Ten Years from Sensemaker to GAMMA

The SuperCIO thread started earlier

One part of the GAMMA story predates Quantiply and Sensemaker. The SuperCIO thread did not begin with agentic AI. It began around 2009, in the early cloud era, when it was becoming clear that the CIO role would evolve beyond managing enterprise IT. Cloud was more than a hosting model. It was the start of a deeper platform shift, in which the enterprise became elastic, programmable, scalable, and increasingly dependent on architecture-level decisions about cost, resilience, speed, and operating leverage. That was the first version of the SuperCIO idea.

Each wave of technology expanded the role rather than replacing it. The traditional CIO managed information systems, the cloud-era CIO had to architect elastic infrastructure, the Big Data era asked the CIO to make the enterprise observable, and the first wave of AI asked the enterprise to become predictive and decision-aware. Cloud changed infrastructure, Big Data changed visibility, AI and machine learning changed prediction, and agentic AI now changes action. That progression is why the agentic enterprise reinforces the SuperCIO thesis so strongly.

The SuperCIO was always about more than IT management. It was about the executive capacity to read technology as enterprise architecture, which in the cloud era meant elastic infrastructure, in the Big Data era meant observability at scale, in the AI era meant predictive intelligence and decision support, and in the agentic era means governed autonomy. That is where GAMMA fits. GAMMA gives the SuperCIO a maturity language for a new responsibility, a way to see where the enterprise is experimenting, where AI is assisting humans, where agents are coordinating work, where bounded autonomy is appropriate, and where adaptive autonomy can emerge without losing control.

GAMMA is the next expression of the SuperCIO thesis rather than a separate idea. The SuperCIO began as the executive archetype for the cloud-platform enterprise, evolved through Big Data and AI into the architect of enterprise intelligence, and in the agentic enterprise becomes the executive architect of governed autonomy.

2014: Quantiply and the decisioning thesis

The more direct origin of GAMMA goes back to Quantiply, founded in 2014. That same year we secured E\*TRADE as a design partner and joined the SAP Startup Focus program, which gave us both a serious enterprise design context and a strong technology foundation. Our vision for Sensemaker was broader than AML, KYC, fraud, or compliance from the start. We did not set out to build an AML product. We set out to build enterprise decisioning infrastructure on SAP HANA, because enterprises needed a way to turn fragmented signals into better decisions.

SAP HANA gave us the substrate to think differently, letting real-time processing, large-scale data fusion, graph-like relationships, rules, models, risk signals, and human feedback come together in one decisioning architecture. The ambition reached past analytics and alerts toward decisioning itself. We were asking what it would mean for an enterprise to have a layer that could sense, contextualize, reason, recommend, support action, and learn. The market did not yet use language like agentic enterprise or governed autonomy, and yet the early pattern was already there: enterprise intelligence was moving toward decision infrastructure.

2015: the Big Data Foundry and the first Sensemaker

In 2015, still connected to the broader PARC work, we launched the Big Data Foundry, a petabyte-scale real-time algorithm machine built with SAP, Cisco, and Hitachi. It reinforced that enterprise intelligence was moving past reporting toward decisioning, and it clarified that infrastructure alone was not the answer. The real question was what the enterprise would do with that intelligence. That same year we built the first working version of the Sensemaker Decisioning Platform, which was already trying to solve what GAMMA later formalizes as the decision-action loop, even though we did not use that language then.

This is why 2015 matters so much in the ten-year arc. In 2015 the decisioning infrastructure thesis became real. In 2025 that same thesis returned as governed autonomy.

2015 to 2016: the SAP roadshow and the market-readiness lesson

In 2015 and 2016 we went on the road with SAP to test product-market fit. The conversations were valuable, because enterprise leaders could see the power of the thesis. They understood that decisioning was fragmented, that data alone was not enough, and that risk, compliance, operations, and customer intelligence all suffered from the same problem: signals did not naturally become trusted decisions. The roadshow also exposed a hard truth. The enterprise was not ready for a broad decisioning infrastructure platform.

The thesis was right and the category was too horizontal. The market did not yet have the language, operating model, buyer clarity, or governance maturity to adopt a broad enterprise decisioning platform. This became one of the major lessons of Quantiply, and I have carried it since: a technology vision can be right before the market is ready to adopt it. The issue was not only technology. It was enterprise readiness, whether the enterprise could absorb the system, change its decision workflows, trust AI-supported recommendations, govern what the system discovered, and reorganize responsibility around decisioning. At the time the answer was often not yet, and that lesson became one of the foundations of GAMMA, which is less about what autonomous systems can do than about whether the enterprise is mature enough to let them do it.

2016: the wedge became Autonomous Risk Intelligence

By 2016 we found the more compelling wedge, Autonomous Risk Intelligence. AML, KYC, fraud, and compliance gave the broad decisioning thesis a high-consequence starting point, where the pain was clear, the buyer was clearer, the workflows were under pressure, the cost of error was high, and the need for explainability and auditability was real. In hindsight this taught another lesson. Broad decisioning infrastructure may be the ultimate need, and adoption usually starts through a high-consequence wedge.

Autonomous Risk Intelligence also taught us what governed autonomy would eventually require. In financial crime the system could not simply generate a score or automate a decision. It had to provide context, explain why something mattered, preserve a trail, keep humans in the loop at the right points, and support decisions that could be defended. That is very close to what GAMMA now formalizes. The original Sensemaker thesis was broad decisioning infrastructure, the market required a narrower wedge, and Autonomous Risk Intelligence became the starting point, which shows that GAMMA continues the same thesis under a more mature context rather than arriving as a disconnected new idea.

2016 to 2017: Pilot Purgatory

From 2016 to 2017 we secured dozens of pilots. At the time it felt like validation, and in one sense it was, because enterprises were interested and could see the potential. Sensemaker surfaced risk, detected relationships, enriched context, and showed insights the existing workflows were missing.

We were also being pulled into what I now call Pilot Purgatory. The pilots proved that the technology was interesting; they did not prove that the institution was ready, and that distinction matters. A pilot can show that a system works without proving that the enterprise has the decision rights, governance, process maturity, human oversight, and operating discipline to absorb it. That is one reason GAMMA needed to be a maturity model, because technology validation and enterprise readiness are not the same thing.

2017: Series A and the responsibility to prove

In 2017 we secured Series A funding, which validated the seriousness of the opportunity and let us keep building the product, the team, and the market. It also raised the responsibility to prove that the platform could move beyond pilots into scalable adoption. The subtle lesson here is that capital can extend runway without solving market maturity. The enterprise still has to be ready, the buyer clear, the adoption path practical, and the operating model able to absorb the system. The journey from experimentation to adoption has stages, and those stages have to be visible, or companies confuse activity with maturity.

2017 to 2019: sidecar projects and the accountability problem

From 2017 to 2019 we kept building and validating Sensemaker, including more than a dozen sidecar projects to prove it worked as promised. It did work, and in some cases it worked in ways that created new organizational discomfort. Sensemaker found things outside the original AML scenarios, surfacing patterns, relationships, anomalies, and risk signals beyond the initial scope. Technically that was powerful. Operationally it raised a new question: what happens when intelligence reveals risks the organization is not ready to own?

That was a profound lesson. When a system makes hidden risk visible, it creates new accountability, and the organization has to decide what to do with that knowledge, including who owns it, who investigates it, who reports it, and whether discovery itself creates obligation. This is directly relevant to the agentic enterprise, because AI agents will not only automate known work. They will surface new patterns, expose hidden dependencies, and reveal gaps in policy and accountability. That is why GAMMA puts so much weight on operating evidence, decision provenance, and human accountability.

The lesson under the wedge

AML and KYC were teaching environments as much as use cases. They taught us what happens when AI enters workflows where consequences matter. A false positive wastes time, money, and customer trust. A false negative creates regulatory and reputational exposure. A black-box recommendation, an alert without context, a decision without provenance, and automation without human judgment all fall short in the same way. The system had to help humans decide better instead of replacing them, and that became central to how I now think about governed autonomy. The best enterprise AI systems do not strip out human judgment indiscriminately. They reposition it at the right point in the decision architecture.

That idea became one of GAMMA’s core principles. Autonomy is worth pursuing only where it can be trusted, and the appropriate level varies by decision. Some decisions should be automated, some augmented, some coordinated by agents, some executed autonomously under governance, and some kept deeply human. Sensemaker taught this in the language of risk intelligence, and GAMMA generalizes it for the agentic enterprise.

2020 to 2021: COVID, funding, and market maturity

In 2020, COVID changed the environment. It became much harder to secure the funding and market maturity needed to push the platform forward, as enterprise buying cycles turned cautious, uncertainty rose, capital tightened, and the gap between technical promise and institutional readiness grew more visible. From 2020 to 2021 we explored several M&A options.

The period reinforced a hard lesson: technology maturity, market maturity, capital timing, and institutional readiness all have to converge. Being right architecturally is not enough on its own. The market has to be ready, the buyer clear, the adoption path practical, the governance model credible, and the capital runway long enough for the market to catch up. GAMMA is partly a response to that lesson, giving enterprises a way to evaluate readiness before they overcommit to autonomy, and giving builders a way to judge whether the market is ready for a broad platform or needs a wedge.

2022 to 2023: Eventus and the integration lesson

By 2022 and 2023, Quantiply merged into Eventus, and that transition brought a different kind of learning. When AI decisioning infrastructure enters a broader enterprise platform, integration matters as much as invention, because the system has to fit into workflows, product strategy, customer expectations, regulatory needs, and operational trust structures. AI does not live in isolation. It lives inside business processes, institutions, risk appetite, customer commitments, compliance obligations, and board-level accountability. GAMMA reflects that. It is a model for enterprise readiness around governed autonomy, not a model of AI capability alone.

2023 to 2025: 451 Ventures as the experimentation machine

After the Eventus transition, 451 Ventures became the place where I could step back from a single product, platform, or market wedge and ask a broader question: what venture architecture the next generation of intelligent, regenerative, and high-trust systems would need. From 2023 to 2025 it grew into more than a venture-building vehicle. It became my experimentation machine for testing a broader systems thesis. This was when my thinking about regenerative systems became structured, as the Regenerative Prosperity Model, the RPM Flywheel, and the broader 451 architecture took shape, and I stopped thinking about AI, decisioning, or enterprise platforms in isolation.

That mattered for GAMMA. One lesson from regenerative systems thinking is that maturity cannot be measured by static documentation or isolated output alone, because a system’s health shows through its flows, feedback loops, resilience, and capacity to adapt. GAMMA’s idea of living operating evidence did not come only from enterprise AI governance. It also came from regenerative systems thinking. A living system cannot be understood through a periodic checklist. It has to be sensed continuously, its signals interpreted, its feedback loops monitored, and its capacity to adapt governed. That is what the GAMMA Nervous System tries to do for the agentic enterprise.

2022 onward: the farm that became a systems laboratory

If 451 Ventures was the experimentation machine, the experiment that taught me the most began on October 10, 2022, when I wrote the first executive brief for Agsiri. The idea on the page was modest. Agsiri would make U.S. farmland accessible to accredited investors through fractional ownership, place capable farmers on the land, and coordinate the relationship between capital and operation. It looked like an investment platform. It turned out to be a systems laboratory.

The trouble surfaced the moment we tried to make farmland investable. A farm is not ordinary real estate. It is a biological system and an operating enterprise at once, a financial instrument and a source of food, a store of generational knowledge and a member of a community. The instant we reached in to make it investable, we inherited every layer of that complexity. The questions that followed were not features waiting to be specified. They were the architecture of trust: how to value land whose future depends on soil, water, and operator skill; how an absent owner knows whether an operator is improving the asset or quietly mining it; how to turn contracts and regenerative intentions into actions someone can monitor and enforce.

Those questions pulled Agsiri past farmland crowdfunding toward a regenerative land and farm operating system, and they pulled a family of system explorations into being inside 451 Labs. I want to be precise about what these were. The names that follow are working names for theses under validation, explorations rather than shipped products, each one drawn out by a pressure the farm created rather than chosen from a technology menu.

The distance between an idea and a working system produced the thesis we call Pluto, our exploration of disciplined translation from human intent into domain models, schemas, and tests. The question of who is allowed to do what produced the thesis we call Pola, an attempt to lift policy out of documents and memory and make it explicit and enforceable at the moment of action. Work that refused to move in a line produced the thesis we call Satori, less a way to move tasks between queues than a way to orchestrate an outcome across people, agents, and systems.

The long memory of agriculture, set against the short memory of early models, produced the thesis we call Ana, concerned with preserving context, provenance, and the difference between a fact and a verified result. The growing power of agents to act produced the thesis we call Actra, which separates reasoning from authorization and authorization from execution and leaves a durable receipt. The need to keep testing assumptions against signals produced the thesis we call Astra, which later became the spine of the What If Next work and the method I now describe as Recursive Self-Adaptation.

The farm then pushed past software into the physical world, where the field, not the screen, reveals whether a model is wrong. That produced the theses we call Salinas and Tukey, concerned with perceiving the farm as a living physical system and translating approved plans into bounded missions for people and equipment. A later thesis, KOSHAS, took shape as a kernel meant to draw many of these threads together for orchestrated human and synthetic agents.

One pressure mattered more than the rest for this reflection. Agsiri could claim a property was growing healthier and show an investor a confident dashboard, and none of that, by itself, was trust. A platform that raises capital and manages an asset should not be the sole authority certifying its own performance. That conviction produced the thesis we call Wellzai, concerned with independent biological verification, and it pointed directly at GAMMA as the assurance thesis for agents, processes, and operational integrity. The farm taught me that the coming AI economy would rest on more than intelligence and automation. It would rest on proof.

This is where the two origin stories of GAMMA meet. One lineage runs through Sensemaker, decisioning infrastructure, and Autonomous Risk Intelligence, where the lesson was that intelligence in high-consequence workflows has to be explained, governed, and evidenced. The other runs through Agsiri, regenerative land, and the demand for independent proof, where the same lesson arrived from a different direction and on living ground. Both threads insist that autonomy must be earned, bounded, observable, and revocable, and that maturity shows through the living flow of work rather than through a static document. The farm did not contradict the decade of enterprise lessons. It confirmed them, and it taught me to read decisioning infrastructure as part of a living system.

2023 to 2024: reimagining sensemaking in the GenAI era

While 451 Ventures widened the field for regenerative systems thinking, GPT and GenAI reopened the original Sensemaker question in a new technical context. Capabilities that once required specialized architectures became accessible through foundation models, natural language interfaces, retrieval systems, copilots, and emerging agent frameworks. The underlying problem did not disappear. It became more urgent.

GenAI made sensemaking easier to access and governance harder to maintain, because far more people could now ask questions, generate analysis, summarize information, draft decisions, and build workflows. That raised the old questions again, about which context is used, which sources are authoritative, which policy applies, what the decision trail is, who approves the action, and what happens when the system is wrong. This period helped me see Sensemaker as part of a longer arc rather than a product from the past. The market was finally moving toward sensemaking, and sensemaking without governance could become a new risk.

2025: agentic AI made the question unavoidable

By 2025, agentic AI made the issue unavoidable. Copilots were no longer the whole story. Agents were beginning to enter enterprise workflows, where they could plan, reason, call tools, coordinate tasks, and eventually execute actions. This brought the original Sensemaker thesis back in a new form, and with a reversal. In 2015, broad enterprise decisioning infrastructure was too early. In 2025, the opposite risk appeared, that enterprises might adopt agentic systems faster than they can govern them. GAMMA emerged from that tension.

The final GAMMA paper defines the framework as the Governed Autonomy Maturity Model and Assessment for the agentic enterprise. It argues that autonomy should mature in stages, that enterprises should not jump from human-led workflows to autonomous execution without decision infrastructure, and that maturity should be evidenced from the living flow of work rather than manufactured through documentation theater. It also introduces the idea that GAMMA itself should become agentic, a living governance system that observes decision-action loops, assembles operating evidence, detects drift, updates maturity scores, and guides enterprises toward trusted autonomy. That was the major intellectual output of this reflection.

Looking back, the ten-year arc reads as one progression. The decisioning thesis became real in 2015, met a market that was not ready, narrowed into a high-consequence wedge, passed through pilot interest that was never institutional readiness, survived a hard funding climate and an integration, widened into living-systems thinking at 451 Ventures, was tested on living ground through the farm, and returned through GenAI and agentic AI at a far larger scale. The enterprise now has to mature from experimentation to assistance, from assistance to coordination, from coordination to governed autonomy, and eventually to adaptive autonomous operation. That is the arc from Sensemaker to GAMMA.

The CMM lesson

Part of GAMMA’s thinking came from reflecting on the Capability Maturity Model. CMM gave organizations a useful way to move from ad hoc toward repeatable, defined, managed, and optimizing processes. In many organizations it also became documentation-heavy, where maturity turned into evidence production, process formalism, and audit preparation. I did not want GAMMA to repeat that.

Agentic systems are dynamic. Their behavior changes as models, prompts, tools, data, policies, users, and workflows change, and static documentation cannot govern dynamic autonomy. That led to a key design principle. Where CMM matured software process through documentation discipline, GAMMA matures autonomous enterprises through living operating evidence, collected from the actual flow of work: agent actions, workflow events, approvals, overrides, exceptions, policy checks, tool calls, escalations, audit trails, outcome data, drift signals, and business impact. That is what makes it appropriate for the agentic era.

From assessment to nervous system

The other major 2025 insight was that GAMMA should not stay a static assessment. It needs a nervous system. The GAMMA Nervous System is the operational layer that continuously senses, interprets, scores, recommends, governs, and learns. It watches where decisions are happening, where agents are participating, where autonomy is increasing, where humans are overriding, where policy is unclear, where drift is emerging, and where autonomy should expand or be constrained. That is what turns GAMMA from a maturity model into a living governance system.

The five levels give the enterprise a progression model.

The capability dimensions give the assessment its structure, and the GAMMA Score gives executives a way to communicate maturity. The nervous system is what makes the model operational. Without that layer GAMMA risks becoming another framework. With it, GAMMA becomes a way to govern autonomy continuously.

The companion frameworks

Another set of ideas shaped GAMMA’s thinking, and I do not want to overload the core paper with all of them, so I treat them as companion frameworks. Some are operating roles, and some are theses still under exploration in 451 Labs. The SuperCIO is the executive architect of governed autonomy, a thread that began in the 2009 cloud era and matters more in the agentic enterprise because the enterprise now needs executive architecture for governed autonomy rather than technology modernization alone. The AI Builder is the distributed creator of agents and workflows, an important new persona and also a source of shadow-AI risk when it is ungoverned. Ava, the Agentic Value Architect, is the role that translates business friction into agentic use cases and judges what level of autonomy is appropriate.

Two of the companion ideas are working theses rather than roles. The Clearstory thesis explores decision memory and provenance, preserving the story of how decisions were made. The Tukey thesis, which began in the farm’s physical layer, also explores how weak signals and emerging patterns become inputs to maturity learning. These are connecting tissue, and the core GAMMA paper should stay focused on the maturity model, the decision-action loop, the operating-evidence concept, and the nervous system. The clean framing is that GAMMA defines maturity, the Operating Roles Framework defines responsibility, and the nervous system makes maturity observable in real time.

What GAMMA is really trying to name

At its deepest level, GAMMA is more than another AI governance framework. The market already has many, and many are useful, establishing important principles: AI should be fair, transparent, accountable, safe, secure, and human-centered. Those principles matter, and for the agentic enterprise they are not enough. As AI moves from assistance to action, enterprises need a more operational way to understand how autonomy matures inside real workflows, distinguishing systems that merely support human work from those that coordinate across tasks, act within defined boundaries, or continuously adapt to feedback.

They also need evidence. Before autonomy can expand, an organization has to know what proves readiness: the quality of the data, the maturity of the workflow, the clarity of human accountability, the strength of governance controls, the reliability of the system, and the ability to detect drift, preserve provenance, and intervene when necessary. GAMMA is designed to name that missing capability. It gives enterprises a way to understand not only whether AI is governed in principle, but whether autonomy is ready to operate in practice, and it helps leaders decide when to proceed, when to pause, when to redesign the workflow, and when to strengthen the control environment before scaling further. That is the real purpose of GAMMA: governance as a living enterprise capability, embedded in workflows, measured through evidence, strengthened through feedback, and matured as the organization learns to govern intelligent work at scale, rather than governance as a policy document.

2025 reflection

As I close 2025, GAMMA reads as both a year-end reflection and a ten-year synthesis. In 2015 the question was whether enterprises could use real-time intelligence to improve decisioning. In 2025 the question is whether enterprises can govern autonomous intelligence as it begins to participate in action. That is a much bigger question. The journey from Sensemaker to GAMMA taught me that enterprises do not fail only because their technology is weak. They fail when technology arrives faster than the institution’s ability to absorb, govern, and trust it.

Each chapter left a lesson, and together they form the model.

GAMMA brings those lessons together.

GAMMA is a maturity model for the real operating question of the next decade: how an enterprise knows which decisions can be assisted, coordinated, governed, automated, or kept human. That question had been forming for ten years across both the enterprise and the farm, and in 2025 it finally had a name. The conviction it carries grew clearer to me this year. The aim is trusted autonomy rather than maximum autonomy, reached by helping enterprises move from experimentation to assistance, from assistance to coordination, from coordination to governed autonomy, and eventually to adaptive autonomous operation, always with discipline. Autonomy has to be earned, and in the agentic era trust has to be evidenced from the living reality of work.