Empirical Evidence of AI-Enabled Accessibility in Digital Gastronomy: Development and Evaluation of the Receitas +Power Platform

1. Introduction

Domestic cooking remains a fundamental component of contemporary lifestyles, with nearly half of Portuguese citizens (48.6%) reporting that they prepare meals at home daily [1]. In comparison, studies conducted in the United States [2] indicate that approximately 49% of adults consistently cook dinner at home. Across Western, Southern, and Northern European countries [3,4,5], the average amounts to around 7.8% home-cooked meals per week, suggesting that domestic cooking continues to constitute a salient aspect of everyday life, albeit with notable variations in its frequency.In parallel, the ongoing digitization of gastronomy has fostered widespread enthusiasm for creating, experimenting with, and sharing culinary recipes across online platforms. However, this digital transition has introduced significant accessibility barriers, particularly affecting users with physical, sensory, or cognitive impairments [6,7].

Despite the abundance of online culinary content, a persistent digital divide remains between availability and effective accessibility. For individuals with motor, visual, auditory, or cognitive disabilities, interaction with existing recipe platforms often proves difficult or even exclusionary [8]. Globally, it is estimated that approximately 1.3 billion people, around 16% of the world’s population, live with some form of disability, with prevalence rates varying considerably across age groups, regions, health conditions, and socioeconomic contexts [9,10,11,12,13]. Nevertheless, most culinary websites and mobile applications continue to prioritize content sharing, recommendation algorithms, and personalization features over inclusive interaction and accessibility compliance [14,15]. Consequently, only a small number of culinary platforms currently adhere to established accessibility standards such as the Web Content Accessibility Guidelines (WCAG) issued by the World Wide Web Consortium (W3C) [16].

To address these shortcomings, the Receitas +Power project was developed as an inclusive technological framework aimed at reducing the accessibility gap within the culinary domain. The platform integrates AI components and user-centered design principles to ensure compliance with WCAG 2.2, enhancing perceivability, operability, understandability, and robustness [17,18].

By providing a web-based environment for recipe creation, visualization, management, and sharing, Receitas +Power bridges the gap between technological innovation and digital inclusion in gastronomy. Beyond promoting creativity and participation, it contributes a scalable and replicable model for accessible user experience (UX) design and inclusive interaction on digital culinary platforms [19,20].

This research was guided by a set of questions designed to identify the technological, methodological, and accessibility challenges associated with the development of inclusive digital culinary platforms. These questions aimed to establish a framework for evaluating current practices and defining requirements for a universally accessible, user-centered solution [6,7,8].

Accordingly, the study was structured around the following research questions:

• Q1: Do existing digital platforms enable users to create and customise their own culinary recipes beyond mere recommendation systems?
• Q2: What best practices and design principles ensure web accessibility across different disability types?
• Q3: Which digital solutions are most effective in ensuring accessibility and inclusiveness in culinary applications?

Together, these questions shaped the scope of the study, supporting the identification of accessibility gaps in existing digital culinary ecosystems and guiding the design and evaluation of the Receitas +Power platform. The overarching goal was to integrate inclusive design methodologies with contemporary web development frameworks and AI technologies, thereby contributing to the advancement of digital inclusion and accessibility within the gastronomic context [8,17].

The remainder of this article is organised as follows. Section 2 reviews the current literature on AI and accessible web technologies within the context of digital gastronomy. Section 3 outlines the system architecture and development process of Receitas +Power. Section 4 describes the research methodology and data analysis procedures. Section 5 presents and discusses the empirical results, while Section 6 concludes with reflections on theoretical implications, limitations, and future research directions.

2. Literature Review

The intersection of AI and accessible web technologies has significantly transformed digital interaction, particularly within the culinary and gastronomic domains. AI-based systems have enabled advanced levels of personalisation, automation, and multimodal interactivity, fostering more intuitive and inclusive user experiences [14,15,21,22]. The integration of recommendation algorithms, image recognition, and natural language generation has not only optimised nutritional analysis and dietary planning but also promoted user engagement through dynamic, data-driven content. These developments have broadened the scope of human–AI collaboration in food technology, allowing users to make more informed and healthier dietary choices [23,24].

Despite these advances, many culinary applications continue to prioritise convenience and personal customisation over universal accessibility, thereby excluding significant user groups, particularly those with disabilities [25]. This imbalance reflects a persistent gap between AI-centred innovation and inclusive digital design. A comparative analysis of contemporary platforms such as Mealime [26], SideChef [27], BigOven [28], and Cookpad [29] reveals this limitation clearly. These systems provide comprehensive functionalities, recipe generation, ingredient management, and social interaction, yet their accessibility features remain limited. Few incorporate mechanisms such as voice control, screen reader optimisation, or adaptive visual contrast, which are essential for users with sensory, motor, or cognitive impairments. Consequently, while these platforms succeed in enhancing usability, they fall short of achieving full digital inclusiveness, indicating a continuing disparity between technological sophistication and accessibility integration.

Recent research has demonstrated that deep learning techniques have gained prominence in culinary informatics, encompassing areas such as ingredient recognition, recipe prediction, and automated meal planning [20,22,30]. Convolutional and transformer-based neural networks have achieved high accuracy in food classification and context-aware recommendation [21,22,30]. Nevertheless, most systems remain restricted to efficiency and precision objectives, overlooking accessibility and user diversity as design priorities. Studies such as Pacífico et al. [30]represent preliminary attempts to address this deficiency by applying collaborative filtering to generate allergen-safe recipes, combining AI reasoning with inclusive content adaptation. However, such approaches remain the exception rather than the norm, signalling an ongoing research and implementation gap.

Web accessibility forms the normative and regulatory foundation for inclusive digital development [31]. The Web Content Accessibility Guidelines 2.1 and 2.2, supported by ISO/IEC 40500:2012, ISO 9241-210:2019, and ISO 9241-11:2018, define four essential principles, perceivability, operability, understandability, and robustness, that underpin accessible design [16,32,33,34,35]. These standards have been successfully applied across sectors ranging from business software to conversational agents and assistive mobile technologies, demonstrating their adaptability and long-term relevance [17,18,35,36]. Nonetheless, accessibility research exhibits a structural asymmetry: visual impairments are disproportionately represented, whereas auditory, cognitive, and motor dimensions remain underexplored [32,35]. As Horton and Quesenbery [37] assert, accessibility must be embedded from the earliest stages of design rather than treated as a post-development corrective measure.

Recent scholarships increasingly support this integrative perspective. Chemnad and Othman [17] emphasise that AI can facilitate automatic accessibility evaluation and dynamic content adaptation, enabling continuous compliance with evolving accessibility standards. Similarly, Manca et al. [18] and Vera-Amaro et al. [19] advocate for transparent, user-centred accessibility frameworks that leverage AI to detect and mitigate usability barriers in real time. Within human–computer interaction (HCI), several authors argue that inclusive digital systems require iterative prototyping, mixed-methods evaluation, and participatory co-design to ensure responsiveness to the diverse needs of users [38,39,40]. Collectively, these approaches illustrate a paradigm shift from static accessibility assessment to adaptive, AI-driven inclusion.

In this context, Receitas +Power emerges as a practical response to these conceptual and empirical gaps. The project operationalises the principles of universal design, inclusive interaction, and generative AI (GAI) within a single culinary web platform. Unlike existing applications, it treats accessibility not as an auxiliary function but as a fundamental design criterion. Through the integration of multimedia guidance, speech synthesis, and machine-generated recipe videos, it aims to provide equitable participation for users with diverse abilities: visual, auditory, cognitive, and motor. Moreover, the platform aligns with WCAG 2.2 standards and adheres to ethical frameworks such as the European Code of Conduct for Research Integrity [41] and the ACM SIGACCESS Research Ethics Guidelines [42].

By joining the fields of AI-based personalisation, web accessibility, and inclusive HCI, Receitas +Power contributes to a broader redefinition of digital gastronomy as a site of technological and social equity. It advances the current discourse by demonstrating how GAI, when grounded in accessibility standards and participatory design, can promote not only efficiency and engagement but also fairness, autonomy, and digital inclusion. This synthesis of adaptive intelligence and universal design thus represents a promising direction for future research and development in accessible intelligent systems [14,24,36,43,44].

3. Development of the ‘Receitas +Power’ Platform

The Receitas +Power platform [45] was designed as an inclusive digital ecosystem that promotes culinary creativity while ensuring universal accessibility. Designed for users with diverse abilities, the platform enables the creation, visualisation, and sharing of recipes through an adaptive, AI-assisted interface.

The software development methodology adopted was Prototyping [46,47], user-centred and accessibility-oriented, grounded in the WCAG 2.2 [16]guidelines and the principles of Human–Computer Interaction (HCI). Following the development phases of the methodology illustrated in Figure 1, the defined iterations were: (1) user management and authentication; (2) recipe creation and visualisation; (3) recipe storage and retrieval; and (4) integration of accessibility elements. Each iteration involved user testing and refinement cycles, ensuring that design decisions were guided by empirical feedback.

3.1. Conceptual and Architectural Design

The conceptualisation and system architecture were guided by user-centred design principles [44] universal accessibility standards [48], and usability heuristics [41]. The goal was to build a platform capable of supporting independent recipe creation, editing, and sharing for all user profiles.

Accessibility was ensured through semantic HyperText Markup Language (HTML) 5 markup and Web Accessibility Initiative – Accessible Rich Internet Applications (WAI-ARIA) 1.2 attributes [42], guaranteeing compatibility with assistive technologies such as screen readers, voice navigation, and alternative input devices.

Figure 2 presents the platform’s conceptual architecture, which integrates AI modules within an accessible user interface (UI) and a compliant controller-model structure, unified under a security and accessibility validation layer.

3.2. System Architecture and Implementation

The platform’s information architecture balances simplicity and adaptability, maintaining cognitive clarity while supporting scalability. The database schema is built around four key entities, Recipe, Ingredient, Utensil, and Quantity, providing a normalised relational model for efficient data management and future AI integration [49].

Figure 3 illustrates the modular Laravel-based layered architecture, composed of the Frontend/UI, Application, and Database layers connected through Representational State Transfer (RESTful) APIs [50]. This structure ensures maintainability, facilitates accessibility integration, and supports GAI functionalities.

3.3. Integration of Generative Artificial Intelligence and Accessibility Validation

A feature of Receitas +Power is its GAI module, which assists users in producing coherent, contextually accurate cooking instructions. Based on selected ingredients and utensils, the system automatically generates step-by-step recipes, reducing cognitive load and enhancing autonomy.

Several generative models were evaluated, including ChatGPT [51], Hugging Face [52], and DeepSeek [53] for text generation, and Stable Diffusion [54] and Vidnoz [55] for image and video synthesis. Due to computational and reliability constraints, the final implementation adopted the OpenRouter API [56] using the mistralai/mixtral-8x7b-instruct model. This configuration provided a balance between linguistic coherence, scalability, and performance [39].

The system was implemented using the Laravel framework, chosen for its modularity, maintainability, and adherence to accessibility-oriented design. Its Model–View–Controller (MVC) architecture separated data handling, interface logic, and interaction management, ensuring clean code and scalability[57].

The frontend was designed to be visually appealing, responsive, and fully accessible, as illustrated by Figure 4. The main interface allows users to create and manage recipes intuitively, with consistent navigation, legible typography, and compliance with WCAG 2.2 Level AA.

Security was ensured through Laravel’s built-in protocols for input validation, encryption, and role-based access control. The responsive layout guarantees accessibility across screen resolutions and device types, including smartphones, tablets, and desktop computers.

Usability testing constituted a core validation stage. A pilot group of 87 participants with diverse digital literacy and accessibility needs evaluated the platform across devices. Mixed observation and questionnaire methods were used to assess navigation ease, comprehension, satisfaction, and accessibility perception.

Figure 5 shows the inclusive ingredient and utensil selection interface, which allows users to choose components using visual and textual aids.

Once selections were finalised, the AI module generated procedural text and accessible formatting, as shown in Figure 6.

Figure 7 presents the final recipe visualisation view, combining ingredients, utensils, and sequential steps optimised for screen readers and compliant with WCAG 2.2 AA.

Feedback-informed continuous refinements: font adjustments, enlarged buttons, optional narration, improved keyboard focus indicators, and enhanced colour contrast.

Accessibility validation was integrated across all development phases, using both automated and manual testing. The Web Accessibility Evaluation (WAVE) Tool [25] confirmed full WCAG 2.2 compliance, ensuring conformance to the four POUR principles: Perceivable, Operable, Understandable, and Robust [37,58].

Key accessibility features include full keyboard operability, descriptive alternative text, adjustable typography, focus indicators, and seamless compatibility with screen readers and speech input. These measures collectively ensured usability parity for all users.

3.4. Summary of Development Approach

The Receitas +Power project exemplifies how inclusive design, GAI, and scalable web technologies can converge to create a socially responsible and technologically advanced digital platform. By embedding accessibility as a structural design parameter, rather than as an afterthought, the platform demonstrates that innovation and inclusiveness are mutually reinforcing objectives.

This modular, AI-augmented, and standards-compliant framework offers a replicable model for inclusive digital transformation across other domains of human–technology interaction.

4. Methodology

This section details the methodological framework adopted to develop and evaluate the Receitas +Power platform. It outlines the research design, prototyping process, accessibility integration, participant recruitment, and data analysis procedures used to ensure methodological rigour and empirical validity.

4.1. Research Design

This study adopts an exploratory and multidimensional research design, structured as a case study centred on the Receitas +Power web platform. The approach combines qualitative and quantitative analyses to capture both the experiential and measurable aspects of accessibility and usability in AI-enhanced culinary systems. Given the project’s dual focus, technological development and inclusive user experience, the case study framework was considered the most appropriate for examining the interplay between design processes, user interaction, and accessibility outcomes [38,40,59].

Although the study follows a predominantly exploratory and qualitative orientation, complementary quantitative analyses were integrated to assess the reliability of the measurement instruments and to explore possible relationships among conceptual dimensions. This methodological pluralism aligns with current recommendations for mixed-methods research in HCI and accessibility studies [39,40].

4.2. System Development and Prototyping

The development of Receitas +Power followed an iterative prototyping methodology, in line with agile user-centred design principles [38,59]. Each iteration included communication, rapid planning, modelling, construction, and user feedback. The technical implementation relied on PHP [60] , HTML5 [61], CSS3[62], and JavaScript [63], supported by the XAMPP local server environment [64] and MySQL database management system [65].

The prototyping process enabled progressive refinement of interface components, ensuring alignment between usability objectives and accessibility requirements. Early prototypes focused on user management and basic recipe creation, while later iterations incorporated multimedia elements (e.g., video instructions generated through AI) and accessibility features designed according to WCAG 2.2 and ISO 9241-210 standards [16,33,34].

4.3. Accessibility Integration

Accessibility was operationalised through the integration of W3C’s WCAG 2.2 guidelines and WAI-ARIA 1.2 specifications [16,32]. Design elements prioritised perceivability (clear contrast ratios and alternative text), operability (keyboard navigation and voice-assistant compatibility), understandability (simplified language and consistent layout), and robustness (cross-browser and assistive technology compatibility).

Automated evaluation tools such as WAVE [25] and manual heuristic inspections were employed to identify and address accessibility issues during development. The system architecture was informed by contemporary frameworks for AI-driven accessibility evaluation and content adaptation [6,7,17,18], ensuring compliance with both technical and ethical standards [41,42].

4.4. Evaluation Procedure and Participants

The empirical evaluation phase involved 87 participants recruited through voluntary online participation and community outreach. Participants represented a broad demographic distribution, including individuals with and without disabilities, ensuring diversity in technological literacy and accessibility needs.

Recruitment was conducted through open calls disseminated via educational institutions and members of Associação +Power [66]. Inclusion criteria required participants to be over 18 years old, able to use a computer or smartphone independently, and willing to consent to data collection for research purposes. Exclusion criteria comprised incomplete responses and duplicate entries. The final sample (n = 87) included 9 participants self-identifying with functional limitations (visual, cognitive, or motor).

The System Usability Scale (SUS) [49,57] was employed as the primary quantitative instrument to assess perceived usability, efficiency, and satisfaction. The SUS consists of ten statements rated on a five-point Likert scale, allowing the derivation of a composite usability score ranging from 0 to 100. In addition, open-ended qualitative questions were included to capture user perceptions regarding accessibility, visual clarity, and interaction flow.

Ethical considerations were rigorously observed in accordance with the European Code of Conduct for Research Integrity [41] and the ACM SIGACCESS Research Ethics Guidelines [42]. All participants provided informed consent, and their anonymity and data confidentiality were guaranteed.

4.5. Data Analysis

Quantitative data were analysed using descriptive statistics, reliability testing, and correlational analysis to assess internal consistency and interdimensional relationships among usability constructs, following established usability testing procedures [67]. The SUS was used to compute mean scores and standard deviations, while the Cronbach’s alpha coefficient was applied to evaluate the internal consistency of subscales. Pearson’s correlation coefficients were calculated to examine associations between usability, satisfaction, and accessibility.

Qualitative responses were examined through thematic analysis, following a mixed inductive–deductive approach. Themes were coded manually, verified by two independent researchers, and cross-compared to ensure interpretive reliability. Integration of quantitative and qualitative findings followed a triangulation strategy, enabling cross-validation of usability metrics and user perceptions.

For internal consistency, Cronbach’s α values ranged from 0.798 to 0.849, indicating acceptable to high reliability. Moderate Pearson correlations (0.45–0.55) suggested interdependence between the measured constructs.

4.6. Methodological Limitations

As an exploratory case study, this research does not aim for statistical generalisation but rather for analytical insight into inclusive design practices. The participant sample, although diverse, was not probabilistically selected, and therefore caution should be exercised in extrapolating results. Nonetheless, the mixed-methods approach ensured both depth and reliability of analysis, making the findings relevant for subsequent empirical validation in larger and more heterogeneous populations.

Future studies should incorporate longitudinal tracking and behavioral analytics to observe sustained engagement with accessible interfaces over time. Additionally, comparative experiments across different AI models could further clarify the specific contribution of generative intelligence to accessibility outcomes.

5. Data Analysis and Results

This section presents the empirical findings derived from both quantitative and qualitative analyses. It summarises participant demographics, explores responses to each research question, and discusses the implications of the results for accessibility, usability, and AI-enabled inclusiveness.

5.1. Overview of Findings

The findings provide robust evidence that combining AI with accessibility standards can substantially enhance inclusiveness, usability, and user satisfaction in digital culinary environments. Quantitative and qualitative analyses collectively address the three research questions (Q1–Q3), examining user empowerment, accessibility practices, and the effectiveness of AI-enabled inclusive solutions.

5.2. Participant Demographics and Digital Profiles

A total of 87 participants completed the usability and accessibility evaluation of Receitas +Power. As summarised in Table 1, the sample represented diverse age groups, balanced gender distribution, and varying levels of digital literacy. The inclusion of participants with visual, cognitive, and other functional limitations ensured that multiple accessibility perspectives were reflected in the dataset.

Demographic diversity aligns with inclusive HCI research recommendations, strengthening the ecological validity of the analysis and ensuring that findings capture a wide range of user experiences [39,40].

5.3. Q1: User Empowerment Beyond Recommendation Systems

Addressing Q1, participants consistently reported that Receitas +Power extended beyond traditional recommendation systems by allowing genuine recipe co-creation. Through AI-assisted ingredient recognition, allergen detection, and nutritional recalculation, users could modify or generate recipes dynamically.

Thematic analysis highlighted control, creativity, and flexibility as recurring descriptors, suggesting that participants experienced a sense of authorship rather than passive consumption. These results align with Bondevik et al. [21] and Starke et al. [23], who emphasise autonomy as a central motivator of engagement in culinary recommender systems. Consequently, Receitas +Power exemplifies human–AI co-creation by transforming recommendation algorithms into participatory creative tools [44].

5.4. Q2: Accessibility Design Across Disability Types

For Q2, the study investigated how accessibility guidelines could support diverse disability profiles. Automated evaluation using WAVE [25] and manual inspection confirmed compliance with WCAG 2.2 and WAI-ARIA 1.2 standards [16,32].

Participants with visual or motor impairments valued the system’s text-to-speech, voice command, and keyboard navigation features, while those with cognitive challenges highlighted the importance of clear structure, readable typography, and simplified textual content. These results reaffirm Horton and Quesenbery’s [37] assertion that accessibility must be designed proactively rather than retrofitted. They also support Miesenberger et al. [36], demonstrating that accessibility and usability reinforce one another when addressed holistically.

5.5. Q3: Effectiveness of AI-Enabled Inclusive Solutions

The third research question assessed the overall usability, efficiency, and satisfaction with the platform. Quantitative analysis based on the SUS indicated consistently high ratings, as summarised in Table 2.

The distribution of scores is illustrated in Figure 8, which shows a clear clustering between 75 and 90, indicating that most participants evaluated the system as good to excellent.

To ensure internal consistency, Cronbach’s α and Pearson correlations were calculated for each construct, as shown in Table 3.

The scale demonstrated good internal consistency, with Cronbach’s α values ranging between 0.798 and 0.849. According to Tavakol and Dennick [68], α values above 0.70 are considered acceptable, while those between 0.80 and 0.90 indicate strong reliability without suggesting redundancy. Moderate positive correlations (r = 0.45–0.55) among usability, satisfaction, and accessibility constructs confirm that these variables are interrelated but non-redundant. This interpretation aligns with the discriminant validity criterion based on the Heterotrait–Monotrait ratio (HTMT < 0.85) proposed by Henseler et al.[69], confirming that the constructs share moderate but not excessive overlap. Such association levels are consistent with general effect size guidelines, where r ≈ 0.5 represents a moderate relationship [70].

5.6. Qualitative Insights

Qualitative responses further enriched the quantitative interpretation. Four principal themes emerged, as summarised in Table 4. Thematic summary of qualitative responses (n = 87).

These statements reinforce that accessibility features enhanced rather than constrained usability, an observation consistent with contemporary literature positioning accessibility as an enabler of innovation [36].

5.7. Synthesis and Implications

Taken together, the quantitative and qualitative findings demonstrate that Receitas +Power effectively integrates AI-based adaptability with accessibility frameworks to deliver a high-usability, inclusive experience. The excellent SUS performance, solid reliability metrics, and positive user narratives collectively validate the platform’s design principles.

These outcomes substantiate the premise that accessibility, when embedded within GAI systems, not only enhances user engagement but also fosters autonomy and equity. The study therefore contributes to a growing theoretical consensus that inclusive, human-centred AI can bridge the gap between technological sophistication and digital fairness [36,44].

6. Conclusions, Limitations, and Future Work

This study examined how AI can enhance accessibility, inclusiveness, and personalisation in digital culinary environments. Using an exploratory, multidimensional case study centred on the Receitas +Power platform, three research questions were addressed concerning user empowerment beyond recommendation systems, accessibility across disability types, and the effectiveness of inclusive AI-driven solutions.

The findings confirm that Receitas +Power extends the capabilities of conventional culinary platforms by enabling users to create, customise, and adapt their own recipes through AI-assisted ingredient recognition, allergen detection, and nutritional recalibration. This functionality moves beyond passive recommendation models, positioning the platform as a participatory, creative, and inclusive environment. In relation to Q1, the results illustrate how human–AI collaboration can foster user autonomy and engagement while broadening the scope of digital gastronomy.

For Q2, the integration of WCAG 2.2 and WAI-ARIA 1.2 guidelines, supported by adaptive multimodal design and AI-assisted accessibility evaluation, proved essential for accommodating a wide range of disability profiles. The data demonstrate that inclusiveness and usability are interdependent dimensions rather than conflicting objectives. This evidence supports longstanding arguments that accessibility must be embedded at the conceptual and developmental stages of design, rather than treated as a corrective addition [36,37].

Regarding Q3, quantitative analysis based on the SUS indicated a high level of user satisfaction (M = 80.6), supported by internal consistency measures (Cronbach’s α = 0.798–0.849). Moderate positive correlations between usability, satisfaction, and accessibility constructs (r = 0.45–0.55) suggest a coherent and mutually reinforcing relationship among these factors. Collectively, these results validate that AI-enabled adaptive interfaces—when anchored in universal design principles—offer an effective route towards digital inclusion.

From a theoretical standpoint, this study contributes to the emerging discourse on human–AI co-creation by demonstrating that generative intelligence can operationalise accessibility as an intrinsic design property. It reinforces the conceptual transitions identified by McGuire [44] and Miesenberger et al. [36], where accessibility is redefined as a catalyst for innovation rather than as a compliance requirement. Moreover, the results expand Norman’s [71] user-centred design model by embedding GAI to accommodate diverse user capabilities.

From a practical perspective, Receitas +Power establishes a replicable framework for inclusive digital gastronomy, integrating AI adaptability, WCAG-conformant design, and ethical user research practices. The project adheres to the European Code of Conduct for Research Integrity [41] and the ACM SIGACCESS Research Ethics Guidelines [42], ensuring both methodological rigour and ethical accountability.

Despite its strengths, this study presents several limitations. As an exploratory case study, it focuses on analytical insight rather than statistical generalisation. The use of a convenience sample, while diverse, restricts representativeness. Furthermore, the cross-sectional design captures user perception at a single point in time, limiting the understanding of long-term engagement or behavioural adaptation. Finally, while the SUS and reliability measures confirm robustness, future research could benefit from triangulating these results with eye-tracking data, physiological metrics, or longitudinal user analytics to capture deeper behavioural patterns.

Future work should therefore aim to expand the scope and duration of empirical testing, employing longitudinal and comparative studies across domains such as health, education, and personalised nutrition. Integrating multimodal sensory data, such as voice, gesture, or gaze tracking, could further refine adaptive accessibility mechanisms. In addition, predictive analytics and machine learning models could be deployed to anticipate user needs dynamically, enabling proactive adjustments to interface complexity and content delivery.

In conclusion, Receitas +Power demonstrates that the convergence of AI, accessibility, and inclusive design is not only technically viable but socially transformative. The platform exemplifies how innovation can advance digital equity by reimagining accessibility as a foundation for creativity, autonomy, and participation. It thereby contributes to a broader paradigm shift towards intelligent systems that are efficient, ethical, and universally accessible.