Phil Tomlinson, SVP of TaskUs, oversees the company’s global offerings, including Trust & Safety, AI Services, Digital Customer Experience, and Risk & Response. With over 20 years of experience, he develops digital solutions and advocates for better mental health in the workplace.
TaskUs is a global outsourcing company that provides next-generation digital services and customer experience solutions for high-growth businesses. Specializing in Trust & Safety, AI Services, Digital Customer Experience, and Risk & Response, TaskUs helps companies scale efficiently while maintaining top-tier support and security. With a focus on innovation, agility, and employee well-being, TaskUs partners with some of the world’s leading brands to deliver exceptional results.
TaskUs emphasizes a balance between technological innovation and human-centric AI. How does this philosophy influence the development and implementation of your AI solutions across industries?
At TaskUs, we believe AI should augment and empower human expertise rather than replace it. So we design and deploy our AI solutions for that interaction. While we design to automate, we don’t seek to automate an entire workflow. We understand where in the process human expertise is most valuable, then integrate and optimize the AI to benefit from that expertise. This serves to improve the ultimate result, reducing the number of false positives in fraud detection, for example, when compared with fully automated approaches.
We find this integration approach that anticipates entire workflows enhances human decision-making, leading to smarter, more effective solutions, whether it’s in customer support, content moderation, or risk assessment, etc.
Can you elaborate on how TaskUs leverages its proprietary platform, TaskGPT, to enhance both productivity and customer experiences?
TaskGPT is an AI platform that uses GenAI to boost productivity and improve customer experiences. It automates routine tasks so that human agents can handle more complex issues. The platform speeds up workflows and helps agents provide faster, more accurate responses.
TaskGPT helps agents retrieve information and make smart suggestions in real-time, which makes customer interactions smoother and more efficient. The demo on our website shows how the platform can process queries instantly, deliver useful insights, and personalize responses—making customer service faster and more effective. By blending AI-driven automation with human oversight, TaskGPT helps scale operations without sacrificing quality, leading to better customer satisfaction and quicker resolutions.
What role does TaskUs play in helping clients navigate the challenges of deploying AI responsibly while ensuring transparency and ethical considerations?
We develop AI governance frameworks that focus on fairness, accountability, and transparency in decision-making. Our approach includes using diverse training data to help mitigate bias and ensure AI models align with societal expectations. We also make sure AI systems are explainable and their decisions can be easily understood to provide full transparency. As AI continues to evolve, we will keep integrating human oversight and risk management methods to make sure AI deployments don’t just comply with regulatory standards, but are both safe and ethical.
Agentic AI is poised to revolutionize industries with its action-oriented capabilities. What are some of the most promising applications of Agentic AI that TaskUs is currently exploring or supporting?
Agentic AI shifts AI capability from passive responses to proactive, decision-making agents by combining GenAI’s content creation capabilities with some of the principles from more traditional programming, allowing us to empower the AI to act, but within a specified set of circumstances or guidelines. We’re exploring ways Agentic AI can anticipate customer needs and then take action. For example, by scheduling a meeting or issuing a refund without requiring human intervention. In fraud detection, Agentic AI can analyze real-time data and behavioral patterns to anticipate and prevent fraudulent activities before they occur. In healthcare, it can handle aspects of claims processing and appointment scheduling, boosting operational efficiency. These action-oriented capabilities help optimize workflows and deliver superior customer experiences.
How does TaskUs address the potential risks associated with Agentic AI, particularly in industries like fintech and social media?
We mitigate potential risks by taking multi-pronged approach. First, we build and deploy AI solutions aligned with ethical standards, especially in high-stakes industries like fintech and social media. Also we can often tap into an organization’s existing corpus of instructions, guidelines, and regulations that govern human decision-making. Agentic AI can tap those stores to inform its ability to act.
A strong part of our risk mitigation is not designing our AI solutions to automate away the human element – we incorporate strong human oversight into all our AI-driven processes. This ensures that any AI-driven decisions are monitored and can be adjusted or corrected when necessary.
Finally, we test and validate our AI solutions to make sure they operate within safe and effective parameters. We work closely with clients to build AI models that are both efficient and ethical. Our approach ensures that AI-driven actions are in line with regulatory standards and help mitigate risks like bias and misinformation.
AI safety is a growing concern for businesses. What safeguards or best practices has TaskUs implemented to ensure the responsible use of AI in critical applications?
We prioritize fairness and transparency by conducting thorough bias audits and fairness checks to make sure our AI models are equitable. We protect sensitive data with strong cybersecurity measures to reduce the risk of breaches. Explainability is a key part of our approach, making sure that AI decisions are understandable for both businesses and consumers. Human oversight in high-risk situations ensures the AI systems don’t make critical errors. By continuously monitoring AI models and working to meet industry standards, we ensure responsible AI deployment while maintaining trust and regulatory compliance.
With your extensive experience in Trust & Safety, how do you see AI evolving to tackle misinformation, content moderation, and other societal challenges?
TaskUs is a leader in using AI to help tackle misinformation and harmful content. We partner with companies developing the world’s most advanced GenAI models and have highly trained, deeply experienced dedicated teams that safeguard platforms and companies from emerging threats and bad actors. Our AI-driven content moderation solutions detect and tag hate speech and harmful content in real-time, allowing platforms to remove problematic material before it spreads. Our AI models consider both context and intent, helping to prevent over-censorship and ensuring that legitimate content remains accessible. Additionally, we support global platforms by using AI for scalable multilingual moderation, ensuring content policies are upheld across diverse languages and cultures.
How does TaskUs collaborate with clients to ensure that their AI deployments align with emerging standards for fairness, accountability, and transparency?
We work closely with clients to develop tailored AI governance frameworks that prioritize fairness, accountability, and transparency. We build responsible AI policies, conduct fairness audits to identify and address potential biases, and offer compliance training to ensure AI stakeholders are equipped with the knowledge they need. Additionally, we assist clients in complying with global regulations and ethical AI standards, providing the tools necessary for transparency and auditability. Our approach ensures that AI systems are not only effective but also responsible, supporting ethical deployments that meet both business goals and societal expectations.
The use of synthetic and hybrid data is gaining momentum. How does TaskUs assist clients in utilizing these data types to enhance model accuracy and reduce bias?
Synthetic and hybrid data can improve model accuracy and reduce bias. We use synthetic data to augment real-world datasets, ensuring that AI models have a more diverse and representative base for making predictions. By combining synthetic data with curated, real-world data, we enhance the quality of training datasets while addressing biases that may exist in some data sources. These methods also ensure privacy compliance, as synthetic data doesn’t involve personally identifiable information. By integrating these data types into training workflows, we are creating more inclusive, accurate, and secure AI models.
What are the biggest challenges you’ve encountered in introducing synthetic and hybrid data solutions, and how have you overcome them?
Because it’s “synthetic,” or fabricated, making synthetic data “real” can be a challenge. Maintaining compliance with privacy regulations and preventing the replication of biases in the synthetic data can also be challenging. We use real-world scenarios and advanced generative techniques to create synthetic data to make it as realistic as possible. We also collaborate with legal and compliance teams to ensure our synthetic data aligns with privacy regulations like GDPR and CCPA, minimizing privacy risks. Finally, we don’t automatically “trust” our synthetic data to be bias-free – we rigorously validate it to make sure it’s not biased, and we keep refining our models based on what we learn from the validation.
Multilingual AI has the potential to break language barriers. How is TaskUs enabling businesses to create more inclusive and culturally aware AI solutions?
We train our models on diverse linguistic datasets to enhance accuracy across multiple languages. This helps us improve speech-to-text and natural language processing (NLP) capabilities so we can provide real-time multilingual customer support and engagement with global customers. We also integrate cultural adaptations into our AI solutions to align content with local norms and sensitivities. This enables our AI solutions to resonate with a wider range of users and effectively breaks down language barriers.
Thank you for the great interview, readers who wish to learn more should visit TaskUs.
