Operations staff get a hard time. The lowly systems administrator (sysadmin), database administrator (DBA) and all the other operations engineering team members from cyber penetration specialists to user acceptance testing (UAT) and so on are generally unloved.
This reality is why we used to talk about software application developers (who, you may have noticed, have been getting a lot of the limelight in recent years) finishing their application builds and then just ‘throwing it over the wall’ to the operations team to figure out how to provision backend systems and be able to run it.
Things have gotten better and the DevOps developer (Dev) + operations (Ops) drive to form a more coalseced workforce ethic and culture has progressed things, but not always.
Ops champions emerge
Clearly, we need to keep evolving DevOps and at the same time (and this could be the good news part for operations staff) we also need to start creating directly applied Ops functions to specific parts of the modern IT stack.
Key among these functions will be Artificial Intelligence (AI).
Enterprise data platform company Tibco wants to make life easier for the AI operations team (now commonly written as AIOps) with its ModelOps release. This software service is designed to enable businesses to deploy AI models faster across a broader range of devices, machines and user endpoints at scale. Sometimes written as TIBCO to denote the organization’s lengthy acronym (The Information Bus COmpany), the firm is known for its data integration pedigree and its data analytics portfolio.
To put it in really direct terms, this is scalable secure cloud-based data analytic model management, monitoring and governance – now with an increased focus and function set aligned to AI model deployment.
Ingredients in an AI model
When we talk about an AI ‘model’ in this sense, the term is used to encapsulate the algorithmic logic that goes into the AI engine (or brain) and it also straddles the critical life-support systems that the AI will need in order to work without bias, without loss of insight into what it is doing and with transparency of an algorithm’s behaviour within business-critical applications
In this case, Tibco ModelOps addresses the requirement for speed in deploying AI and draws from the company’s work in data science, data visualisation and business intelligence (BI). The software itself works to get AI models to a state where they can be deployed and managed into ‘model pipelines’ (a digital journey that describes the lifeblood, location, lifecycle and lifespan of an AI model) so that they can be moved into production environments efficiently in robust ways.
The Tibco ModelOps solution is format-agnostic, supporting all common model formats, including Application Programming Interface (API)-based models in any cloud service or on-premises. This service promises to make it easy to add governed models to other Tibco products including Tibco Spotfire (technology for data visualization, discovery, wrangling and predictive analytics), Tibco Data Virtualization and Tibco Streaming.
Mark Palmer, Tibco senior vice president of engineering points to New Vantage Partners, 2021 big data and AI executive survey by Tom Davenport and Randy Bean.
“While 92% of firms spent more overall on data science in 2021 compared to previous years, only 12.1% deployed it at scale [according to the above-linked survey]. To help organizations realise the value of their AI deployments, we’ve designed a system that puts self-service access to data science firmly in the hands of teams, including business users,” said Palmer. “This allows decision-making teams to choose the algorithm they want, work from any cloud service, and run it safely, securely, and at scale.”
AI road trip, out of the lab
These product developments from Tibco push AI to a new point according to Palmer i.e. he says that this is the moment where we can witness bold new steps as we now enable business users to take AI out of the lab and out on the road.
A 2022 survey of Tibco customers suggested that it’s no longer uncommon for organizations to manage hundreds – even thousands – of analytic models and workflows. Tibco ModelOps claims to be able to allow any authorized business user, data scientist, analyst, or IT user to manage and deploy thousands of models in production with complete governance and management capabilities.
Users are able to deploy in the cloud or on-premises, highlighting model performance through built-in, customizable dashboards powered by Tibco Spotfire. This means that Tibco ModelOps, clients can now move past the worry of unintended negative consequences of failed automation because of complex or poorly managed AI or rules-based models, making it safer to automate based on validated and secure AI models.
Ops on the up
There’s a bit of the digital democratization transformation message here that we’re hearing from every enterprise technology platform worth its salt. There’s also a bit of the robust scalability message… and there’s obviously the all-important managed intelligence as a means of mitigating AI bias that we absolutely positively need to hear from every company in this space.
But that’s not to say that Tibco isn’t coming forward with a genuinely new developmemt.
The company is (arguably) one of the first vendors many people think of when the subject turns to data integration, so there’s no reason why we shouldn’t now extend that thought to AI model data integration, governance, refinement, augmentation, management and visualization.
Ops is on the up and AI is helping to share the love, let’s never be haters.