From Bronx to Boardroom: Shana Simmons’ Rise to Zendesk Chief Legal Officer

“I am not just a Black woman in tech. I am a Black woman who grew up incredibly poor in the Bronx.” Shana Simmons has been the Chief Legal Officer of Zendesk since 2024. Her journey from the Bronx to California, via law firms in London and New York, from Google Cloud to Everlaw, and finally to Zendesk, represents a unique path that she refuses to consider as the standard. At Denver for Relate, she opened up to BDM about her career, her role as CLO in the era of AI, and issues around the legal responsibility of AI agents.

From Cleary Gottlieb in New York and London to Google Cloud, Everlaw, and now Zendesk, what drove your transition from financial law to tech law, and how do these experiences connect?

I wish I could say my career has been meticulously planned with a clear thread linking each role, always aiming to be a Chief Legal Officer or a law firm partner. That’s not the case. My time at Cleary in New York and later in London was filled with amazing work and even more amazing people. At that time, Cleary was renowned for its global practice and the profit-sharing model among partners, which was based on seniority rather than client portfolios, thus prioritizing client interests. That was the environment I thrived in.

My move to London marked the first personal decision in my career: getting engaged to who is now my husband, a firefighter in California. This prompted a kind of reset to ‘base camp’. Sometimes when climbing towards a peak, it’s necessary to return to base camp to acclimatize. I was fortunate to secure a position at Google Cloud. Starting not as in-house counsel but as head of contracts meant relearning what clients need because I was the lawyer negotiating directly with them. Ultimately, you sign a contract, and a contract is a partnership.

The common thread in all my decisions has been people and their issues.

Shana Simmons

Chief Legal Officer, Zendesk

The common thread in all my decisions has been people and their issues. At Everlaw, we were building AI technology for lawyers like me. And that’s where I find myself today at Zendesk, not out of ambition to be the CLO of a tech AI company, but rather by building the right teams around me.

Becoming Chief Legal Officer at Everlaw, an e-discovery solutions provider for lawyers, immersed you early in AI applied to law. What did this experience teach you that is now valuable at Zendesk?

Two things. First, my experience at Google Cloud made it easy for me to understand the customer’s perspective because we were the underdogs. Google Cloud was not even in the top three at that time. We had to build differently for banks, for healthcare, and for retail. I had to understand the regulations of all these clients worldwide. This is how we build today at Zendesk, and this is how my legal team operates. My lawyers are involved from the beginning of product design.

Second, there have been instances where AI produced non-existent case law, incorrect citations, where lawyers lost a lot of credibility. It’s crucial, as a lawyer, to be able to trace back to the source of a precedent: whether it’s a court decision, a judge’s opinion, or a contract. That’s where I realized the power of Retrieval Augmented Generation (RAG). Some might say it’s old technology, but I still love RAG because as a lawyer, I can ask: where did this information come from? And the technology takes me right there. It was the first time I saw how useful RAG could be for lawyers.

Over the fifteen years of your career, what has changed for women and minorities in American tech?

First, let me talk about myself. I am from the Bronx. So, I am not just a Black woman in tech; I am a Black woman who grew up incredibly poor in the Bronx, in a one-bedroom apartment for five. What has changed is that someone like me can lead the legal department of a global company. I do not take this for granted. It is both an honor and a privilege. This means understanding the clients and their environment, ensuring that everyone in my teams feels included and can thrive is extremely important to me.

I am always very nervous about saying, ‘If I could do it, so can you.’ Because the truth is, not everyone has had the same opportunities I had.

Shana Simmons

Chief Legal Officer, Zendesk

Now, addressing the question about women and minorities, it seems that opportunities are not the same for everyone. It has been much harder for some people, and it appears much easier for others. But behind what seems easier, there are a lot of sacrifices. So, I am always very nervous about saying, ‘If I could do it, so can you.’ Because the truth is, not everyone has had the same opportunities I had. What got me out of the Bronx was the chance to go to a boarding school young, with a full scholarship. All these opportunities then built on each other.

You publicly advocate for diversity and inclusion. Practically, what does this change in your work as CLO? And what would you say to a young lawyer wanting to follow in your footsteps?

I would humbly tell a young lawyer considering this path, or any path, to be themselves. To stay true to who they are. Even though I have some experience now, I still find myself questioning. Should I present myself like this male colleague? I like to dress in colors, should I change that? The truth is, we all do better when we present ourselves fully because what we bring forward is exactly the voices of our clients.

We have 80,000 clients in more than 130 countries serving people like you and me. If we build just for what I think is the archetype of an executive, and not for someone who might share my perspective, I betray both the company and our clients. So it’s super important to be authentic. But also, you need to learn your craft. This is easy advice to give, but with AI today, you need to learn your craft, the technology you use, and you need to understand the industries and sectors you support. And put the client first.

You described AI Trust around three pillars: governance, control, consequence management. You emphasized that trust has become the biggest hurdle to AI adoption in businesses. What fills your days as CLO when your company positions itself as a trust-first provider? What consumes most of your time?

I wouldn’t say that trust is more important than product capability, but it is deeply embedded in it. Now, no two days are the same. I often say that as a leader, you must be incredibly agile. I might spend one moment thinking about what the legal department will look like in 2030 – how to orchestrate a department where humans and AI work together autonomously, while keeping my teams fully engaged in building what I need – to very hands-on topics.

At the company level, Zendesk has been transforming for the past three years. I need to think about the global regulatory requirements that accompany what we build, how we build it, how we think about our collaborators around the world. The rules are not the same in Australia, France, Spain, or California. I have to consider all this globally, at any given moment.

The legal department has been involved from the start. It would have been easy to be the ‘no’ department, but I also think about how we facilitate growth while managing risk. No two days are alike. Sometimes, we just react to the worst thing that happened in the company that day.

You say that one of the biggest barriers to AI adoption is trust. Yet, companies often adopt AI solutions from providers before considering data and governance issues. How do you explain this discrepancy?

It’s quite surprising. I think that if these companies truly put their customers first, this question should come before adoption, even before purchase. You need to have certifications, but that’s not all. Many people buy quick solutions without even wondering if the product works.

I continue to attend product demos from other publishers trying to sell us things. And I often think, ‘This is nonsense.’ Someone built it quickly because they could, but it’s not what we need. It won’t meet the goals we need to meet. And they don’t even think about the data. When I ask simple questions like, ‘How do you protect the data? How do you think about my lawyers who need to retain it?’ they can’t answer.

When an AI agent makes a decision for a client, who is legally responsible when something goes wrong? The software publisher, the client company that deployed the agent, or the provider of the underlying model? Are there any legal precedents?

This is a topic my peers and I talk about a lot. There’s no consensus yet, but the general idea emerging is that ‘liability follows control.’ The more you control, the more responsibility you bear. This is what you want from your suppliers as a company, and what you want to assume for yourselves. Conversely, if you are responsible, you must be in control.

Our solutions are designed to be easy. Companies should not have to retrain our models or build a lot of safeguards. But they remain in control of their customer journey. This is a very important question.

But I’d like to go further. Liability comes up in all contract negotiations. It’s very easy, when a client wants to make a deal and they have little to lose, to say ‘I am responsible for everything.’ That doesn’t help me at all if you have damaged my brand because you didn’t meet your commitments to my clients. When I interview commercial lawyers for my team, I ask them, ‘When you talk with a client, what is the most important part of the contract?’ If they answer ‘liability and indemnification,’ I don’t hire them. That’s a lawyer’s answer. It’s easy to say we negotiate liability, that we won’t indemnify everything. What’s hard is understanding how data is used, how data is shared, how you meet your service commitments. That’s the conversation that’s harder to have, and that’s where we should focus.

You recently wrote in Verdict that ‘the era of optional AI transparency is over,’ emphasizing the need for explainability and the ability to audit an agent’s reasoning. Currently, where does the balance lie between marketing promises and emerging regulatory obligations?

First, the reason why I say this: in a year, when we talk again, nobody will be thinking about transparency because AI will be embedded in all our systems. And I believe we must hold AI agents to a higher standard of explainability than what we might expect from each other as humans. It’s crucial to be able to trace the logic and understand it.

That said, complete explainability of large language models is still out of reach. Researchers still don’t really know why an LLM says X or Y. But the explainability of an agent-based system like Zendesk’s is much more advanced. You can look at interaction logs. You can see that an agent approved a refund because it looked at the customer’s return conditions and applied the company’s return policy. Or you can understand where it went wrong. I want to be cautious: I’m not saying the entire industry is there yet. I’m talking about a smaller subset of the autonomous AI landscape.