AI-Native Marketer

EDITOR’S NOTE

One marketer. More than three years to ChatGPT. A real answer to the question everyone is asking.

Dear Nanobits Readers,

The question underneath every AI conversation right now is not, "What can AI do?" Most people have already seen the demos. The real question is more personal: if AI can do this, what do I actually do?

This edition is a continuation of the AI-native functional experts series on Nanobits. We are talking to functional leaders who have genuinely figured out how to work with AI at a high level. Not people who read about it. People who built workflows, shipped things, changed how their teams operate, and can tell you exactly what broke and what held.

Somya Sinha and I (Monalisa Sethi) sat down to have exactly this conversation. Somya runs AI with Somya on Substack, and this edition is a collaboration between her newsletter and Nanobits.

By the end of this piece, you will know the exact marketing tech stack, what good AI-native marketers do manually versus with AI, and the three things we believe will keep any business leader relevant.

If you are a marketer trying to figure out what to automate and what to keep, a business leader rethinking your GTM stack, or anyone asking whether they still have a place in an AI-heavy workflow, this one is for you.

Where did my AI journey start?

"When ChatGPT-3 launched, I was at a series-B fintech start-up. Our leaders encouraged us to use GenAI for copywriting.

In mid-2023, I took a career break and did a copy traineeship with Terribly Tiny Tales. I started using AI for a lot of writing and quickly saw that purely AI-generated writing looks soulless. Too many emojis. Dead giveaways everywhere.

So I started building guidelines for writing with AI while keeping my own voice. Blake Stockton's blog on avoiding AI slop was useful here.

I also love reading about technology. By the end of 2023, I had subscribed to 25 AI newsletters. Allie Miller, Karen Hao, Rundown, Alpha Signal, and others. My social media feed filled up with interesting AI content, and that kept compounding."

How did I actually learn AI?

"I did not take a single AI course. Two things worked for me.”

The first was learning by writing about AI. Together with Geetika Mehta, I started Nanobits in early 2024. We noticed a gap in Indian AI news coverage and went after it - covering Indian news, tool teardowns, and AI concepts. It pushed me to run deep experiments and write about firsthand learning rather than just consume content.

The newsletter built my credibility as an AI expert and directly led to AI consulting gigs. In one gig, I did ghostwriting for an entire team of CXOs by creating a Custom GPT for each leader, which brought down copy revisions from 'n' rounds to just one.

The second was learning by implementing AI at work. When I joined an industrial saas series-B company as a marketing leader, I immediately got down to researching how to strengthen our GTM stack and introduced new tools for the team. Even now, before starting my next role, I'm already researching how to use Claude Code for GTM."

My marketing stack. What do I do manually vs. with AI?

"The most important thing I can say upfront: when machines create everything, the last scarce resource might be humans. If you do not know what questions to ask, no tool will give you answers on its own."

Here is how my tech stack breaks down across the full marketing function.

Content Creation: Gemini + Claude

Research comes first. For all third-party research, I use Gemini 3 Pro in Deep Research mode. This includes deep dives on people, including profiling 50 webinar attendees ahead of an event.

The first draft is always manual. That is a firm rule.

For refining writing, including newsletters, I use Claude Projects. I enter all the relevant context myself and tweak the memory manually. My usage splits into two modes. Claude on the web is for open-ended thinking and brainstorming. Cowork handles repeatable tasks, such as repurposing a weekly newsletter into posts for Reddit, X, and LinkedIn, or analyzing 20 customer calls, grouping them by theme, and sending those summaries to the sales team on Slack.

I am still working out where Claude Code fits. One experiment I am considering: building a GTMBuddy-style tool to track whether the GTM team is actually using the decks, one-pagers, and assets that marketing creates for them. My view on tool adoption is deliberate. "It is very important not to chase every shiny tool that gets launched. Explore, no doubt, but mindfully."

For instance, platforms like SEMRush, Ahrefs, and WebFX offer native AI features to help with content refinement and distribution. However, their impact on revenue has been incremental so far.

Because everyone is publishing AI-written articles right now, the problem is two-fold. Google penalizes content published at that pace, and AI-written articles carry no authority on their own.

The hard part, and the part very few people do today, is pulling real insight from prospect and client call recordings and getting input from industry voices for writing a good thought leadership piece. That gap is what sets a good marketer apart.

The deeper issue goes beyond AI-generated content. 

There is also a structural problem in how most teams approach content. One blog, one white paper, and one ebook, each treated as a standalone output. Good marketing leaders are moving away from that. They think like a content studio, where everything connects, addresses real audience needs, and fits into a unified view. If you cannot answer why you are writing a particular blog and how it’s going to be distributed, the strategy is fragmented. Start with "why" and build from there.

It comes down to what you are optimizing for. SEO rankings, brand building, or pipeline? An AI-generated thought leadership piece is not thought leadership. You can game SEO and ranking, but if the piece is not unique, no one will read it.

GTM Launch Playbook: Human Judgement + Claude + Notion MCP + Asana MCP

The structure of a launch plan comes from experience. A D-minus-45 to D-zero Gantt chart, the channel-by-channel plan, and the stage-wise task list: these are built by a human. "If you ask someone with 15 years of marketing experience to list everything needed for a product launch, they will have no problem. I am yet to see a junior or mid-level resource produce a solid launch list and execute it flawlessly with AI alone."

Once the plan is in place, Claude, Notion MCP, and Asana MCP take over admin. Tasks push to Asana automatically. Stakeholders get reminders. Overdue items surface without anyone having to chase them manually. "The senior marketer running the launch has a full plate. Offloading the admin frees them to do the actual marketing."

Analytics: Salesforce + HubSpot + Mixpanel

For sales and partnership-led motions, Salesforce and HubSpot mostly cover what the team needs. For product-led growth, Mixpanel predominantly handles the analytics.

HubSpot now has an MCP server. Connect it to Claude and ask questions in plain language: "What was deal velocity last quarter, and how does it compare to the historical average?" No SQL required.

The most important point to remember that applies to every tool in the stack is that "humans operate the tools, not the other way around. If the marketer does not know what questions to ask, no tool will surface insights on its own."

Signals and Prospecting: Clay + ZoomInfo

For lead research and scoring, I use Clay and ZoomInfo, supplying third-party intent signals at the contact level. Clay is the stronger tool. It can pull details from, say, the 10-K reports of prospective customers, allowing the team to ask specific questions: "Which prospect has a stated cost-cutting goal as part of their digital transformation?" ZoomInfo and Clay both depend on Bombora for that contact-level signal layer.

The business logic for scoring stays manual. Deciding which activities count more and which signals to weight is a human call.

Sales Engagement: Gong + Outreach + LinkedIn Navigator + HubSpot Sales Hub

The standout use case here is Gong. If the team is in early-stage talks with a new prospect and already has a similar client in the portfolio, they pull that client's account from Gong and use the native AI feature to surface relevant insights. Those insights feed directly into the new prospect conversation.

If you are using competitive intelligence tools like Klue: "These cost $40,000 to $60,000 a year. I would much rather hire a person to do CI than buy a tool. Most martech tools average $25,000 to $30,000 annually. Vendor sprawl also creates fragmented data. Building any AI layer on top of fragmented data becomes a serious problem."

Sales Engagement: Gong + Outreach + LinkedIn Navigator + HubSpot Sales Hub

The standout use case here is Gong. If the team is in early-stage talks with a new prospect and already has a similar client in the portfolio, they pull that client's account from Gong and use the native AI feature to surface relevant insights. Those insights feed directly into the new prospect conversation.

If you are using competitive intelligence tools like Klue: "These cost $40,000 to $60,000 a year. I would much rather hire a person to do CI than buy a tool. Most martech tools average $25,000 to $30,000 annually. Vendor sprawl also creates fragmented data. Building any AI layer on top of fragmented data becomes a serious problem."

What does the future of the marketing workforce look like?

The role of a marketer is not disappearing. Its shape is changing fast.

Teams will likely be smaller. A marketing leader hiring today does not need as many writers and editors as they did three years ago. AI handles a real portion of what used to require headcount. But that does not make the marketer redundant. It makes the marketer's judgment more central, not less.

Deloitte describes this shift as the rise of "superjobs," roles that combine work from multiple traditional positions, using technology to expand what one person can own and deliver. The work that remains for humans after automation is more interpretive and strategy-driven: problem-solving, data interpretation, and knowing what questions to ask in the first place. In marketing, that translates directly.

The marketer of 2026 is part strategist, part analyst, part editor, and part workflow architect. One person, broader scope.

The job description for a marketer is also changing, and for the better.

A portfolio of standalone content pieces is no longer an effective signal. What stands out when hiring is evidence of content projects, not content outputs. Industry reports such as Amagi's FAST Report and Mailmodo's State of Email Marketing are good examples. These are not one-off blog posts. They are content IPs, built to repeat, compound, and own a conversation in the market over time. Has the candidate worked on something like this? Have they thought about distribution, not just creation?

The parameters for hiring a marketer have shifted. The right candidate today has three things: primary research instincts, distribution experience, and an understanding of how AI fits into their workflow without replacing their thinking. Everyone involved in hiring and in being hired needs to raise their game accordingly.

The bottom line: marketing teams may get leaner, but the marketers who remain will carry more responsibility, not less.

End Note: How does one stay relevant as a marketer in the age of AI?

Three things. I think about these a lot.

Build something and sell it. Anyone can build now with tools like Lovable. What sets people apart is taking something to users and getting them to pay or engage. When I started building this newsletter during my career break, it always opened up deeper conversations in interviews. People were skeptical, but the newsletter proved I knew my subject well. When I was interviewed for my last role, the AI workflows I had built in Make.com and n8n.io got the hiring team's attention. Have something to show, not just something to say.

Write your first drafts yourself. AI works well for research, refinement, and distribution. But the first draft has to come from you. Most people prompt their way from zero to a finished piece today. That is exactly why original thinking is getting rarer and more valuable. The angle, the opinion, and the insight: those have to be yours.

Build your tribal knowledge. AI has no access to what you absorb from your managers, leaders, from customers, or from being in the room when a deal closes or falls apart. The more you observe and learn from the people around you, the stronger your real-world judgement becomes. That is the part no model can replicate.

A note from Somya about this conversation: 

This series will continue. Next up, we will talk to leaders in sales, consulting, design, finance, HR, etc., all asking the same question. Subscribe so you do not miss it. And if you are someone doing interesting AI work, reply to this email. We would like to feature you.