Honu is on a quest to enable full business autonomy 💡

CV Deep Dive

Today, we’re talking with Imad Riachi, CEO and Founder of Honu AI.

Honu is a startup building cognitive layer technology as part of its quest to enable full business autonomy. Founded by Imad Riachi in 2021, Honu started with a mission to reengineer how businesses operate and empower them with self-thinking and acting capabilities. It’s working on bringing this technology to market via a new platform that it dubs Decision Infrastructure.

Currently, Honu’s Decision Infrastructure is in closed alpha and is being tested by a set of customers ranging from startups to mid-sized enterprises across e-commerce and hospitality. As a team of 10, the startup is working to make its’ platform a critical piece of the stack of every AI-first business.

In this conversation, Imad takes us through the founding story of Honu, their new category of decision-making infrastructure, and Honu’s roadmap for the next 12 months.

Let’s dive in ⚡️

Read time: 8 mins

Our Chat with Imad 💬

Imad - welcome to Cerebral Valley. First off, give us a bit about your background and what led you to found Honu?

Hey there! My name is Imad and I’m the CEO and Founder of Honu. A bit about my background. I come from a family of small business entrepreneurs, which has shaped my thinking about work, and is the original inspiration behind Honu. Academically, I have an undergraduate degree in computer engineering, a masters in robotics, and a PhD in neuroscience. I was fortunate enough to be part of the core team that started the Human Brain Project in Europe. I was one of the pioneers of in-silico neuroscience research: using large computational models of the brain to better understand how it functions and form hypotheses that are then tested in the wet lab.

While wrapping up my PhD, back in 2008, I built my first climate tech startup, which didn't really take off. This however, made me realize that I preferred the faster pace of building things and putting them into the hands of people for them to be used. Then, I moved to London and alternated between working in bigger companies and startups. I spent a couple of years at Goldman Sachs as a quantitative algorithmic trader, was on the founding team of a behavioral finance analytics company, and then joined Facebook, where I co-led their data science team in London.

My last stint before starting Honu was in the energy sector, working with a multibillion dollar utilities group in the UK.Where i worked closely with the founder and set up their data and AI function, before heading and turning around their smart grid and smart tech department where we launched a couple of world-first products in London, which was really cool. After taking a bit of time off, and reflecting on what my next project before starting Honu in 2021.

In a nutshell, Honu has been a long time coming. It’s an idea I've been thinking about for 10-15 years. The big vision behind Honu is to build technology that enables businesses to become fully autonomous, allowing them to self-think and execute on their own. Having spent more time around small business owners as I was traveling in asia, I felt that such a technology would have the power to fully democratize entrepreneurship and allow everyone, everywhere to start a business. The impact of such an endeavor, and the magnitude of the problem to be solved resonated with me, and hence I embarked on this journey, and it’s been super exciting one so far!

How would you describe Honu to the uninitiated developer or enterprise team?

Our product is a platform we call Decision Infrastructure. It does not fit easily into the traditional categories of today’s stack. It’s a new ‘cognitive layer’ that sits between the data (or systems of record) and your AI (or systems of intelligence).

The Cognitive Layer sits between the systems of record and systems of intelligence

This layer acts as a digital nervous system for the business, and builds a common shared mental model of the business that is dynamic and intelligible to the various AI ( Agents ) running on top of it. What really sets us apart is that we’ve reimagined what the intelligence stack needs to look like to enable AI to deliver strategic and executive decision-making in business.

Even before generative AI really took off, our view has been that for AI to effectively take on strategic decision-making roles, you can’t rely on ever deeper models, bigger data, and more powerful compute.The current direction of AI is limited and the superintelligence we aspire to will not come from here.

It’s a reimagining of how AI can function to make strategic decisions.We believe that the intelligent stack can be built differently, and by decoupling the representation of the problem space from the solution, you can create an architecture that makes strategic decision-making much easier. This approach is less computationally expensive and scales more efficiently.

That’s why we’ve been in research mode for a while, working on our approach. But now, we’re starting to roll out our MVP, and we’re super excited about it.

The Cognitive Layer - A real game changer for AI in business

What are some of the reasons you’ve decided to focus on the infrastructure for decision making as a problem that you're looking to solve?

We have a framework that we think about when it comes to the types of decisions that happen in a business on a day-to-day basis. It's a simplistic model, but it captures the problem quite well. We call it the decision-making pyramid. If you imagine moving from the bottom of the pyramid to the top, you encounter increasing levels of sophistication.

At the bottom layer, you have the operational level. This includes workflows and manual tasks that need to happen. The next layer is tactical, which involves more task-oriented activities. With the rise of generative AI, a lot of these tasks are being taken over by agents. For example, if you need a blog post written or an image improved, these tasks are directive and straightforward.

At the top, you have strategic decision making. These are the decisions that require full context of the business and the options available to the business owner e.g. what areas of the business to invest in, different supply chain options, objective setting, financing…

The Decision Making Pyramid

Moving above the tactical level to the strategic level requires a significant leap in AI’s sophistication, which we refer to as the cognitive gap. Overcoming this gap means having a system that can look at your business holistically, understand your business logic and processes, and reason through various scenarios. This system needs to comprehend risk, manage finite resources, and integrate all these elements to make informed decisions. Currently, the AI available falls short of doing this, and that’s the problem we aim to solve.

To us, automating tasks like writing blog posts or improving copy is just that—automation, not autonomy. To achieve true autonomy, you need to tie all these tasks together into a coherent representation and deliver strategic decision-making on top of that. To make this a bit more concrete, the biggest problem, especially for small businesses, isn’t just needing help with tasks like creating a blog post or optimizing a campaign. Sure, those are important, but they’re not the make-or-break decisions.

The critical decisions for a business are higher-level ones, like, “I have $5,000 in my bank account—what should I do with it?” Should this money go toward marketing, creating blog posts, or hiring someone to redesign the website? These decisions involve resource allocation and thinking about the business as a whole. They are the kinds of decisions that can make or break a business.

When we talked to small businesses, asking them about their top three problems, we quickly realized that their perceived top problems often weren’t their actual top problems. This insight reinforced the idea that if we truly want to achieve a fully autonomous business, which is our ultimate goal, we need to solve for strategic decision-making. That’s why we introduced a new approach called the cognitive layer, which makes strategic decision-making possible.

This cognitive layer addresses the various challenges necessary to bridge the cognitive gap, sitting between the data and AI. It’s designed to handle the complexities of making high-level decisions that are crucial for the success and autonomy of a business.

Could you tell us a bit about your user base today? Who’s finding the most value in what you’re building with Honu?

Absolutely. As soon as we had a small demo ready, the interest was immediately widespread, cutting across sectors and industries. We had people from enterprises, consulting, e-commerce, restaurants, hospitality, and even hospitals expressing interest. They all saw potential in having something that could help strategize and improve their operations.

That said, although we see broad value across many industries, we’re currently focused on small business owners. The pilots we’re working on right now are with companies that serve small business owners across e-commerce and hospitality. This helps us scale our reach and impact as many small business owners as possible, to help them make better, quicker day-to-day decisions in their businesses.

Given that you’re in alpha mode currently, what are some of the ways you’re measuring the impact that Honu is having on your early customers? How are you measuring success at this stage?

The first and most simple measure is growing the number of businesses actively using the Decision Infrastructure. Growth will be achieved by partnering with some of the most innovative providers of platforms to small businesses.

We will then measure success by the engagement with agents powered through the Decision Infrastructure ecosystem. We have already had many “aha” moments where small businesses say, “Wow, I hadn’t considered that”, and want to delve deeper. As we build the offering and number of services available we’ll be looking closely at how business owners interact with these agents, and what decisions they make based on these interactions. And as a second step allow for these recommendations the option to be implemented automatically through tactical AI agents.

Once the loop is closed from strategic decision, to tactical, action and then measurement of impact, we will be able to measure success directly in terms of $$ uplift to the bottom line of a business.. This will mean we will have one of the purest views of the impact of not only our own Decision Infrastructure, but also of the services and agents that provide value to small business owners.

What's been the hardest technical challenge around building Honu so far, and what do you anticipate will be a difficult challenge going forward?

The important context here is that we had to create technology that didn’t really exist before, including developing new coding patterns. The hardest part was figuring out how to build a layer that can create a dynamic representation of a business and encode it in a way that’s abstract enough to apply to different types of businesses - after all every organization has its own way of doing things and its own unique set of challenges. Building this system has been incredibly difficult because of the multiple cogs that need to work in harmony. We had to make it extensible and marry a lot of different computer science and software engineering approaches and then tie all that together with AI. We currently have 2 patents pending for this pioneering technology.

Honu’s Decision Infrastructure capabilities

Another major consideration was having the team apply their experience with building AI systems at scale, in ensuring that our approach can scale to tens of thousands or even millions of users without compromising on important aspects like security, permissions, and observability. We’ve baked these considerations into our design, but it’s been quite complex to put it all together. We’ve been focusing a lot on the initial setup and challenges, and now it’s going to be very interesting to see how our design scales beyond those first use cases. One of our big strategies is not just to build a layer that works within the existing infrastructure but also to open up our SDK so that third parties can extend and contribute with their own agents and services.

We’ve built this in a way that makes the developer’s life so much easier. We are abstracting away the data layer and communication layers, and taking care of credential management, and fine grained read and write permissions. All of this needs to work seamlessly. I’m really looking forward to seeing how this ecosystem comes together. Once it does, it will serve as the connective tissue for how intelligence is delivered and how people can embed their expertise into our platform, distributing it to the businesses that need it.

We’ve built it with this in mind, so believe it will stand up to the test as we start scaling.

There's been a huge interest in agents and agentic workflows. Has that been a feature in the decision-making infrastructure that you're building at Honu?

We see agents as an intrinsic part of the ecosystem we’re trying to create. Our Decision Infrastructure includes an SDK that can be used to develop agents. We're not looking to replace any existing frameworks. Instead, we’re building a system—think of it like a nervous system—where agents can connect and understand what’s happening in the business, and this allows them to be proactive rather than reactive.

Honu’s Agents SDK

You can use any external framework available to build your agents. For example, your agents could be built using frameworks like Agent Ops or SuperAGI, or they could be simple Zapier automations. They could even be as basic as a small piece of code with an Excel spreadsheet behind it. We’re pretty agnostic about that because the real value we bring is in creating that common mental model of the business. This makes it intelligible to different systems of intelligence—like autonomous agents—so they can understand, collaborate, and make better strategic decisions for the business.

Lastly, tell us a bit about your team. Are you hiring, and what do you look for in prospective members joining the team?

We’re a team of ten people, working remotely from different places. Most of our team members are engineers, and quite a few of them have been CTOs in the past. Many of our engineers have experience working on mission-critical platforms that need to work at scale, and a lot of them have worked on building AI-driven systems.

The mix of backgrounds and experiences really helps us with what we're doing. When it comes to the qualities of our team, we’re passionate technologists. We’re all about technology, and we approach new technologies with a healthy dose of skepticism.

We’re always striving for excellence. It’s not about just getting things done quickly. We really take the time to think things through and understand the systems we’re building. Since we’re dealing with complex systems, it’s important that everything works well together. If you only do 80% of each part right, you end up with a system that doesn’t work most of the time. So, we’re always looking for people who think that way.

Conclusion

To stay up to date on the latest with Honu, follow them on X and learn more about them at Honu.

Read our past few Deep Dives below:

• Zilliz Cloud is powering vector search for the enterprise 🔋

• Kolena aims to redefine quality standards in ML 🎛

• Pinecone Serverless is turbo-charging your time-to-prod ⚡️

• Cleanlab is automating data curation at scale 🌐

If you would like us to ‘Deep Dive’ a founder, team or product launch, please reply to this email (newsletter@cerebralvalley.ai) or DM us on Twitter or LinkedIn.

Join Slack | All Events | Jobs