ï»żIntroduction to AI for Small Business Strategy (00:00)
Brad Eather: Hello and welcome to the Sellings Creative Podcast, a podcast exploring creativity's role in sales. I'm your host, Brad Eather, a digital communications sales enabler helping established businesses sell on social. Today, we're going to tackle a subject that I've deliberately avoided on this podcast until now. One of the reasons is it's very complex, and everybody's got an opinion. That's right, we're going to be talking about AI.
Gaurav: Sheesh.
Brad Eather: AI has made people feel uncertain about the future, but it has also made people excited. What I really want to unpack today is what AI actually means for us right now. It's one thing to project what might happen in the future, but it's another thing entirely to understand how it's already changing the way we think and operate in business today. This is exactly why I wanted to talk to someone who lives and breathes this stuff âsomeone who not only understands AI, but understands it in a business context. He has an MBA, a background in media communications, and an AI-focused data science background. Please welcome to the show the Head of AI Strategy and Solutions at Warp Development, Gaurav Dev Sharma. I hope I got that right.
Gaurav: That is correct, Brad. Thanks for the warm intro, and I'm really excited to be here. As you said, everything is changing so fast in this space. It's always interesting to talk about different verticals or jobs and look at how AI impacts them. I'm very keen to get started today.
How to Use AI in Business Operations Across Niche Industries (02:19)
Brad Eather: Like you said, AI is changing at a rapid rateâso rapid that it's impossible for everybody to understand every single aspect and where it's going. What kinds of businesses do you focus on at Warp Development?
Gaurav: At Warp, we focus on pretty much everything. We don't limit ourselves to specific niches, and that is one of the main things about AI: it affects everyone. It's not something meant for just one industry or another.
Obviously, we do see patterns where the immediate return on investment varies by sector. For example, manufacturing is a very interesting space for AI. Historically, a lot of manufacturers did not make the jump from step one to steps two or three. They skipped things like having a CRM, standard sales products, or an ERP system. Instead, a lot of complex modeling for pricing or operational data was just set up on local spreadsheets.
Quite interestingly, AI now helps them jump directly from step one to step four. Suddenly, you don't need to hire an expert to manage Salesforce or HubSpot. Instead, you can build custom software solutions for these firms based on the data they already have. The quality of that data matters, which I think we will dive deeper into later today. However, you can clearly see the shift in this industry where companies can bypass some of the rigid steps normally forced upon them in a non-AI world.
Contextual Information vs Technical Coding for Problem Solving (05:13)
Brad Eather: Perfect. Before we go any further, I want to kick things off talking a little bit about you. I mentioned that you've got degrees spanning from communications to data science. How do you think that unique mix of knowledge and experience has given you a distinct perspective on how AI is playing out, and how has it set you up for where you are now?
Gaurav: Sure, and that's a very interesting question. When you look at people working in AI, I don't really have a conventional background. But I think that is exactly my advantage: I look at it from a business perspective. It's very easy to get lost in the technical details. Coming from a diverse background, I look at the situation and say, "Okay, let's forget about the AI for a second. What is the actual problem or opportunity here ?" From there, I ask how we are going to solve it, whether we actually need AI, or if there is a simpler alternative. You reverse-engineer from the core issue and apply AI only if it's genuinely needed.
We often see highly talented individuals with PhDs and heavy machine learning backgrounds focus purely on the technical nuances. They are incredible at statistics, math, and understanding complex models. But the reality is that when you are solving a business problem, the stakeholders don't care what models or math you are preparing. They care about the problem that keeps them up at night and how you can solve it. My advantage has been that my background quite naturally forces me to focus on that business layer rather than getting caught up in the underlying code.
Overcoming the Trap of Personalized Outreach Saturation (07:40)
Brad Eather: Yeah, because AI can be confusing for a lot of people. You've brought up two different aspects here. Most people using AI are interacting with tools like ChatGPT for creative outputs, using it as a partner to write, synthesize ideas, or simply get words onto a page. The other side involves utilizing custom models as a knowledge hub , giving organizations access to massive amounts of centralized information.
Where AI has taken a massive leap is in its ability to process qualitative data. Historically, we used quantitative data to find commonalities, but AI now gives us the ability to find rich commonalities within qualitative data. I want to hear your perspective on these two applications. First, put yourself in the shoes of a salesperson using generalized AI for the creative, writing side. Then, let's transition into a business context and talk about AI agents and how we can leverage company IP for a true business advantage.
Gaurav: Absolutely, those are great points. When I answer your first question, I might actually annoy a few people, to be fair. Yes, it is good to use tools like ChatGPT as a starting point for outreach, whether that is via email or on LinkedIn. It helps you develop an initial intuition for what works and what doesn't.
But what excites me much more is how these tools can push a salesperson to develop a deeper technical understanding. Traditionally, unless they are in technical sales, salespeople don't have a deep grasp of the underlying software product they are selling. AI tools change that.
If I am trying to build a software implementation proposal for a client and I don't have a technical background, I can use ChatGPT and other tools to draft those complex details. I can look at our past company projects, pull that understanding into the AI, and get remarkably far. This saves an immense amount of time that would otherwise be spent going back and forth with a developer to explain basic technical concepts for the proposal.
It allows you to be creative. You are pairing your natural sales capability with the AI's ability to understand technical architecture. You can feed it the specific problems the client wants to solve from your meeting notes and generate a highly tailored proposal. Again, it's a starting point. You should never blindly copy and paste stuff; you always need to apply your own mind to it. That's where the real value is.
Balancing the Human Element with Automated Content Slop (11:34)
Gaurav: On the flip side, automated personalized outreach is a space where I think we are witnessing a race to the bottom. There are thousands of companies right now scraping LinkedIn or other social media channels to blast automated "personalized" messages. It feels like a tar pit idea. It looks highly attractive initially, but now 500 new companies rock up every day doing the exact same thing. A lot of them are heavily funded by places like Y Combinator, which I find surprising.
What this ends up creating is an absolute slop of emails in everyone's inbox. It can be the most beautifully personalized email in the world, but if a prospect receives 200 of them every single day, it becomes an annoying problem regardless. Email filters catch a lot of it, but this generic automation is not what will drive future value for AI in sales.
Behavioural Changes and Interactive Voice Tool Simulations (14:21)
Brad Eather: What gets lost in the AI conversation is that this is just as much about humans as it is about technology. Since we both share a background in communications, you might be familiar with the philosopher Marshall McLuhan, who famously wrote "the medium is the message". One of his core philosophies was that as technology develops, it acts as an extension of ourselves. Think of jumping into a car: it gives you the ability to go further, but you also physically extend your awareness to the vehicle. You feel where its edges are and become one with it.
AI is doing something similar, but it is extending a different aspect of what it means to be human. When we look at behavioral traits, oversaturating people with automated emails just makes the human recipient desensitized to it. How should we be using AI to actually stand out, protect our unique humanity, and act as the true author of the tool rather than just having another AI agent talking to an AI agent?
Gaurav: That is a great point. I'll give you a practical example that I find incredibly valuable: using real-time AI voice features. Before an important client meeting, I will feed the AI context about the target company and the objectives of our upcoming discussion. Then, I will actually simulate and roleplay the meeting out loud with the AI inside ChatGPT.
It is absolutely incredibleâand slightly creepyâhow many times the AI has thrown a unique question back at me that I hadn't thought of, only for the client to ask me that exact same question during the real meeting later that day. Having that kind of preparation is like having access to a domain expert 24/7.
I might be a bit biased because I've always been adventurous enough to try things I have no initial idea about. But AI takes that trait to the next level. Even if you lack deep domain expertise in a specific area, you know you have this interactive copilot or intern that can instantly get you up to speed. That shifts your workplace behavior significantly.
Anyone who integrates AI into their daily workflow can relate to this: you can confidently say "yes" to a lot more opportunities. If you have the agency and initiative, you know you can figure things out on the fly. For example, if I have a highly technical sales meeting but our developers aren't available to join the callâwhich happened a lot when I worked at a data consulting firmâI can still prepare beautifully on my own. As a salesperson, you can now spin up functional demos of what a client wants without writing a single line of code. It hasn't happened broadly across the entire corporate landscape yet, but the initial projects we are deploying at Warp show exactly where the world is heading.
Shifting from Static Pitching to Real-Time Client Co-Creation (19:56)
Brad Eather: From a sales perspective, I always tell people entering the industry to relentlessly develop their soft skills. AI doesn't force you to niche down; instead, it gives you highly contextualized information. If you have the initiative to find complex data and distill it in a way that someone else can easily digest, that is far more beneficial for a modern salesperson than having narrow technical know-how. It makes communication simpler. Instead of spending hours struggling to get a developer to explain technical nuances because you aren't speaking on the same plane, you can ask the AI exactly what you want and get clear bullet points. If the answer isn't perfect, you just iterate across a couple of prompts.
Gaurav: 100%. Another major structural shift happening right now is the transition from traditional pitching to active client co-creation. I was actually talking to a friend about this recently. Nowadays, everyone brings AI note-takers into virtual meetings to handle real-time transcription.
Imagine you are a salesperson in a live discovery session. The AI note-taker is transcribing the client's requirements in real time. You can feed that live text directly into an AI-driven design or wireframing tool to mock up visual designs and demos right there in the meeting. Instead of a one-sided pitch, you are inviting the client to actively co-create the solution with you on the spot.
What Is an AI Agent vs Traditional Business Automation? (22:19)
Brad Eather: Let's slow down and explicitly explain what an agent is, because once we define that concept, this entire conversation will make a lot more sense. What exactly is an AI agent?
Gaurav: That is a very interesting question because if you ask ten different people right now, you will get ten different answers. In my definition, it's best to look at the contrast between traditional automation and an AI agent.
Traditional automation relies on completely predictable, rigid parameters. It operates on boxed assumptions with a strictly limited number of predefined paths, meaning you have to hard-code a set solution for every single step.
With an AI agent, you don't need to hard-code every single scenario. An agent can evaluate a situation dynamically and decide on the fly whether it needs to execute path X or path Y to achieve the overarching goal. At a basic level, you have single agents where you connect an AI model to carry out a specific operational task for you, such as automatically updating client records in HubSpot based on the contextual notes of a phone call. They move from passive AIâlike standard ChatGPT where you type a prompt and it passively hands you a responseâto active AI that autonomously executes sets of tasks.
Deploying Multi-AI Agents in Business Workflows (25:06)
Brad Eather: My understanding for the audience is that while a standard Large Language Model answers prompts broadly, an agent contextualizes information within set parameters. We dictate exactly what internal data it can access and what specific task it needs to fulfill.
People are likely familiar with Microsoft Copilot generating a transcript and automated meeting summary at the end of a call. That is a basic agent built for a specific purpose. But in an enterprise context, we are talking about taking every single piece of company IPâfrom the sales funnel to technical documentation and legal requirementsâand connecting them. We can create a chain of specialized agents that act like a master employee with 50 years of experience, passing data seamlessly from one agent to the next to get an optimal business outcome.
Gaurav: Exactly. You don't always have to hook an agent into your own internal knowledge base, but for any established business, that is where the true competitive value lies. It involves securely connecting your centralized company data repositoriesâlike SharePointâto an LLM within protected cloud environments like Azure or AWS. This ensures your data remains completely private and never leaks to third parties. Cybersecurity is the number one question I get from clients, and hosting it securely within enterprise cloud services eliminates that issue entirely.
From there, businesses want to connect these agents directly to their Salesforce or HubSpot pipelines. Over the last 20 years, the corporate world has become entirely "SaaS-ified". There is a separate SaaS application for absolutely everything, leaving enterprise data scattered across 20 different disconnected systems. For AI agents to work effectively and unlock new operational capabilities, you need to bring that data into a central repository. Once centralized, you can blend scattered qualitative data together to uncover insights, notice hidden patterns, and generate predictive information that was previously impossible to find.
Preparing Clean Data Structures and Machine-Friendly Formatting (27:41)
Brad Eather: So if a small business is considering this leap, a massive element of change management involves organizing their existing information in a way that an AI can actually comprehend.
Gaurav: That is easily one of the most important points. The old rule of traditional data modeling still applies: garbage in, garbage out. While modern frontier models are becoming robust enough to navigate messy file structures, maintaining clean data hygiene is still a vital practice.
Most organizations are guilty of saving files with names like "Document 1" and having 20 different versions of the same file floating around with random edits from different team members. That completely confuses an AI solution. You need clear data management policies.
I tell everyone to start intentionally writing documentation not just for human eyes, but specifically formatted for machines. For example, Markdown (.md) files work beautifully for feeding structured instructions to an AI. Traditionally, if you built a complex Excel spreadsheet, you wouldn't write a text paragraph explaining every formula or your underlying financial reasoning. Now, you absolutely should. Write down the plain-text rationale behind your data models, because when you feed that spreadsheet into an AI for analysis, the model has the vital context it needs to deliver accurate answers. That is a fundamental behavioral shift in how we document business processes.
The Future of Corporate Knowledge Retention in Business (31:01)
Brad Eather: That makes total sense. Right now, AI is incredibly proficient at analyzing specific types of modal informationâreading text, seeing images, and hearing audio. AI reads incredibly well, but up until recently, its ability to truly see, contextualize, and understand an image was quite limited compared to text.
Gaurav: It depends, but modern vision capabilities are actually quite good. Look at the real-time video features inside ChatGPT where you can turn on your camera, show the AI your surroundings, and it accurately describes exactly what is around you on the fly. I'm surprised that feature hasn't received more mainstream hype, because its enterprise utility is massive.
If you are a manufacturer with a junior technician out in the field, they can simply point their camera at a machine fault. Because the agent is connected directly to your centralized knowledge base, it acts like a senior engineer standing right next to them, diagnosing the issue and explaining exactly how to fix it without needing to pull another human off the floor.
This feeds back into the widespread issue of knowledge retention. Across almost every industry, you have veteran employees who have been with the business for 20 or 30 years. They are reaching retirement age, and when they walk out the door, the business traditionally loses decades of accumulated knowledge because it was never documented. We are using AI to solve this by explicitly quantifying that undocumented brain trust. We extract that knowledge through guided interviews and documentation, feed it directly into a secured agent, and ensure that invaluable institutional knowledge remains a permanent piece of company IP long after the individual leaves.
Redefining Creative Problem-Solving with AI Micro-Prototypes (38:38)
Brad Eather: If you are a young junior entering the professional workforce or a student at university right now, this landscape represents a historic opportunity. If you develop a deep familiarity with these tools, you can completely flip the traditional hiring and interview process on its head. Instead of just showing up with a resume, you can research a target company's problems, use low-code AI tools to patch together a functional prototype solution, and present it directly to a senior stakeholder.
Even if the prototype isn't perfect, showing that level of strategic initiative forces leaders to take you seriously. It completely skips the traditional boundaries of entry-level devalued work. Given everything you've experienced at the intersection of business and technology, what is your personal definition of creativity?
Gaurav: It's a really interesting question, and I love answering it because I have absolutely zero traditional, artistic creative talent. For me, creativity in the modern age means something entirely different. We live in an era of absolute information abundance where knowledge is completely commoditized and accessible everywhere.
I view creativity as the ability to take odd-shaped pieces of Lego scattered across the internet and fit them together to build a unique solution to a problem. It's about looking at things from an unconventional angle, noticing that a certain piece doesn't traditionally fit, but realizing you can shave off a tiny edge over here to seamlessly merge it with something else.
In a space as chaotic as AI, if you only look at things through a rigid, conventional lens, you will get overwhelmed instantly. True creativity is simply making sense of these fast-moving, disparate pieces and building functional things that solve real-world human problems.
Brad Eather: I love that definition. It's about flipping rigid thinking on its head, shaving those metaphorical Lego pieces, and shaping them to suit a distinct, impactful purpose. Gaurav, it has been an absolute pleasure having you on the podcast. Where can people find you if they want to reach out?
Gaurav: You can easily find me on LinkedIn under my full name, Gaurav Dev Sharma. Alternatively, feel free to email me directly at [email protected]. I am always incredibly happy to connect, grab a coffee, and nerd out about anything and everything related to AI strategy.
Brad Eather: Perfect. Well, to everyone listening out there, thank you for tuning in to the Sellings Creative Podcast. If you enjoyed this episode, remember to subscribe for more, and in the meantime, happy selling.
