SaaS Talks #28: Integrating LLMs in your SaaS product

LLMs are all the rage these days, and for good reason - they have the potential to significantly improve existing software, while also helping re-imaging enterprise workflows.However, using LLMs to deliver robust performance that your customers expect is no small task.

What use cases should one choose, what model should one pick, how does one avoid hallucinations in output, should you consider open source models - these are just some of the questions you need to answer to get LLMs ready for your business.

In this edition of SaaS Talks, we will focus on the different aspects of integrating LLMs in your SaaS product.

00:00 – 02:03: Introduction

‍02:04 – 07:00: Whatfix’s journey with LLMs

‍07:01 – 09:41: Change in the complexity involved in building AI products post LLMs

‍09:42 – 12:57: Good and not-so-good use cases for LLMs

‍12:58 – 17:36: Approach to developing team skill sets for AI

‍17:37 – 20:49: Where does the moat lie for SaaS companies?

‍20:50 – 25:56: How is the space moving, specifically in co-pilots?

‍25:57 – 27:31: Will the edge for fine-tuned, specialized models evaporate with data becoming public?

‍27:32 – 30:46: How will people choose between open and closed models?‍

30:47 – 32:13: Successful use cases in performance marketing

‍32:14 – 35:49: Applying guardrails in production use cases of LLMs

‍35:50 – 37:52: Using point solutions v/s centralizing tooling

‍37:53 – 40:08: Impact on gross margins of SaaS companies deploying LLM use cases

‍40:09 – 42:49: How should one think about pricing?

‍42:50 – 44:31: How sensitive are enterprises to privacy/ security issues?

‍44:32 – 47:51: Emerging industry-specific LLMs, vertical v/s horizontal, prompt v/s fine-tuned

‍47:52 – 50:50: Making choices while integrating LLMs

‍50:51 – 53:10: Co-building with an anchor client

‍53:11 – 55:15: Does using AI give a new SaaS co. an edge over an existing large SaaS co.?

‍