AI Creates Jobs. Want Proof? Here's Haptik's Story


The growth of Artificial Intelligence has spawned a thousand foes and a thousand friends. AI has drawn a line in the sand when it comes to the future of jobs. And it seems like most of the world has picked a stand either for or against it.
So what does Haptik have to say, being a company at the forefront of the paradigm shift to Artificial Intelligence?
Will people from non-technical backgrounds have job opportunities when Artificial Intelligence (AI) takes over?
Can one actually justify this claim?

Our answer is a resounding “Yes”.

And yes, we can justify our claim. We transformed our chat support assistants into AI Trainers who build and manage the company’s chatbots.

Here’s the story.

Automation: The Root Cause

Haptik began as a 24/7 Personal Assistant App that could set reminders, book cabs and handle other everyday tasks over chat. The conversation is initiated and managed by the bot. If the chatbot is unable to understand a certain query, the chat “breaks” and is handed off to a human agent who steps in to take the conversation further.

As our product suite grew, we built over 100+ chatbots dedicated to specific businesses and amassed a wide network of customer interactions across platforms. This meant that we had a lot of ‘training data’ on our hands. Training data refers to conversations and messages that our chat assistants have exchanged with our users over the years. With all this data we could train our bots using supervised and unsupervised machine learning techniques to become smarter and answer difficult queries on their own, i.e., automate our system.

Over time, the bot can answer queries resulting in lesser chats “breaking” to the agent. Ultimately, this means that over time as the machine gets better, the agent’s job becomes redundant. But, here’s where we turned redundant jobs into an opportunity for empowerment.

Chat Agent Then, Bot Master Today

An AI Trainer is someone who essentially ‘teaches’ a chatbot what to say and trains it to respond to different user queries. Usually, the first image that comes to mind when we think ‘AI Trainer’ is that of a super smart engineer who understands complex code.

However, who better than someone who truly understands the user’s needs and the chatbot’s abilities to train the bot?

The revelation: Chat Assistants could be excellent Chatbot Trainers

And so, in 2016, Haptik initiated programs to transform our human chat agents into bot developers and trainers. Our in-house tool to build and manage bots consists of a GUI that’s fairly easy to use. It depicts chat flows as simple flow-charts that can be added to and edited whenever needed.

As of April 2018, we’ve successfully upgraded more than 100 such assistants from simply handling queries to training a chatbot. This has resulted in a win-win scenario for all, particularly for the assistants. The shift diversifies their job profile over an array of roles and empowers them with job skills for the future.

The Jobs of the Future

The AI Trainer role at Haptik is broken down into 3 sub-areas:

1. Bot Builders: Making a bot end to end i.e. developing the character, tone, content, and responses of the bot.
2. AI Bot Trainers: Monitoring user queries, labelling data that is unanswered and feeding it back into the data set for the bot to learn and respond to those queries over time.
3. Bot Quality Analysts: Reviewing the bot from a qualitative perspective, understanding user requirements and relaying that feedback to the product management team to improve the experience.


Haptik’s Chatbot Building Tool: Graphical Representation of a Chat Flow

We trained our assistants by adopting an integrated approach through which every assistant works in tandem with the Machine Learning team. Each assistant is paired with one ML team member, who teaches them the end to end process of bot building. They have weekly meetings for 2 hours each, where the assistants discuss their doubts and receive a progress evaluation report. They are taught the processes of using Haptik’s internal bot platform to create bots and also understand how to enhance the chat experience.

The Results Speak For Themselves

The results of this initiative have surpassed all our expectations so far.

While assistants are immensely enjoying their newfound roles as creators instead of just communicators, they have managed to improve the user experience of Haptik’s chatbots as well. Currently, 90% of the bot building and maintenance tasks are being handled end to end by the erstwhile human assistants.

It not only helps to offload work from our engineers but adds to the existing skill set of people who otherwise come from a call centre background, thereby enhancing their long-term career prospects. A great example: One of our bots was at 80% automation but once we moved an assistant to a bot builder role – the automation went up to 92%.


analytics tool

                                 Haptik’s Analytics Tool: Helps Gauge Bot Performance

The assistants at Haptik have managed to find greater job satisfaction and relevance with their enhanced technical skills, understanding the internal workings between nodes, APIs etc, structuring them to achieve maximum efficiency and exploring the endless possibilities that Machine Learning and chatbots offer.

The Masters Speak

“As travel assistants, we knew the process for flight or train bookings, usage of tool etc.
But as bot trainers, we actually understood the internal workings of each process like the connections between all nodes, APIs, the partner’s expectations and so on.” – Melissa Vaz, Chat Assistant turned Bot QA.

“As an assistant, work was somewhat monotonous but as a Bot trainer, it’s more interesting, because there’s a lot to learn and explore with the Bot. It’s very exciting to train a bot and see it working efficiently based on how you have worked on it. In a way, it’s like making your own Robot and see it moving around! – Akash Singh, Chat Assistant turned Bot QA.

“What I enjoy most about being a Bot Trainer is that I’m now able to teach my bot to think on its own instead of thinking of responses myself. It’s the difference between teaching a man to fish or giving him a fish and I thoroughly enjoy the process. Plus, the boost to my resume is a nice perk! – Nikhil Bhagat, Chat Assistant turned Bot QA.


At Haptik, we often get asked the question: “Will Artificial Intelligence take away jobs?” Our answer to that consistently has been that historically, technological advancements have always led to greater job creation, which will hold true for Artificial Intelligence as well. There is no better example of this than the paradigm shift that has been happening at Haptik.


Related Articles

View All