Accelerating Enterprise AI Decision-Making

Accelerating Enterprise AI Decision-Making

Breaking the Gridlock: A CXO’s Guide to Accelerating Enterprise AI Decision-Making.

Across industries, large enterprises are investing heavily in artificial intelligence as a strategic imperative for digital transformation. Yet despite ambitious visions and substantial investments, many organizations find their AI initiatives stalling—not because of technological limitations, but due to organizational gridlock. Here is a peek into the organizational barriers that impede AI progress in large enterprises and actionable strategies for CXOs to break through these bottlenecks. By reimagining governance structures, decision processes, and organizational dynamics, leaders can transform their enterprises from AI laggards to AI leaders, unlocking the transformative potential that currently lies dormant in countless stalled initiatives.

The Velocity Imperative

Your organization has recognized artificial intelligence as a transformative force. Strategic plans highlight AI as a key differentiator. Substantial investments have been made in technology, data infrastructure, and specialized talent. Promising use cases have been identified across functions, from customer experience to operational efficiency.

Yet as implementation efforts progress, a troubling pattern emerges. Projects that technical teams estimate should take weeks extend to months or even years. Proofs of concept succeed technically but never transition to production. Budget cycles pass without tangible results materializing from AI investments. Meanwhile, more agile competitors—often smaller organizations with fewer resources but more streamlined decision processes—move rapidly from concept to implementation, capturing market opportunities and establishing competitive advantages.

The consequences are significant and compounding. According to McKinsey, organizations with streamlined AI decision processes achieve 3-5x greater return on AI investments compared to those with significant organizational barriers. A Boston Consulting Group study found that companies with inefficient approval processes for digital initiatives take 2.7 times longer to bring innovations to market, with each month of delay representing an average of $15 million in unrealized value for large enterprises.

Beyond financial impact, organizational gridlock creates a cascade of negative effects: demoralized technical talent who see their innovations stalled, diminished executive confidence in AI’s transformative potential, and growing skepticism from business units asked to invest resources in initiatives that consistently under-deliver. One Fortune 500 CIO described this as “the AI credibility gap”—the widening disconnect between AI’s theoretical promise and its practical delivery within the organization.

The following is a practical framework for CXOs to identify, address, and overcome the organizational barriers that impede AI progress. By implementing these strategies, you can transform your enterprise’s approach to AI, ensuring that investments in technology and talent translate into tangible business outcomes at market-relevant speed.

Understanding the Organizational Barriers to AI Progress

The Anatomy of AI Gridlock

To address organizational gridlock effectively, leaders must first understand its underlying causes, which typically extend far beyond simple bureaucratic inertia:

1. Governance Complexity:
   • Multiple oversight bodies with overlapping authority
   • Unclear decision rights and accountability
   • Excessive layers of approval required for progress
   • Misalignment between governance processes and AI development needs
2. Risk Aversion Dynamics:
   • Asymmetric incentives that heavily penalize failure while modestly rewarding success
   • Diffused responsibility where multiple stakeholders can block progress but no single entity can approve
   • “Perfect over good” mentality that prevents iterative implementation
   • Status quo bias in decision-making frameworks
3. Organizational Silos:
   • Disconnection between technical teams and business units
   • Fragmented authority across IT, data, compliance, and business functions
   • Information barriers that prevent holistic understanding
   • Competing priorities and success metrics across departments
4. Resource Allocation Rigidity:
   • Annual budgeting cycles that constrain dynamic resource allocation
   • Project funding models ill-suited to AI’s iterative nature
   • Inflexible staffing approaches that limit cross-functional collaboration
   • Return on investment frameworks designed for predictable initiatives rather than exploratory innovation
5. Cultural Resistance:
   • Discomfort with AI’s probabilistic nature among leaders accustomed to deterministic business cases
   • Fear of automation-driven job displacement creating passive resistance
   • Expertise territoriality among existing analytics and business intelligence functions
   • Reluctance to adopt new ways of working required for AI success

These factors rarely operate in isolation. Instead, they create reinforcing patterns that collectively transform organizational process from a necessary coordination mechanism into a significant barrier to progress.

The Decision Velocity Gap

This organizational gridlock manifests most acutely in what we term the “decision velocity gap”—the dramatic difference in the speed at which technical AI progress can occur versus the pace at which organizational decisions enable implementation:

This gap creates substantial organizational friction, with technical teams working at dramatically different cadences than the surrounding business and governance functions. As one AI leader at a global financial institution described it, “Our technical velocity is measured in days and weeks, while our approval and implementation velocity is measured in quarters and fiscal years.”

Impact

The consequences of this velocity gap extend far beyond frustration and manifest in tangible business impact:

1. Competitive Vulnerability: A global retailer spent 14 months navigating internal approvals for an AI-driven inventory optimization solution. During this period, three competitors implemented similar capabilities, capturing an estimated $300 million in market share through improved product availability and reduced carrying costs.
2. Innovation Atrophy: A pharmaceutical company’s promising AI drug discovery platform languished for 22 months awaiting sequential approvals from IT architecture, data governance, compliance, and business leadership teams. By the time approvals were secured, the original technical approach was outdated, requiring substantial rework.
3. Talent Exodus: A financial services firm lost 40% of its AI specialists over 18 months, with exit interviews consistently citing “implementation friction” and “inability to see projects through to impact” as primary departure reasons.
4. Investment Inefficiency: A manufacturing conglomerate invested $45 million in AI infrastructure and talent over three years but realized less than $5 million in business impact. Analysis revealed that dozens of initiatives were partially completed but stalled at various approval gates, creating a “graveyard of good ideas” that consumed resources without delivering results.

These examples illustrate how organizational gridlock transforms AI from a potential competitive advantage into a resource-consuming liability with minimal business return.

Strategic Framework for Breaking the Gridlock

Addressing organizational barriers to AI progress requires a comprehensive approach that balances governance needs with implementation velocity.

Strategy 1: Reimagining AI Governance

Traditional governance structures often become bottlenecks for AI progress. A reimagined approach creates appropriate oversight while enabling momentum:

1. Streamlined Governance Bodies:
   • Consolidate oversight into a single AI Governance Council with cross-functional representation
   • Establish clear decision authority with explicit delegation thresholds
   • Implement regular, frequent meeting cadences (weekly rather than monthly or quarterly)
   • Create fast-track approval channels for low-risk or time-sensitive initiatives
2. Tiered Governance Framework:
   • Develop risk-based classification for AI initiatives (e.g., Tier 1-3 based on data sensitivity, business impact, and technical complexity)
   • Create differentiated approval paths with streamlined processes for lower-risk initiatives
   • Establish default approval timeframes with escalation mechanisms for delays
   • Implement post-implementation audits rather than pre-implementation approvals for low-risk initiatives
3. Delegated Authority Model:
   • Push decision rights to the lowest appropriate level in the organization
   • Create clear guardrails and boundaries rather than approval gates
   • Establish exception-based intervention rather than step-by-step oversight
   • Implement regular retrospectives to refine delegated authority parameters
4. Outcome-Focused Oversight:
   • Shift governance focus from process compliance to outcome achievement
   • Create dashboards that highlight both velocity and quality metrics
   • Develop incentives for governance participants to enable implementation
   • Implement regular governance process optimization based on velocity impact

A global telecommunications company implemented this approach by consolidating seven separate approval committees into a single AI Governance Council with representatives from business, technology, legal, risk, and data functions. The council established a tiered framework with clear decision rights and meeting three times weekly for 60-minute decision sessions. The result was a 70% reduction in approval times for AI initiatives, with the average time from concept to implementation decreasing from 8 months to 10 weeks.

Strategy 2: Creating Agile Decision Processes

Beyond governance structures, the underlying decision processes require reformation:

1. Decision Path Mapping:
   • Document current state decision flows for AI initiatives
   • Identify redundant approvals and unnecessary dependencies
   • Create streamlined, standardized decision paths
   • Establish clear entry and exit criteria for each decision stage
2. Parallel Processing Implementation:
   • Redesign sequences to enable concurrent rather than sequential reviews
   • Create integrated assessment frameworks that satisfy multiple stakeholder requirements simultaneously
   • Implement collaborative tools that enable parallel evaluation
   • Establish coordination mechanisms to resolve cross-functional conflicts
3. Time-Bounded Decision Framework:
   • Implement default approval timeframes for each decision stage
   • Create escalation paths when decisions exceed timeframes
   • Establish “proceed unless rejected” mechanisms for appropriate decision points
   • Develop tracking and transparency for decision velocity metrics
4. Decision Debt Management:
   • Identify and address accumulations of pending decisions
   • Create prioritization frameworks for backlogged approvals
   • Implement temporary surge capacity for decision clearing
   • Establish regular maintenance to prevent future decision backlogs

A healthcare organization applied these principles to their AI clinical decision support initiatives, mapping the 27 distinct approval steps in their legacy process and creating a redesigned pathway with parallel reviews and clear timeframes. The revised process reduced decision time from an average of 9 months to 7 weeks while maintaining all necessary clinical safety and regulatory compliance evaluations.

Strategy 3: Implementing Innovation-Friendly Funding Models

Traditional funding approaches often misalign with AI’s iterative, uncertain nature:

1. Venture-Inspired Funding:
   • Establish staged funding models with clear evaluation gates
   • Create dedicated innovation capital with streamlined access
   • Implement portfolio approaches that expect some initiatives to fail
   • Develop success metrics appropriate to each development stage
2. Rapid Experimentation Funds:
   • Create pre-approved budgets for low-cost proofs of concept
   • Establish streamlined access to initial funding (under $100k) with minimal approval
   • Implement fast-cycle evaluation for continued funding
   • Develop clear graduation criteria from experiment to formal initiative
3. Value-Based Investment:
   • Tie funding to validated business outcomes rather than technical milestones
   • Implement continuous funding models based on demonstrated value
   • Create mechanisms to rapidly scale successful initiatives
   • Develop flexible reallocation approaches as learning emerges
4. Embedded AI Investment:
   • Integrate AI funding into business unit budgets rather than centralized IT allocation
   • Create co-investment models between technical teams and business units
   • Establish outcome-based chargeback mechanisms
   • Implement shared accountability for return on investment

A financial services institution implemented a three-tiered funding model for AI initiatives: an “innovation sandbox” with pre-approved funding for proofs of concept under $75,000; a “growth portfolio” with streamlined approval for promising initiatives under $500,000; and a “scale investment” category for enterprise initiatives with validated business cases. This approach increased the number of AI experiments by 340% while improving the percentage of initiatives delivering measurable business value from 23% to 67%.

Strategy 4: Building Cross-Functional AI Teams

Organizational silos significantly impede AI progress and require structural solutions:

1. Dedicated AI Product Teams:
   • Form persistent, cross-functional teams around AI use cases
   • Include business, technology, data, and compliance expertise within teams
   • Establish end-to-end responsibility for outcomes
   • Create direct reporting lines to senior sponsors
2. Embedded Business Translators:
   • Identify and develop individuals who bridge technical and business domains
   • Embed these translators within both technical and business teams
   • Create career paths that value cross-functional expertise
   • Establish formal translation processes for requirements and results
3. Capability-Centric Organization:
   • Organize around business capabilities rather than technical or functional silos
   • Create cross-functional ownership of outcomes
   • Establish clear accountability for end-to-end value delivery
   • Implement metrics that reflect holistic capability performance
4. Fusion Team Operation:
   • Implement agile practices across business and technical contributors
   • Create shared workspaces (physical or virtual) that enable continuous collaboration
   • Establish common processes and tools across functions
   • Develop integrated performance objectives and success metrics

A retail organization implemented this approach by creating AI “pods” combining data scientists, engineers, business analysts, compliance specialists, and operations staff working as unified teams on specific use cases. These pods operated with substantial autonomy within clear guardrails, with leadership focusing on outcomes rather than processes. The pods reduced time-to-implementation by 65% while improving adoption of AI solutions by business users.

Strategy 5: Building Decision Velocity Culture

Cultural factors significantly influence organizational gridlock and require deliberate attention:

1. Leadership Modeling:
   • Demonstrate rapid, decisive action at executive levels
   • Publicly recognize and reward appropriate velocity
   • Share personal examples of balancing speed and quality
   • Hold accountability forums on decision timeliness
2. Risk Calibration:
   • Develop nuanced understanding of AI risks across leadership
   • Create appropriate risk frameworks for different AI applications
   • Implement “speed of risk” assessment for initiatives
   • Establish retrospectives that evaluate both action and inaction risks
3. Psychological Safety:
   • Create environments where challenging status quo is valued
   • Implement “learning from failure” practices
   • Recognize appropriate risk-taking even when outcomes disappoint
   • Develop balanced consequence management that doesn’t solely penalize failure
4. Velocity Metrics and Visibility:
   • Establish and publicly track decision velocity metrics
   • Create transparency into organizational bottlenecks
   • Implement regular process retrospectives focused on speed
   • Develop individual and team recognition for enabling velocity

A professional services firm implemented these cultural elements through a “Decision Velocity Initiative” sponsored by the CEO. The program included leadership training on psychological safety, regular publication of decision velocity metrics, and recognition for teams demonstrating appropriate speed without compromising quality. The initiative reduced average decision time for AI and digital initiatives by 58% while maintaining or improving outcome quality.

Implementation Roadmap for Accelerating AI Decision-Making

Transforming your organization’s decision velocity requires a structured implementation approach that delivers incremental benefits while building toward comprehensive capability.

Phase 1: Assessment and Quick Wins (1-2 Months)

1. Current State Analysis:
   • Document existing decision processes for AI initiatives
   • Identify primary bottlenecks and friction points
   • Quantify current timelines and delays
   • Assess opportunity costs of status quo
2. Leadership Alignment:
   • Create shared understanding of decision velocity impact
   • Establish executive commitment to improvement
   • Identify executive sponsors for transformation
   • Develop common vocabulary and objectives
3. Quick Win Implementation:
   • Identify 3-5 high-impact process changes for immediate implementation
   • Target the most severe bottlenecks first
   • Create visible momentum through early successes
   • Develop measurement approach for improvement
4. Change Communication:
   • Develop messaging around the transformation initiative
   • Create transparency into current state challenges
   • Share vision for future state capabilities
   • Establish regular progress communication channels

Phase 2: Foundational Transformation (2-4 Months)

1. Governance Restructuring:
   • Implement consolidated governance structure
   • Establish tiered approval frameworks
   • Develop delegated authority model
   • Create new meeting and decision cadences
2. Process Redesign:
   • Redesign core decision processes for AI initiatives
   • Implement parallel processing workflows
   • Create time-bounded decision frameworks
   • Develop tracking tools for decision flow
3. Initial Funding Reform:
   • Establish rapid experimentation funding mechanisms
   • Implement portfolio-based allocation for AI investments
   • Create value-based funding criteria
   • Develop more flexible budget processes
4. Team Structure Evolution:
   • Pilot cross-functional AI teams for priority use cases
   • Implement initial embedded business translators
   • Create fusion team operations models
   • Develop integrated performance frameworks

Phase 3: Scaling and Optimization (4-8 Months)

1. Enterprise Expansion:
   • Extend transformed governance and processes across all AI initiatives
   • Scale cross-functional team models enterprise-wide
   • Implement comprehensive funding reforms
   • Create standardized tooling and enablement
2. Metrics and Accountability:
   • Establish comprehensive decision velocity measurement
   • Implement regular reporting and transparency
   • Create accountability mechanisms for decision timeliness
   • Develop individual and team incentives aligned with velocity
3. Culture Reinforcement:
   • Implement broad communication of transformation benefits
   • Create recognition programs for velocity enablers
   • Develop communities of practice for continuous improvement
   • Establish learning forums for ongoing optimization
4. Continuous Improvement:
   • Implement regular review of decision velocity metrics
   • Create structured optimization processes
   • Develop innovation pipeline for decision approaches
   • Establish benchmarking against industry best practices

Organizational Enablers for Accelerated Decision-Making

Successfully transforming decision velocity requires specific enablers that support and sustain the transformation.

Technology Enablers

While this challenge is primarily organizational rather than technical, certain technologies can significantly enable improvement:

1. Decision Management Systems:
   • Workflow platforms that digitize and track approval processes
   • Analytics tools that identify bottlenecks and predict delays
   • Collaboration platforms that enable asynchronous decision-making
   • Documentation tools that create transparency and accountability
2. AI Governance Platforms:
   • Model inventories and lifecycle management tools
   • Automated risk assessment frameworks
   • Compliance documentation and audit trails
   • Performance monitoring and alerting systems
3. Collaboration Infrastructure:
   • Digital workspace tools for cross-functional teams
   • Asynchronous communication platforms
   • Document co-creation and management systems
   • Knowledge management repositories
4. Metrics and Visualization:
   • Real-time dashboards for decision velocity
   • Process mining tools to identify patterns and bottlenecks
   • Predictive analytics for decision pipeline management
   • Executive visibility into organizational friction points

Skill and Capability Development

New capabilities are required to operate effectively in an accelerated decision environment:

1. Decision Clarity Skills:
   • Problem framing and decision structuring
   • Separating critical from non-critical decisions
   • Clarifying decision criteria and trade-offs
   • Communicating decision rationale effectively
2. Cross-Functional Fluency:
   • Understanding diverse stakeholder perspectives
   • Translating between technical and business domains
   • Navigating organizational politics and influence
   • Creating alignment across different organizational cultures
3. Velocity-Focused Leadership:
   • Appropriate delegation and empowerment
   • Balancing speed with appropriate consideration
   • Creating urgency without anxiety
   • Managing the psychology of rapid change
4. Risk Intelligence:
   • Nuanced understanding of AI-specific risks
   • Balancing innovation and protection priorities
   • Developing appropriate risk mitigation plans
   • Communicating risk effectively to stakeholders

Policy and Standards Evolution

Existing policies often inadvertently create barriers to AI progress and require thoughtful revision:

1. AI-Specific Governance Policies:
   • Develop tailored governance frameworks for AI initiatives
   • Create appropriate categorization based on risk and impact
   • Establish clear documentation and assessment requirements
   • Implement streamlined approval processes for common use cases
2. Data Governance Acceleration:
   • Implement pre-approved data usage frameworks for common scenarios
   • Create streamlined data access processes for validated use cases
   • Establish clear data quality requirements and validation approaches
   • Develop automated compliance documentation for data usage
3. Technology Standards Modernization:
   • Update security and architecture standards for AI technologies
   • Create pre-approved patterns for common implementation scenarios
   • Establish flexible guidelines rather than rigid requirements
   • Implement regular refresh cycles to keep pace with technology evolution
4. Ethical Guidelines Integration:
   • Develop clear ethical frameworks for AI applications
   • Create practical assessment tools for ethical considerations
   • Establish streamlined review processes for straightforward cases
   • Implement specialized review only for novel ethical challenges

Advanced Strategies for Decision Acceleration

For organizations seeking to push beyond initial transformation, additional strategies can further enhance decision velocity.

Strategy 1: AI-Powered Decision Optimization

Applying AI to the challenge of AI governance itself represents a significant opportunity:

1. Decision Process Intelligence:
   • Implement process mining to identify decision flow patterns
   • Apply AI to predict approval timeline risks
   • Create recommendation engines for optimal approval routing
   • Develop anomaly detection for stalled decisions
2. Intelligent Documentation:
   • Automate creation of governance documentation
   • Implement natural language processing for policy compliance checking
   • Develop AI assistants for governance requirements navigation
   • Create auto-generation of business case elements
3. Predictive Risk Assessment:
   • Apply machine learning to predict AI implementation risks
   • Develop automated identification of similar past initiatives
   • Create intelligent suggestion of appropriate controls
   • Implement continuous learning from implementation outcomes
4. Decision Automation:
   • Identify routine decisions suitable for full automation
   • Develop AI-assisted decision support for complex cases
   • Create confidence scoring for automated recommendations
   • Implement human-in-the-loop processes for appropriate oversight

A global insurance company implemented these approaches, developing an AI solution that automatically generated required governance documentation based on project characteristics, reducing documentation time by 70% and identifying potential regulatory issues with 92% accuracy compared to manual reviews.

Strategy 2: Ecosystem Decision Acceleration

Many AI initiatives involve external partners and require coordination beyond organizational boundaries:

1. Partner Pre-Certification:
   • Establish streamlined approval for pre-vetted partners
   • Create standardized contractual frameworks
   • Implement continuous compliance monitoring
   • Develop shared responsibility models for governance
2. Industry Standardization:
   • Participate in cross-industry governance standardization
   • Adopt common frameworks where available
   • Contribute to developing shared best practices
   • Leverage industry-specific assessment tools
3. Regulatory Engagement:
   • Develop proactive relationships with key regulators
   • Participate in regulatory sandboxes and innovation programs
   • Create transparency into governance approaches
   • Establish early consultation for novel use cases
4. Co-Creation Frameworks:
   • Develop integrated decision processes with key partners
   • Create shared visibility into approval progress
   • Establish joint escalation mechanisms for bottlenecks
   • Implement cross-organization decision forums

A banking consortium implemented this approach by creating standardized compliance documentation and pre-approved architectural patterns for common AI use cases in financial services, reducing partner onboarding time from 6 months to 3 weeks while maintaining regulatory compliance.

Strategy 3: Organizational Resilience Development

Building resilience into decision processes ensures velocity can be maintained even during disruption:

1. Decision Continuity Planning:
   • Establish backup decision authorities for key roles
   • Create delegated decision rights for disruption scenarios
   • Implement clear escalation paths when normal processes are impeded
   • Develop simplified decision frameworks for crisis conditions
2. Decision Stress Testing:
   • Conduct simulations of high-volume decision scenarios
   • Test governance processes under compressed timeframes
   • Identify and address potential failure points
   • Create continuous improvement based on stress test results
3. Adaptive Governance:
   • Develop governance approaches that scale with organizational needs
   • Create flexible frameworks that adjust to changing conditions
   • Implement regular reassessment of governance effectiveness
   • Establish innovation pipeline for governance approaches
4. Decentralized Coordination Mechanisms:
   • Implement distributed decision rights with central visibility
   • Create self-regulating governance mechanisms
   • Develop peer review rather than hierarchical approval for appropriate decisions
   • Establish community-based standards evolution

Measuring Success and Ensuring Sustainability

To ensure the transformation delivers lasting benefits, organizations must establish comprehensive measurement and feedback mechanisms.

Success Metrics Framework

1. Velocity Metrics:
   • Time from idea to implementation for AI initiatives
   • Decision cycle time for key approval stages
   • Percentage of decisions made within target timeframes
   • Resource time spent on approval processes
2. Quality Metrics:
   • Implementation success rate for approved initiatives
   • Incidence of governance-related issues post-implementation
   • Compliance with key regulatory requirements
   • Stakeholder satisfaction with decision quality
3. Business Impact Metrics:
   • Return on AI investments
   • Time to market advantage versus competitors
   • Value captured from AI innovations
   • Opportunity cost reduction from accelerated implementation
4. Cultural Change Metrics:
   • Employee perception of decision-making efficiency
   • Willingness to propose and champion AI innovations
   • Leadership confidence in governance effectiveness
   • Talent retention in AI-related roles

Continuous Improvement Mechanisms

1. Regular Governance Reviews:
   • Quarterly assessment of decision velocity metrics
   • Executive steering committee oversight of transformation
   • Identification of emerging bottlenecks and friction points
   • Prioritization of ongoing improvement initiatives
2. Feedback Integration:
   • Regular surveys of AI practitioners and stakeholders
   • Post-implementation reviews of decision process effectiveness
   • Structured capture of governance pain points
   • Formal mechanisms to suggest process improvements
3. Competitive Benchmarking:
   • Regular assessment against industry peers
   • Identification of emerging best practices
   • Gap analysis against leading organizations
   • Implementation of improvements based on external insights
4. Governance Innovation:
   • Dedicated resources for governance improvement
   • Explicit innovation objectives for governance teams
   • Regular experimentation with new approaches
   • Leveraging emerging technologies for governance enhancement

Leading the Decision Velocity Transformation

Transforming your organization’s approach to AI decision-making represents one of the most significant leadership opportunities for today’s CXOs. Those who successfully navigate this transformation will position their organizations to realize the full potential of AI investments, while those who allow organizational gridlock to persist will find their AI initiatives delivering diminishing returns amid growing frustration.

As a CXO, your role in this transformation is crucial. By championing a strategic approach to decision acceleration, aligning organization and culture with velocity objectives, and maintaining unwavering focus on business outcomes, you can ensure that your enterprise breaks free from the gridlock that constrains AI progress.

The journey requires significant commitment, organizational change, and sustained attention. But the alternative—continuing to invest in AI capabilities while allowing organizational processes to prevent their effective deployment—is increasingly untenable in a competitive landscape where speed of innovation is a primary differentiator.

By taking decisive action now, you position your organization for sustained AI-driven innovation and growth, transforming from a company where good ideas go to die to one where innovation thrives and drives competitive advantage.

For more CXO AI Challenges, please visit Kognition.Info https://www.kognition.info/category/cxo-ai-challenges/