Custom AI software integrations and development with Ruby on Rails.

Why Do You Need AI Integrations in Your Rails Application?

Truthfully - you may not. Using an AI integration just for the sake of it may not be the best method. An AI integration is going to be a viable solution where the process or target process is overly complicated, or if no other tools available have been able to get it right. Beyond that there are practical applications that we are just starting to see the benefits of - and for these it’s a matter of using your imagination, and talking to someone who is willing to think outside the box, but can also stay grounded in the technical limitations.

Improve Functionality and Performance

If you think that an AI integration can help to improve performance and functionality of your app then it is probably a good investment. Both of these items are likely to have a return in terms of better user experience, and hence a better return on investment.

Using AI to replace tedious functions that were taking manual intervention is a good use (see this example here of an email parser), or something that is going to speed up response time on your application.

Another quick win might be an FAQ bot that has cached all of the site data, and can respond to natural language questions about the company or its products just using the site data as a reference point.

New Insights and Capabilities

Using AI to surface data related to your brand or products may allow users and customers to gain new insights by getting answers to the questions they have - each person will have a specific and unique question to them - rather than moving through pages of a website or a wiki, you could provide the exact tailored information to them - potentially leading to more sales/signups/members etc.

AI functionality could even move a person through a sign up or web form faster by understanding the right fields to present at the right time - not wasting time gathering information that is already known or gleaned from elsewhere. of course there are many more capabilities that are limited only by what you think of!

Automate Repetitive Tasks

One of the things computers are really good at is doing simple tasks over and over again. The rule is that if you can work out the steps in a system that are done exactly the same time after time, then you have a good target for automation. With the entry of AI into the sphere we can now look at use cases where the process is almost the same time after time, but there are small changes or edge cases that otherwise would prevent automation. Program the process in the usual way, but have the edge cases caught in such a way that the data or step is normalized and can then process without throwing an error, stopping the flow, or requiring human intervention.

Optimise Productivity

As with any software, the end goal should be to optimise, increase or bring about productivity. If the software is not advancing the purpose of the organization in which it is used, then it is, or course, useless. To this end any Ai integration should be done with the viewpoint of ‘do I really need this?’. We are not going to add AI to a piece of software just for the sake of it - it will need to be serving a purpose, and that purpose will need to be greater than the time and money spent to create it (or at least there is a reasonable expectation that this is the case).

The reinteractive AI Integration Process

Identify the Problem

What is the exact problem you are trying to solve, and what are the best tools available to us to solve that problem. That might not be AI - validating and justifying an AI solution will mean that to some degree you will have to buy in to even a small degree of R&D. The estimate of work may not be a clear cut x - x+1 number of weeks. That doesn’t mean that AI solutions are not possible, or even a good fit for many problems, just be aware of the risk, even if slight.

The other point - already touched on in this article - is, does the ‘problem’ even need to be solved at all. By clearly understanding the idea, and working out a technical solution, you will still have to make a business case for that work to be done.

Research and Scope

Once identified, the next step is to design a path for the solution that fits within budget, and importantly within the current business rules so that current systems and apps will still fit in (unless the solution includes a complete overhaul). This scope can then be, once approved, executed by a skilled developer with some understanding and knowledge of the AI tools available. Testing during work and once complete is expected to be thorough, both on our side and for the client, before the completed work is launched public. AI can sometimes give unexpected results, and every safeguard should be taken to prevent or catch this. For example, an AI bot trained to parse emails and make financial offers based on the information parsed should always contain some disclaimer, such as ‘based on our understanding that the offer is…’ - that way an error can be backed off legitimately where a mistake has been made.

Continue to Monitor

As you find new and interesting edge cases for your AI application or feature continue to test and have the code adjusted if the result is not what you would expect. Also be aware that this is a fast moving field, and so any code implementing or leveraging AI API’s should be written in such a way that it can be easily upgraded, or even rewritten without adversely affecting the whole application. This means that you can move forward with advances in the field, improving the speed and reliability of your application. Any application written with Ruby on Rails, as we do in reinteractive, will have this ability pre-baked in to the application. One of the advantages of writing in Rails is its ability to easily break down to the various components, and allow developers who have never touched your code before to easily and quickly get up and running.

The reinteractive AI Integration Process

Identify the Problem

What is the exact problem you are trying to solve, and what are the best tools available to us to solve that problem. That might not be AI - validating and justifying an AI solution will mean that to some degree you will have to buy in to even a small degree of R&D. The estimate of work may not be a clear cut x - x+1 number of weeks. That doesn’t mean that AI solutions are not possible, or even a good fit for many problems, just be aware of the risk, even if slight.

The other point - already touched on in this article - is, does the ‘problem’ even need to be solved at all. By clearly understanding the idea, and working out a technical solution, you will still have to make a business case for that work to be done.

Research and Scope

Once identified, the next step is to design a path for the solution that fits within budget, and importantly within the current business rules so that current systems and apps will still fit in (unless the solution includes a complete overhaul). This scope can then be, once approved, executed by a skilled developer with some understanding and knowledge of the AI tools available. Testing during work and once complete is expected to be thorough, both on our side and for the client, before the completed work is launched public. AI can sometimes give unexpected results, and every safeguard should be taken to prevent or catch this. For example, an AI bot trained to parse emails and make financial offers based on the information parsed should always contain some disclaimer, such as ‘based on our understanding that the offer is…’ - that way an error can be backed off legitimately where a mistake has been made.

Continue to Monitor

As you find new and interesting edge cases for your AI application or feature continue to test and have the code adjusted if the result is not what you would expect. Also be aware that this is a fast moving field, and so any code implementing or leveraging AI API’s should be written in such a way that it can be easily upgraded, or even rewritten without adversely affecting the whole application. This means that you can move forward with advances in the field, improving the speed and reliability of your application. Any application written with Ruby on Rails, as we do in reinteractive, will have this ability pre-baked in to the application. One of the advantages of writing in Rails is its ability to easily break down to the various components, and allow developers who have never touched your code before to easily and quickly get up and running.

Why Choose reinteractive to Help You With Your AI Integrations?

Our CEO, Kane Hooper, has been studying the field of Machine Language for several years, and as it has morphed into the field we know of today as AI, has continued to study it, writing practical applications along the way to demonstrate its real-world use (not just side-show tricks). In doing so he has imbued the whole reinteractive team with an appreciation for not just the uses, but also the limitations of incorporating AI into our workflow. In doing so we have been able to develop production ready applications for clients - using the rules we discussed above - to provide actual value and purpose to those clients.

We are standing ready to discuss how an AI integration using Ruby on Rails may help your orgainsation.