Dhruv Thakkar knows what it takes to bring emerging technology into the enterprise world in a way that actually drives results. As a Senior Solution Architect, he has helped businesses use SaaS technology and AI to solve complex problems and build smarter, more efficient operations. He also volunteers his tech expertise to help nonprofits amplify their impact.
In this interview, he shares how enterprises are deploying Generative AI and Agentic AI to work inside their SaaS platforms, where the biggest opportunities lie, and how to balance innovation with security.
Q1. How are enterprise organizations starting to incorporate Generative AI and Agentic AI into their SaaS platforms?
Enterprises are bringing Generative AI into their SaaS platforms by building AI Agents for different roles like Co-pilot assistants, employee support agents, customer service agents, and even coaching Agents. The big advantage is that businesses can customize these AI agents to fit their specific needs, whether that’s automating repetitive tasks, improving customer service, or helping with strategic decision making.
Q2. What are some of the most promising use cases where AI is currently enhancing enterprise tools like Salesforce, CPQ software, and billing systems?
The most promising use cases for AI enhancing the CRM and Quote to Cash industries are AI Agents. These are digital employees that help small to large enterprises by automating tasks, enhancing efficiency, and improving decision-making. Some prominent use cases for industry are AI Agents performing various roles like customer service agents, sales coaching agents, and internal employee assistant agents that do knowledge retrieval.
For the CPQ industry, AI-enabled complex quoting, automatic product bundling, personalized guided selling, AI-recommended contract renewals, and upsell opportunities are some use cases where AI has already had a big impact in the billing domain, AI is enhancing invoicing, improving fraud detection, and streamlining payment processes, all while helping businesses stay on top of their revenue cycles.
Q3: In your view, how do technologies like large language models, fine-tuning, and vector embeddings enable more intelligent decision-making across enterprise workflows?
In my view, LLMs act as the brain of any enterprise workflow. LLMs automate any complex tasks while maintaining human-like reasoning. While fine-tuning helps LLMs understand specific business contexts, basically personalizing them for an industry or a specific company, whether that’s understanding industry-specific acronyms or aligning with a company’s processes. Vector embeddings help connect unstructured data (like emails or PDF files) to structured systems, enabling smarter search, recommendations, and decisions.
That said, as these systems become more central, ensuring security and compliance becomes absolutely critical.
At a high level, this starts with governance—understanding what data is being used, where it’s coming from, and how it flows through the system. Fine-tuning is a great way to personalize LLMs to specific industries or companies—helping them understand internal terminology, business processes, or compliance standards. But even before that, companies need to implement guardrails like encryption, data masking, and access controls to protect sensitive information from the start.
Q5: What are the key considerations enterprises should keep in mind when integrating AI-driven automation into existing SaaS ecosystems?
AI models rely heavily on clean, structured, and relevant data; hence, enterprises should focus on their data quality as a priority. They should refine and pre-process their data, and have workflows in place where data can be easily readable to AI. Enterprises must also prioritize a data privacy and security strategy. They would need pre-processing power to only send relevant data to LLMs and save token usage.
Q6: How can businesses ensure data security, governance, and compliance while adopting AI at scale within these platforms?
Businesses should start by identifying their sensitive information upfront and apply security measures like Data Masking and Encryption before that data is exposed to any LLMs or used in any model fine-tuning.
Establishing a Zero Data Retention (ZDR) policy with external AI providers is also critical, ensuring that sensitive customer data is never permanently stored. Additionally, businesses should enforce granular access controls to limit AI system access strictly to authorized users and relevant datasets.
Q7: What impact have you seen AI make on customer engagement and personalization in enterprise environments?
Since AI can process vast amounts of data and understand the patterns and preferences of individuals, customer engagement has seen a transformational impact from AI. AI-driven chatbots have been widespread, and it has helped businesses achieve efficiency, retention and high customer satisfaction. AI has also helped personalize content across digital channels ensuring customers receive relevant messaging. Overall, AI has shifted enterprises from reactive customer engagement to proactive customer engagement models.
Q8: Can you share how you or your team have contributed to advancing the use of Generative AI or Agentic AI in enterprise SaaS applications?
At Salesforce, we’ve been deeply involved in advancing the use of Agentic AI within enterprise SaaS applications. Through our new product ‘Agentforce’, we’re developing AI agents that enable customers to deploy digital employees. These Digital Employees are capable of working independently. The AI Agents act as intelligent assistants, customer service agents and help answer questions, retrieve relevant knowledge, identify and resolve issues by interfacing with support software APIs and automate workflows.
Q9: How do you recommend organizations balance innovation with operational stability when implementing emerging AI technologies?
Organizations should first identify their repetitive business processes where GenAI can be applied. They should build POCs or MVPs and have pilot users use them while keeping their original systems ongoing for stability. They should clearly define success criteria for GenAI projects. Organizations should ensure phased rollouts of their GenAI applications, which gives enough time for teams to adapt. Once AI is rolled out, tested, and iterated, businesses should start including AI in decision-making. AI can provide insights and opinions with clear data and reasoning help organizations make better strategic decisions.
Q10: Where do you see AI in enterprise SaaS heading in the next 2–3 years, and what should leaders prepare for?
In the next 2–3 years, AI is headed to drive core structural changes in the Enterprise SaaS industry. More organizations will adopt AI Agents and shift toward agentic architectures, where autonomous AI agents working as digital employees will significantly enhance operational efficiency.
Leaders will need to focus on data readiness, making sure their data is clean, structured, and AI-readable. Data preprocessing will be a key priority. At the same time, teams will need to be trained to work with AI systems and underst how AI Systems differ from traditional SaaS due to their inherent non-deterministic nature. This enablement will help teams identify high-impact AI use cases within their organizations.
Another important decision for leaders will have to make in the next 2-3 years will be whether to build in-house AI Agents using open-source models or adopt solutions from emerging vendors offering AI agents on a per-inference or per-conversation basis.
Ultimately, this shift is not just about adopting a new technology but it is a fundamental change in how Enterprise Businesses will operate. Those Orgs who will tackle these core challenges early will be best positioned to leverage agentic AI fully.
