Part 2 - How are we doing this?
In our previous article, we mentioned how we developed our idea for Cerebrium and the problems we want to solve.
We want to make data and machine learning more accessible to SMBs, to influence everyday decision making and change the way they operate!
Before I dive into how we are approaching some fundamental issues, I would like to pose 3 questions that have influenced our approach:
- How unique is the data your company monitors when compared to competitors? Are you tracking revenue, return on money spent on Google Ads and the number of nights booked, if for example, you are Airbnb or Vrbo.
- Are many SMBs today built on top of their own custom software or a combination of SaaS tools and low-code/no-code tools that fulfill different business functions?
- Do existing platforms that abstract complexity replace the need for certain individuals entirely? Has AWS replaced the need for DevOps engineers?
There are no straightforward answers, however, below are responses based on our years of experience in the industry and dealing with clients from a diverse set of industries:
- The answer depends on the industry and the maturity of the business. From our findings, it seems businesses in similar industries and maturity track between 60-85% of the same data. If businesses have this much overlap in the data points they are tracking, why are they spending hundreds of thousands of dollars and significant time to get their data infrastructure up and running? There is an argument for developing a data infrastructure from scratch, however, it is a decision that may cost your business valuable time and money without seeing a significant benefit.
- The latter. Low-code/no-code tools are growing at a 45% CAGR (compound annual growth rate) and on average there are 30-50 SaaS tools in a business with 50 employees. There are existing SaaS tools covering particular business functions and most of them adequately cater for enterprise use cases. We predict that custom development will only be prevalent for technology companies, particular use cases or in large enterprise businesses. Low-code/no-code tools are empowering workforces to be creative in their domain without the need for developers, accelerating the pace of growth.
- No. AWS reduced the time to spin up a server and the need to have multiple engineers set up and maintain server infrastructure, however, businesses still require a CTO or engineer to set this up on AWS. This has created experts in the DevOps space and communities around its product while removing complexity and giving smaller business access to infrastructure that wasn't previously available to them. At Cerebrium, we don’t plan on replacing data scientists or data engineers, we hope to empower them!
Our Solution
At Cerebrium, we are experimenting with open schema standards based on industry, business models and operational models that will allow businesses to map their data sources to predefined schemas that have been designed by specialists or experienced individuals in industry. This will radically reduce data engineering time, allow businesses to experiment with data structures and help businesses optimize data structures rather than designing them from the ground up. With this approach we hope to cover 60-85% of a business's data needs and allow them to focus on the 15%-40% that is unique to their business.
In the near future we wish to make these schemas open source to cover all businesses types in order to make sure these schemas are designed with large scale community input from those familiar with the industry and those with experience in the industry while leaving room for customization by businesses.
Our goal is not a “one size fits all” data schema but will allow companies to capitalize on similarities between businesses in their industry and the SaaS tools they use. We believe that approaching data infrastructure this way will achieve a multitude of benefits such as:
- Quicker time to value
- Reduction in resources and costs
- Experimentation of data schemas
- Optimal schema design based on community input
With this approach, even the smallest business will have access to data infrastructure previously only available to large enterprises giving them the ability to become data orientated earlier in their journey. A significant part of the success of a business is based on execution and so we hope by fulfilling our vision of making data more accessible to SMBs that more businesses will be able to execute their strategy more efficiently.
Stay tuned for our third and final post describing our approach to unlocking open schema standards in terms of machine learning for SMBs.
