter.
Progression is the product. You do not get a new framework, you get a higher position on the one you already have. Level 1 to Level 2 on Governance. Level 2 to Level 3 on Data. Pilots turning into production on the driver that counts them. Every Weaver engagement is contracted to move you a named number of rungs on named drivers, and the Index is how you check whether we did.
The genuinely dynamic part is the company you are measured against. As the cohort grows and enterprises mature, the peer median rises. Today the cohort averages 64 out of 100 on strategy and 44 on workforce readiness, and the peer median on governance sits at 25. None of those numbers will hold still, and the ones that move first will decide who is actually compounding a return and who is only funding one. You can hold exactly the same score and still fall in percentile, because your competitors did not stand still. That is not the measurement changing. That is the measurement telling you something you would otherwise find out far too late.
This is why the Index is built to be re-run rather than framed, and why every company that completes it makes the comparison sharper for everyone in it. The scoreboard does not move. Your position moves, and the field moves. Return on Intelligence is not a grade you are awarded once. It is a position you hold, and the only honest way to know you still hold it is to measure again, on the same scale.
WHY THIS WORKS
Most AI firms build first. We prove the ROI first.
The market has been told for three years that AI is a strategy. It is not. AI is a delivery problem wearing a strategy costume, and delivery problems are solved by measurement, not by conviction.
Every AI firm can tell you that measurement matters. Very few will hand you the same six numbers at the free first touch, in the paid assessment, in the build acceptance criteria, and on the dashboard your CFO reads in month eighteen. Fewer still will score those numbers against live production data rather than a survey panel. That consistency is the product. It is why a score from us is comparable to the score you took last year, comparable to your peer set, and comparable to what your own delivery team reports back.
Underneath it: fixed fees, so the cost side of the ratio is known before you commit. A dual-vector financial model, hard cost reduction plus revenue acceleration, before any code is written. Proprietary regression protection that absorbs model drift so production workflows do not. A workforce enablement track that runs alongside the build rather than after it. And a heritage that brings together nearly three decades of enterprise execution from TQuila with frontier AI from You.com.
We don't benefit from dependency. We benefit from referrals.
WHAT THE INDEX WILL NOT DO
A scoreboard is only worth having if it tells you where it is weak. Ours does, in writing.
The opening picture is built from public information, and for a large, well-known company that picture reads like how AI-forward the company presents publicly, not how AI-mature it is internally. The web rewards announcements, appointed AI leaders and published governance frameworks, and visibility scales with company size while genuine maturity does not, at least not as reliably. Expect that opening read to run optimistic, and treat the gap between it and what your own people report as a finding rather than an error.
Where a driver cannot be credibly established from evidence, it is marked unknown and left unscored. Unknown is never scored as low, and an unanswered perspective never becomes a bad score by default.
Every source is captured and kept, and every citation is checked against what was actually gathered. If a source does not appear in the retrieved material, it is dropped rather than shown.
What the public web produces is an informed, evidence-backed starting estimate, not a settled verdict. Your three perspectives are what make the score real, and where they contradict the outside picture, yours win.
“Your data is confidential. We use it for peer comparison and benchmarking, full stop. We will never sell it, share it, or train on it. Trust is the whole model. Lose that and there's no index worth having.”
WHERE TO START
If your AI program has produced demos but not dollars, the gap is almost never the model. It is the absence of a scoreboard everyone agrees on. Six drivers. Four levels. One loop. One number to defend at the board table.
Start with the free Index. It takes 10 to 15 minutes, it costs nothing, and it ends in dollars.
Two to three weeks. Fixed fee. A dual-vector financial model with a defensible payback period. If the number is not there, we will tell you. If it is, we build. Change without the change order.