So, you decided to build a chatbot and are wondering how much it will cost you? I want to say right away that there is no single figure. Imagine you are buying a car. The price depends on the brand, class, and all the extras. The same thing happens with the bot.
To begin with, you have to decide why you need the bot at all. After that, choose the option that meets your expectations and has the best price-quality ratio. So, let’s not rush and explore the topic step by step.
The easiest way to decide on bot’s nature is to answer a few simple questions. Properly defined goals help build a viable development strategy and remove the waste of time and money. At Softensy, we supply our clients with the following checklist:
- How big is your company? Large, medium, small.
- What is the target audience of the bot? End users (B2C communication), staff members (in-house duties), business partners (B2B activities).
- Which tasks do you want the bot to do? Simple repetitive, complex analytical.
- How many users at one time should your bot serve?
- What messengers do you want to use?
- How many languages should the bot speak?
How to develop an enterprise chat bot
Now, when you know the purpose of bot creation, it is time to choose the path of development. Currently, there are a lot of tools that provide ready or almost ready solutions for every taste and budget. Some of them require no programming skills; others expect you to write some code. Besides, you can make a bot from scratch. Choosing this option, you get a unique bot with precisely the features you need. Right now, let’s take a closer look at three ways to develop a chatbot.
Ready to use solutions
Everybody is familiar with the Play Market or App Store. We go there to choose apps for business, education, sport, fun, and other purposes. If you need an AI expert, you can also go to a specialized store. Instead of mobile apps, it will offer you off-the-shelf bots.
Today, Telegram, Kik, and Slack have a large variety of bots. The categories range from Finance and Sales to Dating and Fun. Chatting with the bot is usually as easy as ABC, but the process of adding is different from chat to chat.
Apart from specific shops, you can use bot aggregators, where you can search bots by particular parameters. For example, ChatBottle is a real mine of artificial intelligence. On their site, enter a key phrase, like customer support or data analytics, choose your messenger, and get the list of suitable robots. You will see that some bots are tied to a single messenger while others work on many platforms.
Before we move on, let’s review the pros and cons of box solutions.
- There is a high probability of finding the right bot
- You spend a minimum of time for integration
- You don’t invest in the chatbot development
- You will use a standard third-party thing
- There is no way to customize the bot
Unlike bot shops, the platforms do not offer canned solutions but give all the proper tools to create a chatbot. Typically, you construct it from pre-built templates. Some services allow doing it without programming knowledge; others require coding.
The bot creation algorithm is almost the same across all platforms. For a start, you need to choose a use case – a typical task to solve. For example, automate FAQ or qualify leads. Next, you should follow the tutorial for building the bot of choice. In the end, run testing to ensure the bot acts the way you expect.
The majority of bot services offer free trials so that you can run the bot in real situations and see whether it works for you. Also, the platforms have free standing plans. However, their functions are restricted compared to paid subscriptions. For example, Chatfuel, a popular bot building tool, supplies 100 thousand messages a month for free. So, if this number covers the demands of your business, you can operate the robot without any payment.
In the previous post, we considered the top platforms for building a chatbot. Hence, I will skip this topic right now.
To sum up this part, here are the pros and cons of templated constructors.
- It is easy to build a bot from the ready blocks
- In most cases, you don’t need to hire a developer
- You can opt for cheap or free plans
- You can find a proper bot for any kind of business of any size
- If you run a big company, the free option will hardly cover your needs, and the paid ones may be quite expensive
- There is a small chance to find a bot that will 100% meet your demands
Now, let’s review in detail the bot development phases and estimate how much it may cost you.
Here you will have business logic, services for processing messages, and a conversational intelligence system. It is possible to create a server-side app using various technologies. The choice mostly depends on the expertise and likings of developers. At Softensy, we opted for Ruby on past similar projects. Our bot core is written in Node-RED. It runs on Amazon Web Services, which helps us ensure the stability and responsiveness of the bot.
The time for server setup and deployment is around 4 hours.
Here you need to integrate the channels through which your robot will interact with users. By channels, I mean instant messengers, such as Kik, Viber, Telegram, Slack, and Facebook messenger or any other. You can select as many messengers as you like. The integration differs depending on your choice, yet it has a similar algorithm. You need to set an endpoint to accept and post messages based on security tokens authorization.
The integration of each channel can take from 8 to 16 hours.
Chat in one place
If you want to manage all communication channels in one place, we advise on the Chatinoneplace platform. It is our own solution – a module that brings all messengers together.
Assume you have three messengers where your bots communicate with clients. If the bot faces a tricky question, it passes it to a human operator. In their account, the operator sees the client’s history: what messenger they use, what questions they ask, what problems they face. The human operator can read the chat records with bots and quickly detect the client’s needs. Thus, you can provide quality service by mixing bot and human support. You can hire a part-time human manager and reduce the salary budget.
The time for implementation may vary from 16 to 40 hours for all messengers.
Natural Language Understanding
NLP is a technology that helps interpret human speech correctly. You don’t need to invent it yourself. As a rule, we use third-party speech recognition APIs, such as Java voice recognition API, Python voice recognition tools, Google speech recognition API, and others. Setting them up is not a big deal. The main challenge is to teach a bot to recognize intentions and pick up key elements of dialogue.
The thing is the external tools only provide an interface for data input. However, you should define the entities and their values by yourself and afterward enter data into the system. In practice, you have to study the entities mapped to actual objects such as Stock, Contacts, Staff members, etc.
The NLP setup takes about 8 hours, whereas training and extraction of data may take up to 80 hours.
Understanding human speech is only half the way. The hardest part is teaching a robot to speak naturally, sticking to the point. To do this, you need to create conversation algorithms, chat workflows, decision trees. Besides, you have to provide users with easy navigation so that they start the dialogue right away.
Building bot conversational intelligence may take up to 100 hours in total.
The bots for business aim to solve particular tasks, such as order items, book tickets, compare prices, and make appointments. Based on the goals of your company, you need to connect with the proper services. For example, if you automate customer support, you need to provide the bot with access to a knowledge base. If you want to free the employees from some routine tasks, let the bot enter ticket and hotel reservation platforms.
The integration with internal and external services may take up to 80 hours.
The control panel provides tools to analyze bot activities. There, you can view the chat records, the history of transactions, error logs, and other bot related activities. You can build in some analytics that will show the operating efficiency of AI experts and help in making strategic decisions.
The approximate time for implementation – 30 hours.
A detailed estimate of chatbot pricing
|Communication lines (messengers)||16 hours|
|Chat in one place||40 hours|
|Natural Language Understanding||80 hours|
|Smart conversation||100 hours|
|Integrations with third-party services||80 hours|
|Control panel||30 hours.|
So, how much does it cost to build a chatbot?
Well, it’s quite simple arithmetic. If we sum up the hours listed above, we get 560 hours. Considering that an average salary of the software developer is $20 per hour and that you need a team of at least three members, the total chatbot pricing is 350 * 20 * 3 = $21000
This figure may sound crazy-expensive, but you should remember that it is a one-time investment. Developing the bot from scratch, you can lay down the rule and get full power over the product. Nevertheless, if you are a small start-up or mid-sized business, this path may not work for you. In this case, you can settle on a bot-building platform and enjoy all the benefits of ready-made robots.