Wipro-Backed Avaamo Takes Low-Code Approach With LLaMB To Meet GenAI Enterprise Demand

“For GenAI, last year was the honeymoon year. Which was about being amazed with what you can do with it. Today, we are in the meat and potatoes era. And now enterprises are asking themselves, how can I save money?” — Ram Menon, cofounder and CEO, avaamo

If 2023 was the year of generative AI (GenAI) and GPT models, for Avaamo CEO Menon, it’s been 10 years in the making. Founded in 2014, Avaamo has seen the trajectory of AI-enabled enterprise services up close.

From conversational AI to call centre AI to low-code machine learning and automation skill building, Avaamo has seen the AI ecosystem mature and reach the crescendo that’s ringing today. And now it’s getting ready for the GenAI world with Avaamo LLaMB, a suite of products developed specifically for large enterprise use cases.

While Avaamo has worked with the likes of Pepsi, HSBC, Intel and other massive enterprises, LLamB is debuting with a deployment at Wipro, which Menon says is the largest GenAI deployment globally, addressing 40+ use-cases for more than 2 Lakh employees across the world.

Besides being the first customer for Avaamo LLaMB, Wipro’s corporate venture capital arm Wipro Ventures was the first investor in Avaamo way back in 2014. So, in a way, this is a full-circle moment for Avaamo, as it celebrates its 10th anniversary this year. And it’s this decade-long journey that has given Menon and cofounder and CTO Sriram Chakravarthy a unique viewpoint on the massive GenAI opportunity relatively early.

Before Avaamo, Menon was part of the early team at TIBCO, another enterprise software giant, which was acquired by Vista Equity Partners in 2014 for $4.3 Bn.

Avaamo cofounder Chakravarthy is also a TIBCO alumnus and was the VP of engineering at the time of the acquisition. The duo have over four decades of combined experience in the enterprise tech space.

As Menon tells Inc42, “When we got to GenAI, I dare say that we knew a thing or two about natural language interfaces. Salesforces and other enterprise giants came in 10 months ago, but we’ve been doing it and dealing with all the mistakes and we have all the scars on our back from servicing these large customers. So it was relatively easier for us to incorporate GenAI into existing enterprise frameworks and architecture.”

Avaamo LLaMB is primarily a low-code framework that enterprises can use to build generative AI applications for their specific use cases.

Menon claims this is faster than anything else in the market. He believes that the use of generative AI in the enterprise has been mostly experimental with products around retrieval-based generation use-cases with a few documents. This does not unlock the true efficiencies in productivity and operations that GenAI promises to bring.

The LLaMB framework, on the other hand, is focussed on key challenges in enterprise adoption related to the accuracy of responses and data security — eliminating hallucinations, bringing together siloed enterprise data and content, automating cross-enterprise workflows and allowing large companies to accommodate any large language model (LLM) that suits their operational needs.

In a press statement, Anup Purohit, CIO of Wipro, said, “Avaamo and Wipro have created novel ways to improve employee experience using GenAI. We have been working together to implement an AI playbook and have deployed over 40 use cases for 200,000 employees spread over 53 countries. By building on Avaamo’s GenAI technology in new and secure ways, we plan to continue our digital transformation journey and further enhance our employees’ experience.”

Over the years, Avaamo has raised more than $30 Mn from the likes of Intel Capital, Ericsson Ventures, and Mahindra Partners, among others.

Avaamo’s GenAI Playbook For Enterprises

The central tenet behind Avaamo LLaMB is that enterprises do not want off-the-shelf LLMs that can be arduous and more expensive to manage due to their inability to cohesively parse through critical enterprise data, unsecured prompt environments, as well as lack of data moderation.

Chief information officers across enterprises are excited by GenAI, but almost everyone recognises the dangers of letting an LLM loose on company data. Enterprise-grade accuracy and security are paramount.

Avaamo’s LLaMB’s suite of capabilities is build around four key pillars:

• Trust Architecture: Encryption of all prompts with a zero retention guarantee and GDPR, CCPA, and SOC2 compliance.
• Dynamic Prompts: Lower burden on enterprise developers to craft, manage or update prompts and support for custom libraries for accurate responses
• Data Sync: Connectors for structured, unstructured, or semi-structured enterprise systems
• Data Moderator: Monitoring responses for tracing weak, contradictory, or outdated information and actionable insights to identify and plug gaps

To start off, the LLaMB framework enables self-service experiences across call centres, HR, IT, procurement and patient communication in healthcare.

Menon showed us a demo where an employee could simply talk to their in-house LLaMB-powered travel desk or IT procurement agent to book the right kind of itinerary or get access to the right device, according to their level in the corporate structure or budget allocations.

“The problem with the open model is inconsistency. And sometimes this can be disastrous. Take a practical example such as summarising documents when employees ask the platform a question. AI can summarise a travel policy because this is not critical data, but it cannot, for instance, summarise an FDA warning to a healthcare organisation, where each line and specific wording matters. Or it cannot summarise a lender’s disclosure.”

Enterprises want GenAI tools that understand precisely which information to surface as well as the organisation’s intricacies. This is a critical contributor of trust, Menon says.

Another friction point is the provenance of an answer. Where is the response from GenAI coming from? “For instance, is it a citation and which revised version of the document is it coming from? Enterprises need explicit dates and the context of the creation and how it adheres to that data policy,” Menon says

Avaamo’s approach revolves around the dynamic grounding of the data, where the model doesn’t invent anything or doesn’t create content. It only uses content that’s verified to be from the enterprise.

While enterprise SaaS promises to rid large businesses of inefficiencies, one of the key gaps is engagement and compliance with tech platforms. Often the burden of using a separate platform results in employee engagement dropping off, resulting in low ROI on the spending.

Avaamo’s LLaMB is looking to eliminate this giving CIOs the flexibility to choose between Avaamo’s interface or consume the framework as an API. So if a company relies on Microsoft Teams or Slack for communication and workflow management, LLaMB can be accessed through the chat dialog box like any other contact or bot. Wipro, for instance, is deploying Avaamo LLaMB within Teams, which Menon says results in faster and more immediate adoption.

Menon is confident that the enterprise adoption wave will be the big inflexion point for GenAI applications. As per Inc42’s report on the Indian Generative AI Landscape, the global market opportunity around GenAI is poised to grow to $552 Bn+ by 2030.

Further, beyond enterprises, GenAI adoption among startups is also set to boom with nearly 3/4ths of startup founders surveyed by Inc42 last year eager to increase their technology spending by up to 30%.

While he also expects the adoption to trickle down to smaller and mid-cap companies in the GenAI future, Menon’s sights are largely set on generative AI for enterprises, because as he says Avaamo’s DNA is built around large outcomes.

Besides embracing Avaamo’s LLaMB framework, Wipro has launched Wipro ai360, which is envisioned as an ecosystem of AI products and solutions to be used internally by the company and offered to its clients. The IT giant also committed to making a $1 Bn investment in improving AI capabilities over the next three years.

Other Indian enterprises such as Reliance Jio and Infosys are also ramping up their investments in this space.

Bringing Wipro on board for LLaMB is a big signal about India’s place in the GenAI market, Menon says. The company is eyeing the Indian market not just for deployment but also from a talent point of view. “Roughly 70% of our company is in India. All this technology is being built in Bengaluru. We are a startup, but we are a mature startup. Plus, when you’re profitable, it gives you freedom to build a business the way you want it,” the CEO tells us.

Avaamo’s focus on next-gen tech means it also has to bear the burden of identifying and training the right talent. “We found great engineers in Bengaluru and we also do a lot of recruitment from my alma mater Manipal University, and BITS Pilani, where my cofounder graduated from. We bring graduates from these and other colleges, we train them and we send them to the US. So we have methodically built a startup in a very different way,” Menon adds.

His mantra is not focussing on headlines, not burning money, and being product-first. Of course, the world’s largest deployment of GenAI for enterprises means Avaamo is definitely among the headlines now.