DeepTech Motion, Transform a Research Lab Into an Innovation Factory

• Blaise Vignon, Chief Product Officer at Alice & Bob

One of the hardest tasks in a deep tech startup is to wed research with engineering so that together they can turn ideas into innovation

In essence, we must somehow connect the free spirits of creative physicists with the relentless succession of achievements expected from a modern startup. And this is not something that will just happen of its own accord.

Scientists in deep-tech startups are supposed to invent things. But if their investment in time is to pay off, they must fit into a system they may well not have encountered during their academic careers. In the industry, we call this the R&D pipeline.

And at Alice & Bob we have put together just such a specific methodology. We like to call it “DeepTech Motion.” It’s a never-ending cycle of four phases, as shown in the illustration below: we generate ideas, simulate, build, and experiment, thereby generating more ideas. We think of DeepTech Motion as a product in itself, and these phases are its key features.

Ideation

Ideation, the production of ideas, is inspired by theory and informed by our current state of knowledge. New ideas are formalized into roadmaps driven by an OKR methodology: Objectives and Key Results are applied at the company level and cascaded down through departments to our various working groups. But crucially, even though we are OKR driven, we always allow some bandwidth to entertain and test less mainstream ideas coming from our academic partnerships.

Simulation

Simulation and hardware design are growing at a tremendous rate. The results are carefully stored and used to feed data analysis. This means we can contrast a new idea with all previous ones at the bat of an eyelid and make quicker decisions about where to go next.

Fabrication

Fabrication is carried out in the cleanrooms of our academic partners. And needless to say, the chips we produce, scrutinized in the most exhaustive way, are fast converging on key quality metrics.

Experimentation

Experimentation validates our scientific progress toward fault-tolerant quantum computing (FTQC), and so is one of A&B’s most valuable assets. We track it using state-of-the-art knowledge management systems, hosted on sovereign storage.

It’s like a kind of dance and it creates a natural sprint rhythm within the company.

Having a process is one thing …

… but making it work in the long term and making sure it relentlessly delivers results is a very different matter. Here is how we do it at Alice & Bob.

The cold, cold but fast beating heart of our company is a roomful of cryostats, or in more familiar parlance, fridges. These marvels of engineering are needed to get our chips down to temperatures even colder than outer space so that they can do their work undisturbed. They wouldn’t work otherwise, but it leads to a major bottleneck: whenever we want to test a new design, it takes four days to go through the entire cycle of warming up, wiring, and cooling back down. This is why we consign it to weekends, every other week. It’s like a kind of dance and it creates a natural sprint rhythm within the company.

Leadership and Organization

So, we have built our rituals around this natural heart rate. Inspired by the Agile manifesto, we break our projects into phases and aim for continuous collaboration and improvement. Day-to-day execution is tied to semester level goals expressed in OKRs through widely shared execution plans.

Crucially, even though we are OKR driven, we keep some bandwidth to test more creative ideas originating from our academic partnerships. We need this to enrich our roadmap with groundbreaking ideas. And we can afford to do so thanks to a solid execution pipeline.

Metrics and tracking tools

But we need to keep track of how we are doing. As mentioned, DeepTech Motion is a product, and as such, it has metrics. This is no place to go into details, especially as some of those details are key ingredients in our secret sauce, but here are just a couple of teasers to whet your appetite.

Discipline in the execution of our plans

We use roadmaps to tie the work in a sprint to the company’s OKRs, which are defined at the semester level. We are a hardware company, which comes with its own rigidities. But we draw inspiration from the “responding to change” part of the Agile methodology.

Monitoring the fabrication process

Systematic data collection allows us to think reflectively about our fabrication processes. Today, we are beginning to thrive as a data driven company, relentlessly pushing our plans down the pipeline in ever more effective ways.

Below, for example, is a wireframe of the graph we use to compare chip specifications with actual measurements.

Looking inwards at our scientific methodology

When it comes to our research team with its steady supply of doctoral students, who together make up a significant proportion of our workforce, the emphasis is less on relentless execution and more on making sure they are supported in the right ways. To achieve this, we visualize DeepTech Motion through a live data system which continually gathers and displays all time-sensitive information, including things like cryostat status, chip usage, and qubit interactions.

So, how are we doing?

With this system in place, we can get from idea to performance results in less than a month. It means we can afford to explore pioneering qubit designs because we are better at managing the process uncertainties associated with our innovation pipeline. The risks involved in a novel undertaking can be better measured and contained, without jeopardizing an entire execution plan. Our streamlined process has allowed us to pursue the greenfield ideas which led us to our two new cat qubit designs, the concept cats, affectionately named AutoCat and TIGRO. It’s initiatives like this that enable us to file more than one patent every month on core components and architectures.

Clearly, there is still a long road ahead. Even though DeepTech Motion is being more and more widely applied and still gaining in features, we are already thinking about the next step. Now that we have a handle on the uncertainties associated with our innovation pipeline, our next target will be to tackle those associated with our understanding of physics. There is much more we can integrate into this effort, so stay tuned for the next step.