Digital transformation -> AI transformation
Uber is the world’s top taxi company, but doesn’t own taxis. Airbnb is the world’s top hotel company, but doesn’t own hotels. Before digital transformation, companies used software. Now, companies are software, as demonstrated by Uber and Airbnb. So, what’s next?
As rapidly as companies became software during digital transformation, they will become hyperconnected, AI-transformed software even more rapidly. The hyperconnected shift has already started. For example, businesses connect to Stripe’s APIs to process over $500 billion of online payments per year. Now, neural network based large language models (LLMs) are multiplying the value of integrating APIs and (connected) data from many sources, while also multiplying the speed at which the integrations can be designed, implemented and iterated, which then multiplies the value of the integrations. Yes, a potential flywheel. But the AI transformation flywheel can be impeded, like any other flywheel.
AI transformation flywheel obstacles
Like digital transformation, AI transformation is motivated by goals of software-enabled innovation and agility. AI transformation also has similar obstacles as digital transformation: the need for speed can conflict with needs for reliability, resiliency and security. Speed functions as a stressor on the skills, processes and tools which were designed to achieve reliability in a different world. For the AI transformation flywheel to turn, we need to find a way to avoid a tug-of-war between speed and reliability.
To help optimize trade-offs between speed and reliability during digital transformation, and prevent tug-of-wars which neither side can really win, we “shifted left”. The shift left model moved traditional “day two” practices like ops and security (which are critical for reliability and quality) into the design and development phases, ensuring that ops and security needs were accounted for upfront. Similar to Benjamin Franklin’s saying of “an ounce of prevention is worth a pound of cure”, upfront design decisions are often more effective and efficient. Can we expand on this shift left model to enable AI transformation to overcome a similar set of obstacles as shift left did for digital transformation?
Shift left to enable hyperconnected, AI transformed progress?
The good news is most of the shift left paradigm extends into an AI-transformed, hyperconnected world. The bad news is the hyperconnected part is mainly sealed off. We have software for compute, app dev, app security, orchestration and ops. This means dev and ops teams can interact with that software via code, and therefore can shift left. But the secure networking parts required for hyperconnected are a string of islands which are configured and infrastructure dependent, and/or consumed as black box SaaS. These islands include WAN, firewall, proxy, LB, VPN, IDS, IPS, ZTNA. To make it even worse, these islands often depend on more non-software constructs such as IP addresses, ports, NAT, certificates, encryption, access control lists (ACLs) and DNS. Proactively integrating these non-software islands into the software process is very difficult, and hence it is an obstacle for our attempts at extending the shift left paradigm to help with the hyperconnected parts of an AI-transformed world. However, we don’t need to solve the shift left problem on the existing playing field – we can change the playing field to better match the needs of a hyperconnected, AI-transformed world. In fact, we have seen this movie before.
A platform to transform secure networking into software, enabling AI transformations
Prior to the cloud paradigm, compute infrastructure was anything but software. In fact, it was an obstacle to speed and innovation. Cloud software and platforms of course changed that playing field, giving us the models and abstractions necessary to control compute with code. Now it is time to change the playing field for secure networking.
The Ziti platform turns secure networking into software in a similar manner as cloud platforms turned infrastructure into software. Even better, the Ziti platform turns secure networking into open source software. Here too, Ziti follows the footsteps of giants. Linux, Docker, Kubernetes, Mongo, Elastic and Kafka are well known examples. Every major cloud platform is built on open source. In fact, secure networking was the only major infrastructure category which didn’t have an open source platform until Ziti emerged.
AI transformation requires network transformation
Digital transformation was quite a shift. AI transformation will be even faster, partially because it builds on digital transformation. During digital transformation we used platforms to transition to virtualized environments, cloud native apps and hybrid architectures. AI transformation requires network transformation, as the result is an AI-accelerated, hyperconnected world. Secure networking is at the foundation of that world and it requires platforms such as the ones we developed during digital transformation – platforms which are purpose built for its needs. The Ziti platform is the first secure networking platform to step up to this challenge. Try OpenZiti (open source, self-hosted) or CloudZiti (SaaS, network fabrics hosted by NetFoundry and partners) today to simplify your journey.