How Whatfix Is Going All-In On AI As It Eyes A $150 Mn ARR Milestone

Amid the accelerating adoption of AI across sectors, the CEO of Microsoft, Satya Nadella, is quite vocal about the end of the pure-play SaaS era. Of course, this has sparked a discourse of sorts, with many arguing that AI will only enhance the traditional SaaS industry rather than become the harbinger of its end.

The argument holds heft, especially when SaaS organisations are doing everything in their might to avoid becoming obsolete in the age of AI. A great example is that of legacy SaaS giants like Workday and Salesforce, which have dedicated AI teams and are investing in R&D to offer AI-first capabilities to their users.

As far as many others of similar nature are concerned, a clear metamorphosis is visible from SaaS to AIaaS (AI-as-a-Service). 

Witnessing this shift, companies are focussed on automating workflows or training bots and analytics assistants and integrating AI models like GPT, Claude, or Gemini. Such ready-to-use AI tools are helping businesses integrate intelligent features without the pain of building complex AI infrastructure from scratch.

In India, Whatfix is one of the few SaaS companies leading this change. Founded in 2013 by Khadim Batti and Vara Kumar, Whatfix is a B2B SaaS startup that enables organisations to make the most of software tools across applications/software like CRM, ERP, and project management.

Simply put, Whatfix is a B2B SaaS company that helps organisations improve software adoption and user productivity. The company makes it easier for people to use complicated business software, using in-app guidance and AI to boost productivity.  

Its existing products include a digital adoption platform (DAP), a product analytics platform, and Whatfix Mirror, which creates an interactive simulation of web applications for virtual hands-on user training.

However, Whatfix’s tryst with AI began  in 2019. Nearly five years later, all its products are AI-enabled.

Headquartered in Bengaluru, the startup currently receives a majority (about 68%) of its revenue from the US market, while it generates around 22-24% from the UK, Germany, and France. India, Australia and Singapore contribute nearly 10% to its revenue stream.

In FY24, Whatfix’s operating revenue stood at INR 425 Cr versus a net loss of INR 263 Cr. Speaking with Inc42, the company’s founder and CEO, Khadim Batti, said that its current annual recurring revenue (ARR) stands at $58 Mn. 

The startup aims to cross $150 Mn in ARR ahead of its IPO, expected within two years. Its AI playbook is expected to play a crucial role in this quest.

Whatfix’s Brush With AI 

In 2019, when Whatfix first realised the universe of opportunities waiting to be unlocked with AI, it acquired Airim, a tech startup specialised in building AI-powered personalisation engines for users and customers.

At the time, the startup was still working with old models to develop element detection tools. Elements in any tool refer to buttons, input fields, dropdown menus, tabs, etc. The company started working on element identification via computer vision to help users navigate through its tools and software more seamlessly, effortlessly and with precision. 

However, before Whatfix could completely roll out the feature to its customers, OpenAI started launching its models around 2020-21, which changed Whatfix’s AI game altogether.

“With the launch of OpenAI models, we saw a significant opportunity to upgrade our traditional AI approach. Initially, our use of AI was limited to element identification. But once OpenAI entered the picture, we began integrating a range of AI capabilities — summarisation, email drafting, and generating dashboards and reports — into our products,” Batti said, adding that Whatfix, in the last few years, has spent a few million dollars each year to make its products fully AI-enabled.

While Batti did not disclose how much the company spent on R&D, he said that initially, R&D on AI was only 5% of the company’s endeavour. However, today almost every project has AI embedded into its core. After OpenAI’s model launch, the company’s total spending towards AI has almost tripled.

Over the years, Whatfix has also secured several patents. These patents include a method for creating steps in providing digital guidance to an application, personalisation of digital guidance content, and more.

Making strides in its AI journey, the startup launched ScreenSense earlier this year. The tech uses patented algorithms to create next-gen digital experiences. It is more intuitive to the end users, understands the context of a user activity on a software, and helps the user with recommendations to unleash any tool’s full potential.

What’s Whatfix’s Stand On AI?

Like most AI services companies, Whatfix leverages existing LLMs, such as GPT 3.5 and Google Gemini, and tunes them on its existing datasets.

“Initially, we relied purely on training models on tens of thousands of processes, but eventually, we are using a mixture of training and augmented retrieval on more than 50K processes, over 80K knowledge repository content, and 1 Mn+ unique screens across more than 750 software applications,” said Batti

Given that Whatfix works with multiple companies that already use various LLMs in their internal workflow and have different requirements in terms of web browsers, the startup is more flexible in terms of choosing the foundational model.

The startup might consider building its proprietary small language models (SLMs) in the future. However, it has stayed away from building SLMs so far because they might not deliver the kind of accuracy required for element detections and other use cases Whatfix solves for its customers, the founder said.

Whatfix is also spending on hiring data scientists and upskilling its existing team of engineers to build its AI capabilities. Besides, Batti said, each of Whatfix’s departments has a proper budget allocation for using AI. 

Its engineering teams use AI for code improvisation or coding efficiency and the product management department uses it for product requirements document (PRD) reviews. 

In addition, like most companies, its customer support teams use AI for ticket analysis, responding to specific resolutions and summarisation purposes. Whatfix’s marketing team uses it for the campaign creation and account research.

While some of the departments are in their early stages of adoption, many others have generated steady ROI with AI adoption in internal workflows.

While business enterprises comprise the majority of its customer base, Whatfix has also started working with government organisations to deliver its AI-enabled SaaS products. 

The startup has created a team solely for working on federal contracts in the US. Recently, it closed a deal with the US Army.

Batti said that large enterprises and government agencies look for companies that can continuously innovate and ensure high data privacy and security. 

Whatfix, with its 26 patents in total, new product launches every year, and a proven track record of complying with regulatory requirements, has been able to crack high-profile contracts like that of the US Army.

Whatfix In The AI Wonderland

Though Whatfix has upgraded its SaaS capabilities with AI, it hasn’t yet increased the prices of its products. The company is currently looking for more engagement from its customers, and it will decide on a rate increase in the future, depending on the consumption pattern.

Batti is confident about the pace of AI adoption at global enterprises. He said that 18 months ago, these large enterprises were only curious about developments in AI, but they stayed away from deploying AI-based products, due to data security concerns.

“Fast forward to Q4 of 2024, when I spoke to these enterprises, most of them had IT policies evolved to evaluate AI. So, now they are AI-ready in terms of evaluation. Many of these companies have created AI task forces and committees to identify use cases for deploying AI,” he said.

Whatfix is also looking to double down on its businesses in the APAC region. The startup eyes the APAC region to soon start contributing at least 15 to 20% to its revenue from the current 7%-8% level.

In the short term, Whatfix aims to completely move away from traditional element identification algorithms.

[Edited By Shishir Parasher]