Click to learn more about author Prasanna Kumar Illa.
We all are aware that AI (Artificial Intelligence) plays an increasingly significant role in automating highly repetitive tasks, at scale. By repeatedly analyzing the data, AI adapts progressive learning techniques and discovers new crucial information, which was unnoticed before. It includes new capabilities to existing platforms greatly enriching the user experience. In any organization, between applications, most of the data flows through the integration platform and this is one such area where AI adds incredible value in enhancing enterprise’s application and Data Integrations. This article shows how AI and ML techniques are used in Enterprise Integrations and highlights the research by various Cloud Integration platform providers.
AI and Machine Learning (ML) techniques when applied at different stages in an integration development cycle results in different benefits. ML algorithms applied at execution time leads to self-healing integrations. It helps in identifying Data Quality issues and suggesting data corrections. It auto-detects sensitive data in integration flows and offers data masking recommendations, which is of immense value in conforming to regulatory and security compliances such as GDPR, HIPAA, PCI-DSS etc. ML utilized in identifying pattern recognitions and anomaly detections on streaming data, helps in Predictive Analytics. Self-healing techniques adopted at platform level helps to auto-scale based on capacity spikes – minimizing service interruptions and avoiding outages. Likewise, ML techniques applied during design time leads to self-defining integrations. Currently it helps in performing auto-matching and applying auto-type conversions when designing mappings and transformations.
Incorporating AI in integrations significantly boosts the organization’s productivity and speed. AI driven chatbots based on conversational and voice commands self-integrates to pre-defined set of applications. However, this area is currently in research.
Informatica recently showcased AI driven CLAIRE platform at Informatica World May 2018 that builds self-defining integration flows on its own and automatically connects different systems based on voice commands.
Organizations can quickly leverage AI-enabled Cloud APIs to augment superior functionalities to their existing applications. These APIs transform their business systems to cognitive applications that self-learn and enrich user experience. The Cloud giants such as Amazon, Google, IBM and Microsoft offer subscription-based AI-APIs with diverse functionalities such as facial recognition, Natural Language Processing (NLP), interactive conversations, sentiment analysis etc. Let’s look at AI APIs offered by the Cloud giants:
- Amazon AI Services such as Lex, Polly and Rekognition provide advanced Deep Learning and NLP functionalities to enable conversational interfaces and chatbots. Using these APIs, applications can also leverage image analysis such as object detection, facial analysis and facial recognition.
- Google Cloud ML Services enable conversational interfaces to recognize the user’s intent and context. It’s REST API enables to detect occurrence of key nouns in a video file and classifies images into different categories.
- IBM Watson Services offer APIs for visual recognition, language interpretation, language translation, conversational chatbots etc.
- Microsoft Azure AI offers Cognitive Services for vision, speech, language, knowledge and search that can be used as-is or customized with few lines of code.
Even small cloud vendors such as Wit.ai, api.ai, Diffbot etc. offer AI-driven APIs to enable advanced user experience in traditional applications.
AI-driven “Self-Service” integrations
Cloud Integration Service providers, referred as iPaaS (integration Platform-as-a-Service) providers, have simplified the perplexing job of application connectivity and made effortless integrations possible. They introduced “self-service” capabilities in the platform, so everyone in the organization can build their own integrations. iPaaS enabled browser based development and drag-and-drop features with intuitive user experience. Now, any user such as business user, analyst or project manager with clear requirements can experiment or build integrations using self-service features. This significantly improves speed and agility in the organization as any member can take part in the development. It greatly reduces the queue for IT resources and allow the IT teams to focus on more sophisticated integration requirements.
The integration providers further enhance the “self-service” capabilities by introducing AI. As these platforms are Cloud-based and widely deployed, they analyze the shared metadata and data usage patterns across multiple tenants to build reliable recommendation models. As it still involves laborious manual steps and conformance to standards during development, there is room for automation. The recommendation engines in the integration platform guides the business users and offer suggestions to develop the integrations more intuitively.
For example, SnapLogic claims that its Iris AI learns from metadata and data flows of its Enterprise Integration Cloud. It’s Integration Assistant utilizes recommendation engine and machine learning algorithms to provide step-by-step guidance for building data pipelines.
Informatica, another iPaaS provider, claims that its AI driven CLAIRE engine leverages the Cloud metadata capabilities to automate Data Management and Data Governance processes. It uses Machine Learning to identify data similarities, derive entity discoveries and offer data recommendations to accelerate developer productivity.
Likewise, Dell Boomi Suggest offers mapping suggestions with confidence rankings, data transformations and error resolutions through correlations for effortless development.
These “self-service” capabilities help non-technical users to utilize AI in the background to improve their productivity. As the recommendation engines generally gets more efficient on usage it leads to the next phase, building autonomous integrations.
AI-driven integration platforms automate repeated low-level development activities of same or similar integration patterns. They further simplify the development of integrations and offer step-by-step guidance requiring minimal technical expertise. Based on the recommendation models enhanced by collective multi-tenant deployments, autonomous integrations are generated based on conversational commands. Just as using self-driving cars, non-technical users can use integration or application platforms to build self-driving integrations and avoid manually coding the integration flows. Currently, this is at an early research stage and following capabilities will be launched by integration platform providers.
SnapLogic claims its self-driving software for Enterprise Integration enables business users to manage their own data pipelines.
Workato claims it interactive, intelligent chatbots enables users to complete tasks and workflows directly from chat.
In conclusion, organizations are continuously exploring ways for automating repeated functions to gain new market opportunities and increase efficiencies. To meet these demands, Cloud Service Providers are researching to enhance their integration platform, where most of the organization’s data moves, to self-define and self-heal. Here, AI & ML plays an important role in enterprise integrations to automate the tasks with little human involvement. And how little?