APIs and AI. Challenges and Opportunities
David Roldán Martínez
Integrations Technology & Governance Strategic Advisor | APIs | AI | Smart Digital Ecosystems ?? Innovation Evangelist | Tech Writter ?? ??!???ds??d ????ou? ?o?? ??!|??? ?no? ??s no? d|?? ! '!? pu? s!d? ?u!sn
Introduction
The use of AI APIs is not new and has the aim of making super-smart apps that give customers a great experience. But, what happens the other way? How APIs can be benefited from the use of AI?
In this article, I'm giving you some highlights of both ways: AI APIs, which have enabled providing AI as a Service, and AI for APIs, a much more unexplored territory where AI power is about to revolutionize the whole API landscape.
Artificial Inteligence as a Service (AIaaS)
Simply put, Artificial Intelligence as a Service (AIaaS)?is the third-party offering of artificial intelligence (AI) outsourcing.?For a long time, Artificial Intelligence was cost-prohibitive to most companies and this is where AI-as-a-service comes in, as it allows individuals and companies to experiment with AI for various purposes without large initial investment and with lower risk. Having the opportunity to try the algorithms and services of different providers (Amazon Machine Learning, Microsoft Cognitive Services, and Google Cloud Machine Learning, for example) can allow?businesses to?find what works and allows for scaling before committing.
To name a few, of the benefits of applying with AIaaS:
Regarding to APIs, an immediate question arises: what if they expose AI through APIs? The advantages of APIs are well-known. In essence, APIs are a way for services to communicate with each other and allow devs to add a specific technology or service to the application they are building without writing the code from scratch.
Let's explore two types of APIs exposing AI services!
Cognitive APIs
AIaaS allows users to upload and stream data, run cutting-edge models, analyze data using a cloud platform, and consume via APIs. Standard options for APIs include Natural Language Processing (NLP), computer speech and computer vision, translation, or emotion detection.
The following list, far from being exhaustive, includes some of the best examples:
Machine Learning APIs
ML and AI frameworks?are tools that developers can use to build their own model that learns over time from existing company data without needing a big data environment. Thanks to AIaaS, companies can manage Machine Learning without having any particular technical expertise. One example of ML APIs is BigML.
As shown in the picture below, the most basic flow consists of using some local (or remote) training data to create a?source, then using the?source?to create a?dataset, later using the?dataset?to create a?model, and, finally, using the?model?and new input data to create a?prediction.
Artificial Intelligence for API
However, there is a more interesting (and we would say challenging) topic. How can AI help in managing your API ecosystem and your API Lifecycle? This is what we've called "AI for API".
领英推荐
Let us list a couple of “AI for API” examples!
AI In API Design
AI and?online deep learning?techniques can function with near-human cognitive accuracy and is much better with quantitative and numerical data analysis.?AI can analyze the data model you want to expose through APIs and, based on AI and ML, give you a strong head-start for each of the three pillars of a brilliant API design guide: operations (or methods), error handling and, data formats (request and response).
But it doesn’t end there. AI can also help with modeling and documentation, which are fundamental to developing successful APIs.?
AI for API Operations
APIs are the backbone of digital transformation. Via APIs, you can securely share data and functionality.?Any downtime or performance degradation can lead to a significant loss in revenue, customers, and brand value. Therefore, there’s mounting pressure on operations teams to ensure that APIs are always available and performing as expected. APIs are the cornerstone of customer experiences and if they go down, disaster comes.
For example, Apigee applies ML to API metadata to improve anomaly detection by:?
Also, other companies like Akita Software have products able to make a per-endpoint API monitoring by watching API traffic live and automatically inferring endpoint structure, which endpoints are slowing down the performance or which are causing errors, and solve these issues fast bypassing the complexity of implementing and maintaining monitoring and observability.
AI for API security
APIs are the open door to sensitive data and are vulnerable to different forms of cyberattacks.?Don't panic! AI is here to augment the average API security system by analyzing security threats and detect malicious cyberattacks such as Data Exfiltration, Advanced Persistent Threats (APT), Data Integrity, Memory Injection, DDoS API attacks, Login service DDoS, and so on.
The advantage of using AI for blocking security attacks is twofold:
AI also detects and prevents intrusion by conducting static security checks on all incoming data and access attempts. This process examines access patterns while simultaneously scanning payloads to validate all requests and scan them looking for harmful content in API interactions. As a result, AI-based static checks can identify and prevent gateway attacks.
But AI can also help with throttling to prevent high-volume cyberattacks. AI systems can dynamically inspect rapidly changing data so that AI can identify and pick up signals that may spell unauthorized access for intruders or monitor API calls and terminate interactions that may endanger system security.
References