Ema’s recent viral blog post, on How to Build a 100-person company without HR, inspired us to share more on our philosophy of using Agentic AI wherever possible, to streamline our administrative overhead and create a culture of autonomy. We’re committed to practicing what we preach at Ema, and so we create and deploy AI Employees that can automate mundane tasks, boost productivity, and free our teams up for high-value, creative work.

This is the story behind two of Ema’s internal AI Employees—Ema’s Employee Assistant and Sales Assistant—what they do, how they were built, the results they’ve achieved, and what we’ve learned along the way.

While this blog may spark some ideas for existing Ema customers, it can also provide inspiration for other leaders wanting to know how to implement Agentic AI for tangible results in their organization.

How did we discover Agentic AI use-cases within Ema?

In order to identify the most high-impact use-cases, we analysed our team’s workflows and processes and interviewed key management stakeholders. We found a few patterns across functions that senior leaders reading this would probably resonate with:

• Siloed data: Information locked in clunky, unwieldy SaaS platforms
• Tribal knowledge: Several key operational and technical learnings and best-practices hidden away in people’s minds
• Inefficient communication: Significant overlap in information discussed across the company, leading to key team-members repeating the same explanations over and over, serving as bottle-neck in efficient operations
• Repetitive, routine tasks: Significant time spent in mundane web-searches, information finding and summarisation tasks
• Overburdened teams: Certain functions, like security and compliance, are especially faced with heavy and unpredictable customer demand
• Loss of time spent in awaiting async responses from other teams

At Ema, Employee Experience (HR) and Sales were some of the key functions that dealt with the above patterns. We decided to tackle them first: Employee Experience is important because of its widespread impact on the productivity and well-being of everyone at Ema. Sales, of course, is as a strategic revenue function, which could be most affected by the above dysfunctional organizational patterns. Information silos and bottle-necks, especially in an industry as fast-paced as ours, can actively compromise a company’s ability to move fast, drive sales, and succeed.

Work Better, Faster: How Ema’s AI Employees help

Three months down the line, we at Ema have completely transformed our Employee Experience and Sales processes using Agentic AI. Both these AI Employees are available 24x7 on Slack, and perform a whole host of functions that streamline our team’s workflows.

Ema’s Employee Assistant:

• Provides personalised (geo-specific and role-specific) answers to all questions linked to Ema’s many policies, from insurance and benefits to ESOPs, vacation and sick leave policies
• Educates and onboards on Ema’s product offering and the AI landscape
• Performs HR actions, including leave application, fetching payslips, updating tax information, and more
• Streamlines onboarding for new joiners by creating a personalized onboarding guide for every new employee, also serving as a digital buddy to answer any questions about Ema’s culture, organisational principles, internal tools, etc.

In summary, we no longer have a traditional HR team for our team of 100+ employees, spread out across India and the US. Our Agentic AI-first approach has enabled us to be among the first few companies globally to achieve this. HR in Ema is therefore a more strategic function than usual, involved in building culture and hiring the best talent.

Ema’s Sales Assistant:

• Creates sales briefs with company background and buyer background alongside tailored pitch points for every customer meeting
• Autonomously answers all vendor questionnaires (including those on security & compliance) and writes responses to RFPs
• Provides intelligence across all customer call transcripts to identify sentiment of every customer, key selling propositions, key product questions & concerns etc.
• Provides intelligence on all of Ema’s Hubspot, allowing deep pipeline analysis using simple, natural language questions
• Connects with calendar and email to auto follow-up with prospects at regular intervals with personalised emails
• Creates pitch-decks ahead of every customer meeting tailored to the customer’s industry and scale utilising compendium of Ema’s existing decks
• Provides instant access to all sales SOPs, product documentation, customer onboarding guides etc.

“Ema’s Sales Assistant has transformed my entire sales process - I save about an hour before critical client calls typically spent on research and creating a starter deck. For each deal that progress to later stages, our pre-sales and sales team collectively saves around 4 hours on filling vendor questionnaires for clients. I have so much more time to focus on actual sales now.” — GTM Team Member, Ema

Using Ema’s Generative Workflow Engine™ to easily build AI Employees

We initially explored building Agentic systems to tackle our use-cases with Langchain. But after deploying some engineering bandwidth for a trial, we realised we were progressing at a painfully slow pace. We had to manually:

• Build, optimize, tune, and maintain AI Agents for every use-case
• Maintain data connectors with the SaaS platforms we were integrating with
• Create and configure AI Agents with tools needed to perform actions for each SaaS platform (e.g. RazorpayX for applying for leave, Slack for communication, etc.)

This is when we understood the need for a platform that would speed up our pace of development. And soon, Ema’s Generative Workflow Engine™ was born, to solve the challenges we encountered in our quest to create simple, effective, enterprise-grade Agentic Business Automation.

With Ema’s GWE, we brought over a dozen use-cases to life in just a couple of weeks—without writing a single line of code.

As the builder of Ema’s internal AI Employees, our Employee Assistant and our Sales Assistant, here’s what I loved about using Generative Workflow Engine™:

• No agonizing over communication and data-flow between agents
• Not having to struggle with reading documentation of various SaaS API providers to figure out authentication, data syncing or taking actions
• Not worrying about the impact of the latest LLM launches on our use-cases. EmaFusion™ intelligently combines the output of over 100+ LLMs, optimizing for accuracy, costs, and latency—while keeping AI investments future-proof

The ability to focus purely on the business logic and end-user experience whilst abstracting away all of the technical tedium is a magical experience. As a business-user, being able to build out these Agentic AI use-cases I’d conceived, without any technical dependencies, was liberating and enabled delivering of impact in record time.

Learnings from our journey, and how to start with Agentic AI

Using Ema’s AI Employees to streamline our internal productivity is a strategic priority, as we develop and sell our platform to customers with similar goals—and at a much larger scale.

If Ema’s AI Employees can save a few hours each week for 100-people—while unlocking new efficiencies and making space for high-value, creative work—imagine what is possible for tens or hundreds of thousands of employees, with Agentic Business Automation across functions.

Some of Ema’s customers have been leading the way. Here are common lessons from our journeys, for other leaders looking to achieve transformation in their organizations:

• Approach the problem customer-first, not tech-first: Involve the user in the solution process from Day 1; Leverage their insights and think impact-first when planning your solution instead of obsessing over the best LLM to use; Let solutions like EmaFusion™ handle the LLM-level optimisation for you
• Take a conscious decision on building from scratch vs utilising an agentic business automation platform: Think through the benefits and drawbacks of each approach in your context before finalising an approach to automate your workflows; Ensure that you account for developer hours, ongoing maintenance and development delays when thinking through the RoI on building from scratch
• Build incrementally starting with solving a single problem and build momentum with quick wins: Instead of launching the perfect AI Employee to automate all woes of a function, start by tackling a small yet thorny user problem and launch incremental features; Ramping up usage within the organisation takes time and multiple quick wins will help accelerate it
• SaaS is well and truly dead: We’ve said it before and we’ll say it again. Our entire journey so far has proven that in the new AI-first future, all data intelligence and manipulation can be done from the comfort of a native chat interface, without needing to navigate and click through mazes of SaaS platforms while paying for more seats than you need. Instead, opt for end-to-end agentic automation, delivered at a fraction of the cost.

We hope that our journey has given you the motivation to start experimenting within your enterprise today. If you wish to join the list of enterprises 10x-ing their productivity with Agentic AI, book a demo of Ema’s AI Employees today.