Design Thinking for AI: Experiences Centered on People, Not Technology

The Risk of Prioritizing Technology Over the Business Problem

In today's corporate environment, many artificial intelligence initiatives start with a flawed premise: finding use cases to justify adopting a trendy tool. When business leaders focus solely on model capabilities or computing power, the result is often a collection of isolated proofs of concept that create frustration and deliver little operational impact.

For midsize and large organizations, the real challenge of artificial intelligence is not software availability, but effective adoption by employees and end customers. If an automated solution does not address a specific pain point or is confusing for the people who use it every day, the investment loses its purpose. Designing experiences around people aligns business objectives with users' actual needs, turning technology into a support tool rather than a bureaucratic obstacle.

• Projects lose traction when they solve problems that end users do not actually have.
• Complex interfaces or a lack of transparency in results create resistance to change.
• Technical teams become disconnected from the operational teams that manage the business day to day.

What Design Thinking Brings to Artificial Intelligence Projects

Design thinking is a structured approach to solving complex problems through a deep understanding of human needs. Applying this methodology to artificial intelligence development shifts a company's focus from algorithms to the context in which people interact. This means observing how employees make decisions, which information sources they typically rely on, and when they feel uncertain or overwhelmed by their workloads.

Unlike traditional software, artificial intelligence systems are probabilistic: they do not provide fixed answers, but estimates based on patterns. This variability demands a much more rigorous design process for managing user expectations. Design thinking helps define the acceptable level of accuracy for each task, how to present results clearly, and which controls must always remain in the hands of a human professional.

Key Stages in Developing User-Centered Solutions

Combining human-centered design with data science requires an iterative workflow in which ongoing feedback narrows the gap between expectations and operational reality. This process does not replace engineering best practices; it guides them to maximize their relevance.

• Empathize and observe: Study teams' workdays without assuming that automation is always the best answer. Identifying the actual operational pain point before writing a single line of code is essential.
• Define the problem clearly: Frame specific challenges where artificial intelligence offers a measurable advantage, ruling out processes where conventional business rules would be more efficient and predictable.
• Design interactions, not just models: Plan how users will submit data to the system and receive recommendations, considering visual alerts, brief explanations, and workflow shortcuts.
• Build low-fidelity prototypes: Simulate system responses before training the model to assess whether the information provided genuinely supports decision-making.
• Evaluate and adjust continuously: Monitor user behavior alongside model performance to address usability issues as soon as operating conditions change.

Designing for Trust: Explainability and Human Oversight

At midsize and large companies, adoption of predictive or generative models depends directly on the confidence they inspire. If a credit analyst, supply chain manager, or human resources specialist does not understand why a system suggests a particular action, two outcomes are common: outright rejection of the tool or blind acceptance of potentially flawed recommendations.

User-centered design addresses this problem by introducing contextual explainability. This means the interface does more than display a result: it shows the main factors that influenced that conclusion. A human-centered approach also ensures that critical decisions always include a review and approval step, allowing employees' judgment and accumulated experience to correct omissions that can occur in any probabilistic system.

• Visible uncertainty: Communicate honestly when the system has low confidence in an answer or calculation.
• Simple correction mechanisms: Allow users to edit, reject, or provide feedback on results directly within their workflows.
• Clear traceability: Provide a transparent record of the sources of the data used to generate a specific recommendation.

Privacy, Governance, and Ethics from the Start of Design

Experience design goes beyond a screen's appearance or a dashboard's usability; it also encompasses data protection and respect for user privacy. When integrating corporate information flows into advanced architectures, establishing strong standards from day one is essential to prevent unintended bias and leaks of operational data.

Organizations must recognize that regulations governing personal data processing and the responsible use of automated technologies are evolving rapidly in Latin America and the United States. Incorporating ethical considerations during the ideation phase helps mitigate operational and reputational risks. Because each industry and jurisdiction has its own regulatory requirements, companies should support these design efforts with qualified legal counsel to verify compliance with the legal frameworks that apply to their markets.

• Data minimization: Collect and process only the information strictly necessary to complete the defined task.
• Preventing operational bias: Assess whether the historical data the system learns from reflects unwanted distortions or disparities in decision-making.
• Consent and transparency: Clearly inform customers and employees when they are interacting with automated or machine-assisted processes.

The 1labs Approach: Strategic Consulting Focused on Tangible Results

At 1labs, we are an independent artificial intelligence consultancy focused on helping leaders of midsize and large organizations navigate technological complexity with a practical approach. Our role is to bridge analytical capabilities and strategic business objectives, avoiding redundant investments in tools that fail to gain internal traction.

As an independent, vendor-neutral team, 1labs does not sell software licenses or promote the interests of specific platforms. This gives us the freedom to put each organization's interests first, supporting executive committees and innovation teams as they structure viable, secure initiatives focused on strengthening their employees' capabilities.

Frequently asked questions