In the rapidly evolving world of AI and machine learning, businesses are trying to keep up with ever-changing needs. However, there is only so much experienced talent to go around. A company called Tribe AI focuses exclusively on staffing and training this tsunami of needs across the tech talent ecosystem. Through Tribe AI, businesses get access to the top 1% of AI experts from companies like OpenAI, Google, and other leading innovators.
In this Q&A session, co-founder, and CEO Jaclyn Rice Nelson dives into her journey from Google to founding Tribe AI, the role of her company within the tech talent ecosystem, and its impact on businesses. As AI enters a new era, Jaclyn envisions Tribe AI becoming the orchestration layer that empowers every company to become an AI-driven entity, making the technology’s benefits accessible to all.
Gary Drenik: Tell us about your background at Google and the journey that led you to start Tribe AI.
Jaclyn Rice Nelson: I spent nearly eight years at Google incubating products and working with growth-stage companies at CapitalG, Alphabet’s late-stage venture fund. I took two important things away from these experiences. The first was a conviction that AI was going to be the defining technical innovation of our lifetime. At Google, there wasn’t a single part of the company that wasn’t run with AI or ML. The second takeaway was a belief in the power of expert networks.
I helped incubate Google Helpouts – an expert marketplace within Google. Ultimately it struggled to get traction, as startups within tech behemoths often do, but the kernel of the idea stuck with me. Then, at CapitalG, I saw how even highly successful growth-stage companies like Airbnb and Stripe were struggling to hire for data science and machine learning roles and we built our own expert network of top Google AI talent to advise them. It felt like there was a very clear problem and solution – better utilization of existing technical talent to bring the power of AI to all companies, not just Google.
I teamed up with my co-founder, Noah Gale, around a belief in a massive opportunity in AI, a shared understanding of what technical excellence looks like, and a conviction that a network-based approach to talent would be the best way to accelerate AI in the market.
And, in the last few years, the market has really supported our thesis. Covid-19 followed by the Great Resignation shook some of the most talented engineers free from big tech and we’ve been able to turn that into a strategic advantage for our customers.
Drenik: Explain the role of Tribe AI within the tech talent ecosystem today and highlight your progress with the company to date. Include examples to illustrate how Tribe AI works.
Nelson: When we started Tribe in 2019, we believed every company would need to become an AI company. Now, thanks to Generative AI, every business wants to become an AI company overnight. But most companies are still building their data foundation, grappling with IP and privacy concerns, and struggling to source the right data and AI talent. This is especially true for mid-market and non-tech enterprise companies, who have tremendous data assets but lack the technical capabilities to extract the value from their data.
Tribe is exclusively focused on AI and machine learning and has been since our founding. We’ve built hundreds of AI products and have 400+ top AI engineers on our staff. We use this base of expertise to help companies build their AI roadmap, scope, and build PoCs, automate time consuming, manual processes and scale with AI instead of headcount.
ChatGPT is profound because it captured the imagination of the world and made AI tangible. According to a recent Prosper Insights & Analytics survey , 27.7% of people over 18 in the U.S. know about ChatGPT and have already used it or are excited to experience how it works. Another 31.9% are familiar with the tool, and just don’t know how to use it yet. But – it’s the more boring, less sexy applications, where I see the most value for businesses and where I get the most excited.
For example, we worked with a large food distribution company that was losing market share due to a highly manual bid development process. Tribe built and delivered an AI-powered software solution that provided advanced analytics to automate part of the process, increasing the efficiency of the bid team by 50% – this means an additional 60 bids or up to $40M in revenue over five years.
Another example, we worked with a public medical device company looking to predict and reduce customer churn. They had data and built a few data models but didn’t have the lent to operationalize them. Within 12 weeks, Tribe had launched and automated a successful churn model that also reduced engineering spend by 20%.
This is where the AI hype is warranted. There is value that can be created today for every company, and Tribe is here to help companies move from AI aspirations to reality.
Drenik: You also run a VC firm, Coalition Operators, what led to that decision? How does your work in venture support your work with Tribe AI? Are there any overlaps that make you better at each job? What about challenges?
Nelson: When I left Venture Capital to start Tribe it was the landscape view, the ability to see patterns across companies that I missed the most. At Tribe we sit in a really interesting spot in the AI ecosystem that gives me a unique advantage as an investor and made it too hard to sit on the sidelines. Tribe partners with the top AI platforms and hyperscalers like OpenAI and AWS, leading enterprises who are building AI solutions and the best AI talent in our industry. One of the most unique things about Tribe is our view of the market and our deep understanding of needs and gaps across these groups.
This vantage point allows us to build Tribe in a way that addresses the market needs and accelerates our growth as a core node in the AI ecosystem. So much of my work at Tribe is about bringing brilliant technologists together to solve customer needs. Investing is just another way for me to do that and help accelerate the overall market adoption of AI in the process.
As I began investing in early-stage companies, I didn’t just want to invest for my own financial gain, I wanted to build a fund that broadened access to the most talented women in tech, my smartest peers who were too busy in their day jobs to invest but whose expertise I knew would be hugely valuable to founders. Similar to Tribe’s talent network model, at Coalition we’ve built a network of top women operators across functions. In addition to investing in companies directly, we also help founders tap into this talent pool to accelerate their businesses and diversify their cap tables in the process.
The synergy between my work on Tribe and Coalition is massive. We’ve had Coalition Network members take jobs at AI companies who have then become partners with Tribe. Our fund has hired Tribe members to do due diligence on companies where we know they have expertise. I’ve invested into companies through Coalition that have then become Tribe customers. And I’m a better CEO and partner to our customers at Tribe and the companies I invest in at Coalition because I have so much context on the AI market as a whole, not just the role Tribe plays in it. Between Tribe and Coalition, I feel like I’m finally doing the work I was meant to do.
Drenik: We’ve been particularly interested in the topic of ethics in AI. How can a company like Tribe ensure AI is being built with ethics in mind? How does “talent” come into the ethical equation?
Nelson: I think talent is hugely important when it comes to building ethical, unbiased AI systems. It’s no secret that there’s a diversity problem in tech, and it gets worse in specialized engineering areas like AI. There’s an opportunity and an obligation to be aware of the bias that can be present in these systems because of the people who build them and to foster a diverse group of builders to combat this.
For Tribe this means building a diverse network of engineers. And it means creating opportunities to have these diverse builders at the forefront of building products that could be biased, having them in regulatory discussions, and making sure that the incentives are there to build with ethics in mind. By allowing our customers to tap into a diverse network of talent we can ensure that together we’re building the type of solutions that can build powerful and fair AI systems.
Drenik: Another area we’ve been fascinated by is the data sources that feed AI models. As a company that gets to see firsthand how dozens of companies are training models, what best practices are cropping up to ensure it’s fair and equitable, while still being powerful and effective?
Nelson: I am also fascinated by the data sources that feed AI models. In a category with few moats, data just may be the most defensible asset.
Many large companies are sitting on tremendous data assets that are going unleveraged. At the same time, we’re seeing startups, that don’t typically have a data advantage, do extremely unscalable things like going door to door to collect their own proprietary data that can give them an unfair advantage. In both cases the key is to unlock the data you do have to be able to extract insights. The next step is to enrich that data with external data sets that can give you even more signal and predictive power.
When thinking through internal data collection and public data sets that you leverage, being thoughtful around how diverse that training data is and if it’s representative of the real-world population is critical to ensuring you will get results that represent your user base. That said, bias is often assumed to be purely a data problem, but it can also be presented in the problem definition. In this case, no amount of data balancing will fix bias if the criteria is inherently skewed against a particular group.
Once you are at a point of training or fine-tuning models, we recommend companies exclude any sensitive data, like names, not just for privacy reasons but also to avoid the model picking up any racial bias. Looking at data and privacy concerns from a consumer angle, the majority of people in the U.S. intentionally take steps to protect this. Based on data from Prosper Insights & Analytics, only 22.5% of people in the U.S. over 18 have not taken any steps (i.e., turned on private browsing or turned off mobile tracking) to protect their online privacy.
Measuring and evaluating results for bias is important to correct any issues as they arise. We recommend having a separate team conduct a review of the model and the results to better spot any model bias. Continuous monitoring and evaluation are important for maintaining a fair and effective model.
Drenik: AI is advancing at lightning speed. As AI enters a new era over the course of the next several years, what’s next for Tribe? What’s your vision for the company?
Nelson: I believe the future of AI is multi-model, multi-cloud, and customized and open source. Our ability to be agnostic across platforms, pull from the best of each model and customize where needed is a big part of our value to companies and will help us continue to drive outsized results. Right now, one of the biggest opportunities in AI is the need for services to help companies make their AI aspirations a reality, and this is what Tribe does best.
In addition to building products for other companies, we’ve also been building our own AI products to run Tribe and give our engineers more leverage on their time.
Over time I see Tribe as the orchestration layer to enable every company to become an AI company, which will likely be a mix of SaaS and tech-enabled services. The advances in AI are still so new and the underlying technology will continue to change. Our most defensible asset is our experience building AI products, deep expertise in the space and continued work at the cutting edge, which will enable us to build a generational AI company.
Drenik: Thanks for taking the time to discuss your background, AI, and the tech talent ecosystem. I look forward to seeing how Tribe AI continues to bridge the gap between businesses and AI talent from industry giants, fueling innovation and driving real-world impact.
Read the full article here
