inFeedo’s Amber conversational bot taps into employee wellbeing to highlight disconnected employees through sentiment analysis and predictive models
The focus for inFeedo is on massive distributed teams since predictive models built around human behaviour and deep learning take time to refine themselves at scale
With a focus on embellishing its core product, inFeedo is looking to effectively tackle the challenges posed by remote employee engagement, the efficacy of managers and pulse surveys
The Centre for Monitoring Indian Economy data shows that the average employment reduced from an estimated 404 Mn during 2019-20 to 396 Mn in March 2020. In April it came down to 282 Mn. The overall job industry in India and the world has hit a wall with the pandemic rendering several roles useless in cost-cutting measures leading to layoffs.
The biggest concern for employees had been lack of clarity of the future— while most businesses are themselves unaware of what comes next due to the unprecedented crisis, the employee morale is dependent on the leadership. With the workforce moved to homes, the missing office culture, which could address concerns and help employees at an instant, has raised communication gaps. Here, players like inFeedo are trying to fulfil this address gap.
Being a part of Y Combinator’s Winter 2020 cohort, Tanmaya Jain-led inFeedo has been building an employee experience bot— Amber— to enable conversations with remote employees. It mainly focuses on predicting employee morale levels, identifies attrition levels, highlight employee pain points across teams and their managers and identifies culture, diversity and leadership gaps that often go unnoticed.
This seven-year-old startup began with the idea of allowing companies to harness their people’s power to drive bottom-up innovation, crowd solve key problems from teams across multiple locations, and collect anonymous feedback.
Today, with the major focus on Amber, inFeedo has built an AI-enabled engagement bot who periodically chats with employees and helps HR proactively find those who are disengaged or about to leave.
To build trust, the data Amber collects is used solely to help employees in distress with a link to Amber’s Trust Statement that highlights exactly who in the organization can see this feedback. Jain told Inc42 that Amber comprises two tech layers for these users — the first which analyses the sentiments behind the conversations and the other which gives actionable insights based on these.
Sentiment Analysis + Predictive Analysis = Amber
“Amber connects with employees at specific and customized touchpoints to collect their experience data through a chat module. The conversations are based on the employee experience framework which has the entire question bank curated by People Science team based on seven experience drivers and 52 corresponding elements to map sentiment for HR on the dashboard,” Jain told us.
Amber’s natural language processing and sentiment analysis model is built using inFeedo’s proprietary LexScore algorithm. On top of sentiment analysis is predictive analytics. Jain explained that sentiment analysis data is broken down into actionable reports with key insights into individual and overall organisation sentiment across demographics.
“Our sentiment analysis is built on a lexicon-based algorithm called TexSens which can decipher positive, neutral, and negative sentiment for HRBPs to take quick action,” Jain added.
The AI bot also measures actionable insights via various reports from engagement, mood, people-to-meet (high-risk employees), driver-element heat maps (on experience strategies. “Amber works on the data she collects from employees and her analysis is logic-based, pre-configured and recalibrated with time-based on organizational needs,” Jain explains.
The Complications Of Large Workforces
Currently, Amber engages with over 300K employees across 100 enterprises such as Airtel, GE Healthcare, Nivea, Lenovo, AXA Affin General Insurance, Tata Group, Unilever, Puma, and OYO. However, notably Amber works only with companies with minimum workforce strength of 150 or more around the world.
But predictive models built around human behaviour and deep learning take time to refine themselves and work most effectively at scale. So the focus for inFeedo’s Amber is on larger distributed teams. The company initially started with Google, Watson & Stanford libraries to analyse sentiments however it had to rely on its in-house algorithm to improve the accuracy rates. “It took us a year and training over a million employee responses to make our sentiment analysis one of the most accurate domain-specific algorithms in the corporate context,” Jain added.
The concept for Amber was to bridge the human resource business professional (HRBP) and employee gap from 1:300 to 1:11 on average. This helps understand individual sentiment better and create better experiences.
Currently, Amber integrates with most HRMS in the market and reaches out to employees through automated emails, SMS where her chat link is shared. Going forward, the idea is to integrate Amber seamlessly into work tools like Slack.
The Roadmap To $10 Mn Revenue
Jain told us that the company has raised INR 5 Cr ($700K) from Y Combinator and angels including Dr. Ritesh Malik (founder, Innov8), Peyush Bansal (CEO and cofounder, Lenskart), Vinod Muthukrishnan (CEO at Cisco-acquired Cloudcherry), Anand Chandrasekaran (former Snapdeal CPO and Facebook director), Sunder Nookala (CEO at OC Tanner-acquired Kwench) and Anil Advani (CEO at Inventus Law) and Siddhartha Ahluwalia (Former Co-Founder at Sheroes).
The company plans to use the funds to hire senior executives like COO, VP Engineering, VP Marketing and more to scale the organisation to 1000 customers and $10Mn Revenue.
The company closed FY20 at ARR of $1.6 Mn and claims to be growing at 100% Y-o-Y. Jain said that 30% of its revenue comes in from international customers. As per company filings accessed by Inc42, the company reported a revenue of INR 3.78 Cr in FY19 with expenses of INR 2.75 Cr, as against INR 1.2 Cr of revenue in FY18 with expenses of 1.02 Cr.
The company’s FY20 performance couldn’t be verified with the relevant filings at the time of publishing.
With a focus on embellishing its core product, inFeedo is looking to effectively tackle the challenges posed by remote employee engagement, the efficacy of managers and pulse surveys. Amber will reach a candidate even before onboarding to capture their experience at the pre-hiring stage.
At the same time, manager effectiveness with the upcoming My Manager product as a key driver to understand employee-manager dynamics. “What will be added to this is personalised views which allow the HR admin to share detailed feedback through reports to the manager. Further, managers will be nudged with insights and lessons on performing better,” he added. It is also looking to add Pulse surveys with email automation and integrations and let organizations customize these templates for multiple sorts of templates.
As per Deloitte India’s 2020 Workforce and Increment Trends Survey, attrition at an all India level in the current fiscal year is about 15%. Involuntary attrition (layoffs, restructuring, etc.) has increased to 20% of the total attrition which will go up further.
The negative sentiment combined with the remote workforce has increased the need for better employee engagement and wellbeing management.
“With COVID-19 setting in the new remote work normal, how Amber stands apart from its competitors is its empathetic persona that allows her to not just ask but also respond to the feedback from employees. It helps HR establish a human connection and solve the mental wellness issue that social isolation can exacerbate,” Jain explained.
