Achieving the Self-Thinking Business

Imagine an enterprise that can think for itself. Where every decision, from the most tactical to the most strategic, is informed by a deep, real-time understanding of the business and its environment. Where silos of data and functionality give way to a seamless flow of intelligence and action. And where human ingenuity is amplified, not replaced, by artificial intelligence.

This is the promise of the Self-Thinking Business.  

We started Honu in 2021 with the mission of delivering full business autonomy and the promise of a Self-Thinking Business. Back then, we saw that deeper networks and more powerful chips were not the answer to unleashing AI’s full potential in businesses. Fast forward to now and, despite AI’s progress, current approaches to enterprise AI remain fundamentally limited. We see the answer emerging from a reimagining of the intelligence stack to bridge this ‘cognitive’ gap. 

In this article, we delve into the gaps in AI’s abilities and introduce a new approach that elevates AI from limited tactical automation to strategic decision-making, paving the way for fully autonomous organizations.

GenAI and the rise of ‘Autonomous’ Agents

The rise of GenAI, and specifically LLMs, has brought mass awareness of AI and dramatic increases in productivity in specific areas, and propelled 'AI Agents' (also known as 'Autonomous Agents') to the center stage. The cognitive architectures of these agents hold LLMs as their core reasoning engine. While these agents excel in all language processing and information retrieval tasks: the success in rational decision-making has been very limited. 

The dust is now beginning to settle around some of the superfluous claims about generalized agent autonomy, thus giving us a clearer idea of the limitations of the current direction.

Despite some frameworks showcasing reasoning-like behaviors, they still completely miss the mark on what we would consider the characteristics of reasoning: i.e. an ‘understanding’ of data, context, business logic, risk, scenario of possible futures, active experimentation and impact measurement. The eloquence of the LLMs output, by the very nature of how it is trained, is disguising a lack of coherent reasoning. 

The premise that LLMs are capable of building a world model through complex pattern-finding and abstraction of huge corpus data is questionable. The plateauing of the performance of newer foundational models - mostly driven by the limitation of structure that can be extracted by the corpus of data - only further corroborates this hypothesis.

The Cognitive Gap

The biggest business-defining decisions are strategic by nature. For example, instead of asking ‘how to get better website copy or images?’, the more important question should be ‘how can we effectively spend our resources on product development, marketing, or getting organic traffic?’ While the former could introduce a few % points to the bottom line, the latter is usually the difference between a business shutting down or 10x-ing its revenue. 

At Honu , we map the reasoning capabilities of systems onto a decision-making pyramid framework (see below), which represents the different level of decision making sophistication in a company.

Today all AI used in the context of business ( through RPA and task oriented AI agents ) sits, at best, at the tactical base level of the Decision-Making Pyramid. It leans heavily towards automation rather than autonomy, as it stops short of the scope and interpretation of the problem being solved.

Achieving true autonomy requires gluing the base-level components together in order to make the jump of technology delivering holistic decision making (the upper half of the pyramid). 

We’ve labeled the gap in the capabilities of current AI and the shift to move from tactical ( task oriented, fragmented ) to strategic (responsibility/objective oriented, holistic ) decision making as the ‘Cognitive Gap’. 

There are 5 key requirements for any technology to bridge that gap:

1. Breaking the silos: A comprehensive, shared, holistic view of the business.
2. Business logic: Business logic and rules as first class citizens
3. System 2 Reasoning: Scenario analysis, forecasting, simulation, risk understanding and complex planning. Fully explainable.
4. Proactive: Agents appearing dynamically at the right place and time based on business state and context without user input.
5. Smart Ecosystem: Making the larger ecosystem intelligible to agents and how it pertains to the business.

Reimagining the intelligence stack: introducing the Cognitive Layer

It is our conviction that the current trajectory of AI won’t naturally lead to an emergence of these capabilities, regardless of the data and computing resources thrown at the problem. The superhuman AI decision making leap is within reach- but it requires a rethink of the intelligence stack. 

What is needed is a new, supplementary layer that sits between the Systems of Record and Systems of Intelligence (e.g. AI Agents). We call it the Cognitive Layer.

The Cognitive Layer holds a common dynamic representation of the business ( along with its objectives, processes, practices ) that maps to its data, and is intelligible and extendable by the Systems of Intelligence running on top of it. The cognitive layer also provides full contextualisation of the data and information in the business. 

Think of it as a digital nervous system of the enterprise.

Superhuman Capabilities with Less Data and Compute

With deep architectures, such as those found in the transformers space, the representation of the problem and the process of determining the solution are tightly intertwined and largely indistinguishable from each other. In the case of LLMs, we are boiling the ocean with terabytes of data and millions of watts of computing resources trying to build a quasi-generalized world model that is then fine-tuned for specific business decisions. 

The alternate approach we suggest decouples the mechanism that holds the representation of the problem space from the systems aiming to solve, optimize, that problem. This reduces the data and compute requirements by orders of magnitude, making it possible for decision making to be delivered in scarce data and low compute environments.

Furthermore the Cognitive Layer introduces, by design, explainability and interpretability to its reasoning

Flipping from a Pull to a Push Model

The presence of a cognitive layer provides agents with dynamic context and an understanding of the problem space, as well as their role within the larger business. This also means that the context that the agent gets about the business happens dynamically and constantly, without the need for text input from the user. It makes these systems proactive rather than reactive, and able to show up and contribute at the right place at the right time in the decision making cycle. This represents a huge shift from a pull to a push system.

Businesses in the context of the cognitive layer are no longer static. They are always ’on’. It is dynamic in its representation of the business that mirrors what is happening within a live, operating business (be it an additional sales channel, new supplier, etc.). As the organization grows, the business capabilities morph and change, so it is necessary to adjust the processes and practices being applied - and change the way decisions are being made according to the business structure.  

Honu’s Decision Infrastructureᵀᴹ

At Honu we are trailblazing this new approach and soon releasing the first-of-its kind Cognitive Layer technology, which we dub the Decision Infrastructureᵀᴹ.

Currently in closed alpha, our platform is sector and AI-agnostic. Served as a PaaS model.  Our Decision Infrastructureᵀᴹ technology will have an SDK that can be used by application and agent developers alike to leverage this new Cognitive Layer and build superior AI capabilities for businesses.

Honu’s Decision Infrastructureᵀᴹ is an asynchronous, event-driven platform that works with standard Python tooling (JavaScript to follow next).  The platform can be easily extended by developers. Agents are not hosted in the Decision Infrastructureᵀᴹ but run remotely on the providers’ infrastructure - and can be leveraging any of the already existing agent development frameworks.

We are currently working in closed alpha with a few enterprises that span multiple sectors, working on delivering superhuman decision capabilities to their client base and operations. We aim to open up the platform for wider use by the end of the year. 

For further information about the Decision Infrastructureᵀᴹ and the SDK, please visit our website for more info, or follow us on our company page ( Honu ).

If you have an exciting project in mind, or just want to get in touch we'd love to hear from you.