Usable AIOps

Both AI and AIOps are terms that have experienced inflated expectations and market disappointment. Part of the problem is these terms cover a broad range of technologies from insight engines to AI general intelligence, as illustrated by the Gartner 2020 Hype Cycle. What any specific vendor may mean by these terms varies, and these technologies will mature at different points in time. The other challenge is taking technology created for developers and delivering it as operations usable solutions.

Selector’s company vision is IT solutions that are usable: productive and compelling across all aspects of the solution. Productive is the ability to complete a desired task quickly and efficiently. Compelling refers to significant impact. Focusing on these qualities delivers the promise of AI/AIOps and crosses the chasm from early adopters to mainstream adoption. This blog explores the ways Selector’s solution, Selector AI, enables compelling productivity through operations analytics, and the spectrum of AI/ML/Data science technologies utilized.

Correlation — Automatically Connecting the Dots

The history of IT tools is one of each vendor providing its own alert monitoring dashboard. This has resulted in two problems. A proliferation of monitors/dashboards, and siloed data. Some solutions have put significant work into consolidating alerts from different sources, but have provided limited value-add due to a large number of hierarchical dashboards that are difficult to navigate, and static in nature.

Many operations tools focus on one specific type of data, or are just monitors not doing significant analysis at all. The Selector vision is that IT teams should NOT be running between different monitors to try and correlate different lenses on an issue, or slowly navigate complex dashboard hierarchies. Instead, data from different sources should be correlated to quickly and efficiently focus IT experts in the optimal area for their experience driven deep-dive analysis.

Selector AI connects the dots across a growing number of data sources: telemetry, logs, configuration changes, synthetic testing, meta data, and inventory. This is not just collecting and displaying. This is analyzing to connect the dots, dramatically reducing the amount of time operations teams take in deciding where to focus their experience, skills, and energies.

Automation — Eliminating Solution Administration

Automation of core IT assets is a significant trend. Little will be gained in automating the core assets, if operations tools require significant additional manual overhead. Selector AI is automated by design.

One example is alarm threshold setting. Instead of IT teams having to set thousands of thresholds manually, or relying on inaccurate heuristics, Selector AI auto baselines and dynamically sets thresholds using self-supervised machine learning. Another example is automated synthetic testing. Most importantly, the entire analytics workflow is automated. Data is automatically ingested, analyzed, and available for visualization, query, northbound interfaces, and automation playbooks.

While Selector AI does provide the ability for data science experts to adjust workflows, Selector AI’s default approach makes a spectrum of data science processes and technologies usable by mainstream IT adopters.

Dynamic ETL — Extensible Data Sources

Many data science pipelines begin by transforming different data types and structures to a common structure that is optimized for query and analysis. This simplifies the analytics logic by eliminating the need for analytics logic to know the details of different data sources — different data sources can be added without changing the analytics code. Selector’s focus on customer productivity led to a YAML-based approach to defining data schemas. As a result, the Selector solution not only comes with pre-integrated data sources, it is quick and simple to add new data sources without changing, recompiling, or reinstalling the solution. Extensible without impacting the availability of the solution. Productive and compelling.

Query Driven Visualization — Eliminating Complex Dashboard Hierarchies

Solution suppliers invest significantly in anticipating what dashboards operations teams will need. These efforts are well-intended, but ultimately, only operations teams know what they will need for any given anomaly. That is why Selector has enabled users to dynamically create, as needed, visualizations and dashboards from queries honed to the specific anomaly they are dealing with. This gives users the information they need, without having to navigate complex hierarchies.

Natural Language Queries — Human Usable Query

Natural language processing is an important aspect of AI that has experienced significant progress over the last decade. Selector believes this is the best way forward for query interfaces, replacing complex, static, and fragile command line interface code. Instead of asking users to adapt their way of thinking to unnatural computer query languages, Selector believes query languages should adapt to the way humans already think. Not only does this hide the complexity of the underlying data store, it makes powerful query capabilities usable, with a significant reduction in or elimination of training.

Immersive Collaboration — Powerful Capabilities Where Teams Already Work

While there is a role for expert portals, solutions that only provide yet another additional expert portal are continually taking teams out of the environment they already work in, collaboration tools, and increasing the difficulty of sharing operations analytics. Both of these complicate the operational environment and increase inefficiency.

Some solutions enable alerts to be sent to a collaboration environment. Selector AI goes far beyond that, enabling complex queries to be executed within collaboration environments, including the dynamic generation of visualizations. The generated content is easily, efficiently, and naturally shared with other team members. As collaboration tools already operate on many different screens: laptops, monitors, and smartphones, so too do Selector AI capabilities. Selector AI operations analytics can be dynamically generated and shared from any screen, anywhere, anytime — increasingly important in a world of work from home and work from anywhere. The Selector AI operations analytics is productive and compelling — usable in the environment that teams are increasingly turning to.

AI/ML Observability — Seeing the Unseen

Selector believes that there is a new generation of data science-based approaches that are going to have an increasing impact on the productivity and effectiveness of operations teams. Not because they are going to replace operations teams, but because they allow these teams to see that which would not have otherwise been seen; to observe the hidden connections, patterns, and trends.

Humans can not process the amount of operations data available in a timely manner, and arguably not at all. Selector AI’s range of data types, filtering, correlation, ranking, and knowledge graph, connects the dots across diverse and previously siloed data sources in a way that has not been available previously.

What are the normal characteristics of resources? Selector AI’s self-supervised machine learning provides automated answers to this question, without weeks of model training, in a way not deployed by operations teams today.

Natural language query simplifies the process of interacting with a knowledge graph that organizes data for quick answers without walking a long and time consuming hierarchy; a knowledge graph that connects the dots across many different types of data.

Today, there is great debate about what AI/ML means and whether it is real or marketing snake oil. Definitions range from insight to general intelligence. Selector can say with confidence that new computer / data science approaches are enabling operations teams to see what they previously could not see, allowing them to efficiently focus their experience and skills, so time to restore service is dramatically reduced.


Easy and efficient data ingestion. Automated analytics. Human usable query and visualization. Immersive collaboration. All these approaches are designed to broaden the usability of the Selector AI solution, delivering improvement in productivity, with compelling actionable insights that enable IT teams to rapidly work around and resolve anomalies as they arise.

Selector AI — The most Usable AIOps solution.