Integration of AI services. 5 steps to identify and implement integration with AIaaS
In this talk, I describe 5 steps to integrate AiaaS in your projects.
Here I'm considering architecture implementations, automatic tests, business rules, cost, good practices and the main possibilities on Google, OpenAi and Microsoft.
artificial neurons do not work as human neurons. This is just a comparison. input processing output processing input output The first concept considers just the electricity model.
it become a topic? AiaaS OpenAi demystified AI recently with ChatGPT In the 1940s, when Walter Pitts and Warren McCulloch developed the concept of artificial neurons. And in 1950, when Turing published a paper called "Computing Machinery and Intelligence." Revista Exame. ENGENHEIRA.COELHO
tools through of cloud computing services. Google, IBM, Intel, Microsoft, and Apple Inc. are some of the major players in the global AIaaS market. Research Markets - Global Opportunity Analysis and Industry Forecast, 2017-2025
only focus on scenarios that don't have existing solutions, but also those that do. By looking for optimization opportunities and service quality improvements, we can maximize the value of AI in our solutions. STEP 1 Understand the problem ENGENHEIRA.COELHO
product or UX team, as a software engineer you need to understand how each part works and communicates in the system. STEP 1 Understand the problem ENGENHEIRA.COELHO
improvements. Talk to customers and understand their main difficulties and pain points. Identify any recurring issues or errors. Use data analytics to identify where users tend to abandon the process or web page. ENGENHEIRA.COELHO
to be sent to the AI services and define the expected format of the response data. Additionally, determine how the response data will be used in your project, such as for decision making or generating insights. STEP 2 Understand the data
(text, image, sound) Graph Service availability Possibility of use cases Investment Technical compatibility Here we defines the best platform or service, considering:
a filter for inappropriate content. /fine-tunes - Allows you to train an existing model. /files - import files and use them in training with end-point fine- tuning. /audio - Used to transform audio into text. /embeddings - Creates numeric representations for words, sentences, and documents. /images - Processes images, including capture, generation, background removal, and adaptations. /completions - Sequential conversation
and detects feelings. Speech-to-Text: Automatic recognition, real-time transcription and audio call enhancement. Text-to-Speech: Convert text to speech. Voice user interface on devices and applications. Carry out personalized communication. Translate: Real-time translation that can be used to internationalize products. AI Vision (image): Vision API to detect objects, understand texts, celebrities, brands, faces, writing and others. AI Vision (video): Extract rich metadata at the video, scene, or frame level.
openAi services Search: Service that an AI-powered information retrieval platform. Vision: Identifies and analyzes images, making it possible to train the model Speech: Audio to text, text to audio, speech transcription and speaker recognition. Language: Text analysis for mood and sentiment. Bots: Robots that can assist in customer service tasks. Metrics advisor: Metrics extraction monitoring and problem diagnosis.
using automatic tests is not an easy task, data is always changing, and providing different response possibilities. In this case, a good way to test whether actions are performed correctly is to consider just immutable data.
is to use a new request to AIaaS, which will guarantee that the response has the same type of text requested previously. Conducting this type of testing can be risky if you are using an AI service that charges per request. Therefore, it is important to use this test only when it is really necessary.
text model, the model receives and responds in natural language. To get your response in a specific model, you need to explicitly request it. As in this entry, which specifically determines the correct answer: const textType = "sci-fy" const textRecommendation = "A blue house with birds" const orientations = "Do not do comments about your response, do not write any different orientation, just return a " + textType + "text considering the recommendation. The recommendation are:" // Considering we have the fetch encapsuled inside a method const response = fetchPost("/completions", "text-davinci-003", orientations + textRecommendation)
will happen in the cloud, using enterprise SaaS platforms. AIaas now looks set to become part of standard operating procedure CISCO ANNUAL INTERNET REPORT (2018–2023) WHITE PAPER
satoshi7
keeg
terapyon
riyaamemiya
ishikawa_satoru
syuchimu
occhi
willnet
kworkdev
kentarofujii
bengo4com
yamanoku
jonoalderson
pwiseman
jakevdp
palkan
aarron
lara
ryanjones
cassininazir
nmsamuel
badams
eileencodes
![@ENGENHEIRA.COELHO Thanks! [email protected]](https://files.speakerdeck.com/presentations/f63e1c1b31174443a7474a2c426add6f/slide_43.jpg){kind=link}