TL;DR

• The internet is shifting from human centred use to AI agents doing tasks for us.

• New design patterns and protocols from Anthropic and Google let these agents talk directly to websites and each other.

• This creates two parallel Internets: one for people, one for machines.

• Agents skip ads and web pages, making things faster but disrupting traditional economic and business models.

• AI agents are also changing shopping, media, and education practices, creating new challenges with the efficiency gains.

• Businesses are being forced to adapt or risk being left behind.

• There are also new governance challenges: who’s responsible for agent mistakes, and will people lose important skills?

• The “agent divide” may widen the established “digital divide” between those with informed access to performant and transparent AI agents and those who don’t.

• We need fair rules, better education, and to encode within these protocols the ability for humans to maintain agency.

The Internet stands at a turning point. How we use our connected devices, browsers, mobile phones and the infrastructure of the Internet itself is being transformed by the interactions of agent-based AI systems.

Anthropic’s Model Context Protocol (MCP) and Google’s Agent-to-Agent (A2A) framework are a necessary development that supports this structural shift. What interacting with the Internet will look like will change from human-centric browsing to agent-mediated interaction—and it will transform how information is consumed, services are delivered, and power is distributed across the digital universe.

This essay argues that the convergence of protocol-driven agentic systems will lead to a parallel “agentic web” layer optimized for machine-to-machine communication, rendering traditional presentation-layer interfaces obsolete for transactional and informational tasks. This naturally raises existential questions about economic models, governance frameworks, and human agency.

I. Protocol Foundations: MCP and A2A as a Potential Digital Nervous System

The technical architecture of tomorrow’s agentic web is being built today through complementary protocols.

Anthropic’s MCP, launched in late 2024 as an open standard, solves the “M×N” integration problem that previously required custom connectors between individual AI models and data sources. By providing a universal JSON-RPC interface to serve tools, and SDKs to access resources, and prompt templates via standardized endpoints, MCP allows any compliant AI system to dynamically access external data and execute actions—from querying SQL databases to posting Slack messages—through standardized client-server interactions. Crucially, MCP’s security model enforces granular permissions and local-first execution, ensuring sensitive enterprise data remains under organizational control even as models gain contextual awareness.

On the client-side, Google’s A2A protocol, just released in April 2025, addresses the orthogonal challenge of inter-agent coordination. Where MCP vertically integrates individual AI systems with tools, A2A enables horizontal collaboration between autonomous agents through three core components:

• Agent Cards: Machine-readable metadata profiles advertising capabilities and authentication requirements
• Task Lifecycle Management: Standardised and structured workflows with states (submitted, working, completed) and artifact outputs
• Hybrid Communication Channels: Combining request/response patterns for immediate tasks with server-sent events for long-running processes

This protocol closely resembles the established microservices architecture pattern—where an MCP server functions as the API gateway layer handling external resource integration, while A2A implements the service mesh pattern for inter-service discovery and communication.

Together, they could eventually allow for the translation of a new agentic backbone for the long predicted “Fourth Web”—an internet where the majority of interactions occur between autonomous entities rather than between human users to facilitate real world outcomes.

II. Presentation Layer Obsolescence: The Dual-Web Architecture

The existential threat to traditional front-end design stems from the current fundamental mismatch between human-centric interfaces and the operational requirements of AI agents.

Where the browser once served as a universal portal for humans to consume information and services on the Internet, increasingly they are seen as a legacy system from an era when visual layout dictated information hierarchy.

This was an intuitive solution to the problem of human sense-making but is an extremely resource intensive interface for agents to currently parse.

The response, then has manifested most visibly with the rise of dual-layer architectures, where websites will increasingly need to simultaneously serve human users and machine agents, as first party, through parallel data streams, and where the browser based experienced will shrink in importance over time.

(As an aside, will we be seeing the rise of the “agent wars” like the browser wars of the 1990s-2000s? It seems likely.)

The human layer may persist for the express purpose of being a theatre of brand experience, employing rich media and interactive elements to cultivate emotional engagement. For example, luxury online retail eCommerce marketplaces like Farfetch continue to strongly invest in 3D product visualizations and virtual try-ons, recognizing that tactile experiences remain critical for high-value purchases.

But for more transactional and informational businesses with websites, there is increasingly, an optimized data layer available for agent consumption. News media publishers, such as The New York Times, The Guardian, and Reuters, provide APIs that deliver structured news content. These APIs are increasingly being designed not just for human developers, but for automated agents and AI systems to ingest, process, and redistribute content in new contexts (e.g., news aggregation, voice assistants, chatbots).

This bifurcation is a response to the divergent experiential vs. data requirements.

Human-facing pages prioritize aesthetic polish with median page weights now exceeding 3MB (and many pages exceeding many multiples of this).

Agent feeds deliver equivalent information in sub-10KB payloads.

The engineering implications could be stark: using dual architectures would lead to resources more appropriately deployed to minimize API response times while reducing cloud infrastructure costs compared to traditional sites which see non-human user agents as adversarial, instead redirecting them to these much more efficient data feeds, which also increase the prominence of this data in generated outputs.

In my mind, this transition also mirrors earlier platform shifts—desktop to mobile first design requiring much more compact methods of data presentation—but with greater speed (and isn’t a JSON response, right now, the ultimate compact presentation of data?).

Google’s 2025 Web Vitals update penalizes sites lacking structured data feeds, and there have already been comparative studies published showing machine readable protocols such as Anthropic’s MCP protocol led to preferential indexing.

Enterprises face a paradoxical design mandate: human interfaces must become more visually immersive as their functional purpose changes, even as their importance diminishes.

III. Industry Disruption Vectors

Media & Advertising: The Attention Economy’s Unravelling

The challenge to programmatic advertising’s current methods was foreshadowed with cat and mouse ecosystem with ad-blockers but it has accelerated under agentic mediation.

When ChatGPT’s “Link Previews” feature reduced publisher click-through rates by 58% in 2024, it exposed the fragility of impression-based models. AI agents now bypass display ads entirely, extracting needed information while discarding promotional content. This carving away of attention from the browser presentation layer to interfaces closer users by agentic content aggregation and synthesis threatens the foundations of data-driven quantitative marketing and media business models with obsolescence:

1. Attention Metrics Collapse: Tools measuring dwell time, scroll depth, and click-through rates lose relevance as AI agents bypass visual interfaces entirely. A hotel booking agent, for instance, extracts pricing and availability data without loading a single pixel of a webpage.
2. Performance Optimization Paralysis: A/B testing frameworks built on human behaviour patterns fail when agents make decisions algorithmically. Conversion Rate Optimization (CRO) tools like Hotjar, which rely on session recordings and heatmaps, cannot interpret agent interactions that occur via API calls.
3. SEO’s Existential Crisis: Keyword rankings and backlink strategies collapse as agents prioritize structured data feeds over less efficient HTML content.
4. Front-End Irrelevance: Interfaces optimized for human engagement (e.g., responsive design, interactive elements) become vestigial quaint manifestations of brand as most of attention is piped from agents aggregating and presenting machine-readable payloads.

In response, media conglomerates have started experimenting with three survival strategies:

1. Provenance Watermarking: The use of Blockchain or similar cryptographic protocols, could be pivotal in this regard, by embedding signatures into content, allowing agents to verify authenticity while generating summaries. Studies have shown this approach also increases prominence in agent-generated summaries (a fantastic rundown of current approaches: https://equilibrium.co/writing/blockchains-cryptography-authenticity-challenge).
2. Microlensing APIs: Axel Springer’s per-summary licensing model mirrors Spotify’s micro-royalty approach. Despite initial agent traffic drops, the program stabilized at 54% of pre-monetization levels within six months.

We are just at the start of reinventing monetization models for agent-based interactions, and so there are still problems in implementing these current approaches.

When The New York Times attempted the same Axel Springer inspired dynamic paywalls for agent access—$0.01 per summary for basic content, $0.05 for investigative reports—it spawned legal suits over fair use exemptions.

The ensuing court battle highlights the precarious balance between the sustainability of media creators and the voracious appetite of AI companies and the models they train (that power these agentic interactions).

Implications for Marketing Economy:

• Agencies: SEO and CRO consultancies face obsolescence unless they pivot to MCP-compatible structured data engineering to present data to agents in ways that maximize prominence.
• Platforms: Google Analytics and similar tools must reinvent themselves as agent-traffic auditors, tracking API calls rather than page views.
• Metrics: Attention-based KPIs give way to agent trust scores and API call yield—metrics measuring how often agents select a brand’s data feed over competitors.

Businesses clinging to human-centric optimization risk becoming digital fossils, while those embracing agentic traffic could gain first-maker advantages.

eCommerce: The Quantification of Commerce

The agentic transformation of eCommerce mirrors the seismic shift that occurred in financial markets during the rise of quantitative trading. Just as Wall Street’s floor traders yielded to algorithm-wielding quants in the 2010s, today’s “Add to Cart” interfaces are being supplanted by MCP-powered agent negotiations—a transition that centralizes market power among technologically equipped firms while marginalizing analogue competitors.

In quantitative finance, firms like Renaissance Technologies achieved dominance by replacing human intuition with algorithmic arbitrage, capturing micro-inefficiencies invisible to traditional traders.

Similarly, Amazon’s investor earnings reports revealed a growing proportion of transactions now occurring through APIs, bypassing traditional cart flows, leading to faster checkout times compared to human-driven purchases.

This agentic commerce concentrates gains among API-enabled enterprises.

Small businesses face existential pressure reminiscent of smaller traditional brokerages during the quant revolution starting in the 1960s. SMBs lacking the internal technical resources to implement MCP/A2A endpoints, mean they are invisible to agent-driven markets. Efficiency gains will then accrue disproportionately to vendors who control the protocols through which these transactions flow.

This evolution demands a re-evaluation of antitrust frameworks for the API age, where market power resides not in inventory, or selection, perhaps even in logistical efficiency, but in protocol dominance.

Education: Pedagogical Foundations Undermined

The integration of AI tutors and assignment assistants has triggered a paradigm shift in educational practice, demanding a re-evaluation of core pedagogical principles.

There is of course the current cat and mouse game between students using ever more sophisticated writing agents, and educators creating ever more convoluted ways to detect and dissuade students from using them. However, the opposite situation of embracing AI as a part of the learning process also requires careful adjustment of teaching methods as it introduces problematic dynamics to learning efficacy.

Harvard’s 2025 pilots showed AI tutors doubled student engagement and learning gains compared to traditional lectures.

Troublingly, though, educators have repeatedly stated fears of students demonstrated agentic dependency- in that they cannot explain agent-generated output, creating a surface-level competency that masks deficiencies in deeper understanding.

This disruption requires a more thorough and nuanced analysis than this essay and my expertise can provide, but what these experiments demonstrate is that the integration of AI will require a philosophical recalibration of the pedagogical tradition.We’re not assessing knowledge retention anymore, but knowledge sovereignty—a student’s capacity to govern AI tools without being governed by them.

IV. Power Reconfiguration: Protocol Wars & New Hegemons

If we think about the transformation of the Internet from a first principles perspective, software applications were originally created to facilitate humans to interact with, manipulate and transmit data, with networks (eventually the Internet) being the pipes through which data was transmitted.

In this ecosystem, power resided in platforms that allowed humans to either build these applications or consume them-the application layer.

Agentic systems disrupt this model. While power still resides in those who control the data layer, and these are largely the same platforms, because everything above this layer becomes autonomous, such an ecosystem (e.g. MCP/A2A) favours players controlling protocol layers over traditional IaaS and SaaS platforms:

Anthropic’s decision to open-source MCP client libraries mirror Google’s Android playbook at the beginning of the smartphone revolution—that is to win through ecosystem ubiquity creating a monetizable marketplace, rather than directly monetization user consumption, demonstrating the prioritizing of protocol control over application-layer dominance.

Other governments are also proposing different regulatory frameworks.

For example, China’s “Great Firewall” has all domestic A2A traffic route through state-approved hubs, creating a splinter agentic web, following the same network pathways as the currently splintered web ecosystem.

In the EU, the Digital Services Act now requires agent transparency reports, forcing disclosure of content summarization policies. This is on top of other EU regulations like GDPR and the AI Act, which will likely be extended to A2A agents.

This segue’s us into the next section on the challenges of governing such an agentic ecosystem.

V. Societal & Governance Challenges

The Accountability Vacuum

When a medical diagnostic agent misclassifies tumours using outdated MCP-connected lab data, liability becomes fragmented. Does responsibility lie in:

• The hospital’s data governance team for failing to update feeds?
• The AI developer for inadequate temporal validity checks?
• The protocol maintainer for permitting insecure API handshakes?

The 2024 EU v. OpenAI lawsuit established precedent through a $2.3M settlement over hallucinated legal advice, but broader questions remain unresolved. Proposed solutions include:

• Agent Notarization: Blockchain-sealed audit trails documenting decision logic, data sources, and confidence metrics. Estonia’s e-Governance Academy now mandates this for public-sector AI deployments.
• Liability Tokens: Parametric insurance smart contracts triggered by error type and severity. Lloyds of London’s 2025 “AI E&O” product covers prompt injection attacks up to $50M.
• Federated Oversight: Interpol’s “AI Agent Watch” program coordinates cross-border incident response, though participation remains voluntary.

Digital Literacy’s New Frontiers

A 2025 Stanford study exposed alarming disparities in agent proficiency. While 73% of high-income users customize workflows via natural language prompts, 89% of low-income users accept default configurations uncritically.

Mozilla’s does have the “Agent Literacy Project” which provides community based upskilling covering:

• Prompt engineering for critical thinking
• Bias detection in agent outputs
• Privacy-preserving configuration

Early adopters in Detroit and Nairobi show 41% gains in agent-assisted task efficiency, but scaling remains constrained by income derived infrastructure gaps. This socio-economic “prompt divide” risks encoding algorithmic performance differences at population scale, extending the current “digital divide” into the agent usage as well.

Human Agency’s Erosion

The phenomenon of “agentic learned helplessness” emerges when users outsource cognitive functions to AI, but this trend builds upon a broader pattern of cognitive erosion driven by algorithmic content delivery systems.

Social media platforms, e-commerce sites, and search engines have already been shown to reduce critical thinking capacities through their dopamine-driven engagement models. Studies reveal that excessive social media use correlates with decreased grey matter density in the prefrontal cortex and impaired working memory performance, while constant digital distractions erode attentional control and information retention.

This suggests agentic AI systems may compound rather than create cognitive decline, transitioning users from passive content consumption to active delegation of decision-making authority.

Countermeasures require deliberate friction:

• Cognitive Checkpoints: South Korea mandates confirmation delays for AI-mediated financial transactions over $10,000.
• Agent Gradients: Educational platforms like Khan Academy gradually reduce assistance as learner competency grows.
• Epistemic Tagging: The W3C’s “Truthiness” standard requires visual indicators distinguishing human vs. AI content.

VI. Pathways Forward: Architecting the Agentic Future

The transition to an agent-mediated internet demands coordinated action across three societal pillars: enterprises, policymakers, and civil society. Each faces distinct challenges in balancing operational efficiency with ethical responsibility, technical innovation with human dignity.

Enterprise Strategy - Beyond API-First to Agent-Centric Design

Modern enterprises must reorient their digital strategies around two parallel realities: the persistence of human-brand relationships and the ascendancy of machine-driven transactions. This dual mandate necessitates agent-centric design—an architectural philosophy prioritizing structured data feeds and API ecosystems alongside traditional user interfaces.

As mentioned, leading publishers like Reuters exemplify this approach through content atomization, decomposing articles into standalone “knowledge units” enriched with semantic metadata. Each unit includes temporal validity windows, source attribution chains, and confidence intervals for statistical claims.

However, the transition risks alienating human audiences if executed haphazardly. Like Farfetch, Sephora’s Virtual Artist app still offers a human centred experience where users can digitally try on makeup products with their skin tone, while a backend agentic system uses product data and user preferences to optimize recommendations and suggest to the business inventory management.

Enterprises lagging in API modernization to serve agentic systems face existential risks.

Governing

As MCP and A2A and similar protocols become digital public infrastructure, policymakers must balance innovation incentives with safeguards against centralized control. The EU’s Digital Services Act does this through three key mandates:

1. Protocol Neutrality: Requiring interoperability between competing agent standards, akin to telecom interoperability rules. This prevents “walled garden” ecosystems where Google’s A2A agents privilege Google Cloud services.
2. Agent Literacy: Supporting explicit education and training programs on AI interaction, covering prompt engineering and bias detection.
3. Transparency: Directing LLM developers to require public disclosure of external auditing of models, especially for critical workloads.

At this very early stage most governments have no opinion on the matter, but those nations that do, have wildly divergent strategies.

China’s “Great Agent Wall” routes all A2A traffic through state-sanctioned hubs, enabling real-time content filtering. The EU treats AI adversarially under its AI Act—categorizing systems by threat level and mandating strict compliance checks for high-risk applications through centralized bodies like the AI Office. The US adopts a lighter touch, self-regulated approach, relying on sector-specific guidance from agencies like the FTC and voluntary corporate commitments.

Democratizing Agent Use

Non-state actors face dual responsibilities: holding corporations and governments accountable while pioneering inclusive agent ecosystems.

The Solid Project “Agent Pods” (https://solidproject.org/) are an early example of how open-source can lead to the creation of user-controlled data structures or vaults allowing granular permissions, for example:

• Restricting healthcare agents from sharing diagnostic data with insurers
• Allowing eCommerce agents to temporarily access purchase histories
• Revoking access to deprecated agents automatically

As previously mentioned, Mozilla’s “Agent Literacy Project” complements this technical infrastructure with community education.

Synthesis: The Tripartite Mandate

The agentic web’s evolution hinges on interdependent advancements:

1. Technical: Enterprises must treat APIs as core revenue channels, not tack-on afterthoughts to internal IT systems.
2. Regulatory: Policymakers should view protocols as critical infrastructure that require both regulatory oversight, interoperability standards and policy harmonization.
3. Social: Civil society must position itself as informed users, not passive consumers as social media platforms have conditioned, and lean on open-source initiatives to democratize agent use.

Failure on any front risks bifurcation into a digital caste system: agent-empowered elites who can leverage the required protocols, while marginalized groups navigate decaying traditional web applications and increasingly static HTML pages.

The path forward isn’t merely technical adoption but cultural change. As Stanford’s 2025 Digital Civil Society Index concludes: “The measure of agentic success isn’t task completion rates, but whether human dignity scales with machine efficiency”. In this light, the protocols we encode today become the inescapable architectures enabling the future of the Internet, and shaping what it means to know, decide, and interact with the digital world.

Conclusion: The Unbundling of Interaction

The emergence of the agentic web, powered by protocols like Anthropic’s MCP and Google’s A2A, heralds a transformation as profound as the invention of the World Wide Web itself. No longer a passive repository of human-generated content, the internet is evolving into a dynamic ecosystem where machine agents negotiate, transact, and curate reality on behalf of users, organizations, and, increasingly, for themselves. This shift dismantles long-standing assumptions about primacy of the human in human-computer interaction, economic value creation, and societal governance, demanding a reimagining of digital life.

At its core, the agentic web represents a duality: a bifurcation into human-centric experiences and machine-optimized data flows. The presentation layer, once the crown jewel of digital design, now serves a diminishing role as an experiential surface, and less favoured for transactional and informational tasks. Yet this duality is not merely technical—it reflects a deeper epistemological shift. Truth itself becomes protocol-dependent, with blockchain watermarks and cryptographic provenance tags serving as the new arbiters of authenticity as the web becomes a swamp of infinitely plentiful synthetic content.

Whole parts of society and industry sectors face existential recalibration. Media’s reliance on attention economies will crumble as agents bypass ads and paywalls, forcing publishers to monetize through microtransactions and API licensing. eCommerce is morphing already into hyper-efficient agent marketplaces where AI intermediaries broker deals across federated supply chains, privileging enterprises with MCP-integrated infrastructure. Education grapples with the paradox of personalized tutoring at scale, where cognitive offloading risks eroding critical thinking in our future generations, even as accessibility inexorably expands.

Power dynamics will undergo realignment. Traditional gatekeepers like Google and Amazon face disintermediation from protocol-centric players—such as Anthropic with its open MCP ecosystem and Microsoft through strategic acquisitions like MCPHost.

The Internet will continue to balkanize as China’s “Great Agent Wall”, the EU’s transparency mandates, and increasingly techno-feudalistic US-centric web, splinter into jurisdictional fragments. In this landscape, control over protocols supplants dominance of platforms, with protocols the agentic equivalent of maritime trade routes: invisible yet inescapable conduits of power.

Societally, the agentic web poses dilemmas that defy conventional governance. Liability frameworks strain under the weight of fractal accountability chains, where a single misdiagnosis by a medical agent implicates data providers, model trainers, and protocol designers. The “prompt divide” exacerbates inequality, as privileged users sculpt bespoke agent workflows while marginalized groups uncritically accept opaque outputs. Most critically, human agency itself hangs in the balance—neurological studies already show atrophy in decision-making regions among habitual users, foreshadowing a future where the agency of human decision-making is ceded to agentic surrogates.

These are the challenges of the agentic revolution we are currently undergoing, no doubt, akin to the challenges the industrial and digital revolutions similarly created.

The path forward demands symbiotic advancement across three axes: technical, regulatory, and cultural.

As nascent protocols slowly rise to primacy, they will become more than technical standards—their architectures will determine whether the agentic web elevates human potential or entrenches algorithmic determinism. The choice is not between progress and preservation, but between building ecosystems that scale efficiency with dignity, and those that sacrifice humanity for optimization. They will be the social contract of the 21st century: one that doesn’t just govern machine intelligence but encodes the irreducible value of human agency.

Citations

1. https://wandb.ai/onlineinference/mcp/reports/The-Model-Context-Protocol-MCP-by-Anthropic-Origins-functionality-and-impact–VmlldzoxMTY5NDI4MQ
2. https://www.infoq.com/news/2025/04/google-agentic-a2a/
3. https://meetcody.ai/blog/top-ai-web-browsing-agents/
4. https://coin-workshop.github.io/coine-2023-london/Papers/Paper-10.pdf
5. https://www.linkedin.com/pulse/what-ai-agents-could-make-internet-disappear-heres-means-our-cuzhf
6. https://www.decidr.ai/blog/the-three-eras-of-the-agentic-web
7. https://www.sciencespo.fr/psia-innovation-hub/news/five-aims-lessons-from-internet-governance-for-artificial-intelligence-management-strategies/
8. https://www.math.pku.edu.cn/teachers/linzq/teaching/stm/references/Agent-mediated%20Electronic%20Commerce%20-%20A%20Survey.pdf
9. https://www.governance.ai/analysis/computing-power-and-the-governance-of-ai
10. https://github.com/google/A2A
11. https://openai.com/index/browsecomp/
12. https://www.linkedin.com/pulse/revolutionising-enterprise-architecture-governance-genai-munish-gupta-htm6c
13. https://www.the-blueprint.ai/p/web-40-the-rise-of-the-agentic-web
14. https://www.diplomacy.edu/blog/reducing-terminological-confusion-is-it-digital-or-internet-governance/
15. https://www.imbrace.co/future-trends-in-agentic-ai-development-whats-next-for-intelligent-automation/
16. https://www.aiacceleratorinstitute.com/agent-responsive-design/
17. https://www.anthropic.com/news/model-context-protocol
18. https://docs.anthropic.com/en/docs/agents-and-tools/mcp
19. https://www.theregister.com/2025/04/21/mcp_guide/
20. https://www.linkedin.com/pulse/autonomous-agents-anthropic-mcp-gives-glimpse-future-vikram-ekambaram-gcvfc
21. https://www.reddit.com/r/ClaudeAI/comments/1gzv8b9/anthropics_model_context_protocol_mcp_is_way/
22. https://www.codiste.com/how-ai-agents-are-changing-digital-marketing
23. https://a16z.com/a-deep-dive-into-mcp-and-the-future-of-ai-tooling/
24. https://www.youtube.com/watch?v=GUozMSpnmcc
25. https://www.computerworld.com/article/3837918/opera-adds-ai-agent-to-its-browser.html
26. https://pubs.opengroup.org/togaf-standard/adm-practitioners/adm-practitioners_15.html
27. https://www.forbes.com/sites/aytekintank/2025/03/20/how-ai-agents-can-help-you-manage-your-social-media-in-2025/
28. https://www.runtime.news/why-ai-infrastructure-companies-are-lining-up-behind-anthropics-mcp/
29. https://www.crayon.com/resources/insights/visions-2025-data-and-ai/
30. https://www.deloitte.com/lu/en/our-thinking/future-of-advice/agentic-ai-the-new-frontier-in-ai-evolution.html
31. https://academic.oup.com/jcmc/article/25/1/89/5714020
32. https://carnegieendowment.org/research/2024/10/the-ai-governance-arms-race-from-summit-pageantry-to-progress?lang=en&center=europe
33. https://www.forrester.com/blogs/interoperability-is-key-to-unlocking-agentic-ais-future/
34. https://www.ibm.com/think/topics/ai-governance
35. https://www.jstor.org/stable/27750935
36. https://www.linkedin.com/pulse/platform-power-shift-how-ai-rewrite-rules-digital-titans-tony-moroney-azzsc
37. https://www.forbes.com/sites/clorischen/2024/12/23/ai-agents-101-the-future-of-agentic-web-and-onchain-ai/
38. https://policyreview.info/articles/analysis/contesting-public-interest-ai-governance
39. https://www.sciencedirect.com/science/article/abs/pii/S0953543802000280
40. https://architecture.digital.gov.au/responsible-use-of-AI-in-government
41. https://www.presentations.ai
42. https://www.insightpartners.com/ideas/ai-agents-disrupting-automation-and-reimagining-productivity/
43. https://www.digitaleducationcouncil.com/post/solving-the-ai-governance-problem-what-should-institutions-look-out-for
44. https://www.cprime.com/resources/blog/building-resilient-enterprises-with-ai-powered-adaptation/
45. https://www.visme.co/ai-presentation-maker/
46. https://bernardmarr.com/forget-chatgpt-why-agentic-ai-is-the-next-big-retail-disruption/
47. https://www.youtube.com/watch?v=Ax9qNfL9BGQ
48. https://slidespeak.co/blog/2024/03/15/3-ways-to-design-powerpoint-slides-with-ai/
49. https://www.cnet.com/tech/services-and-software/you-can-create-a-powerpoint-presentation-with-ai-heres-how-i-did-it/
50. https://www.aitoolssme.com/comparison/ai-tools-for-presentations
51. https://www.linkedin.com/pulse/ai-governance-ethics-education-charting-responsible-innovation-gavor-u0b5f
52. https://geodesiccap.com/insight/ai-efficiency-the-next-frontier-for-enterprise-ai-adoption/
53. https://www.presentations.ai/compare/tome-vs-presentations-ai
54. https://www.linkedin.com/pulse/forget-genai-hype-ai-agents-2025s-disruption-pr-industry-david-ko-ovypc
55. https://www.finance.gov.au/sites/default/files/2024-06/National-framework-for-the-assurance-of-AI-in-government.pdf
56. https://www.linkedin.com/pulse/enterprise-ai-layer-david-armano-jvslc
57. https://www.superside.com/blog/ai-presentation-makers
58. https://www.flywheeldigital.com/blog/ai-agents-commerce-retail-media-impact
59. https://ai.uq.edu.au/project/ai-and-government-accountability
60. https://www.datastax.com/blog/ai-agents-to-blow-up-the-business-process-layer
61. https://www.forbes.com/sites/jeannemeister/2025/02/15/the-rise-of-the-hybrid-workforce-humans-and-ai-working-together/
62. https://www.salesforce.com/au/news/stories/agentic-maturity-model/
63. https://www.auda.org.au/public-impact/internet-governance-and-public-policy/internet-governance-roadmap/
64. https://solace.com/blog/patterns-event-apis-unified-apim/
65. https://cadeproject.org
66. https://www.linkedin.com/pulse/ai-agents-enterprise-technology-ecosystem-strategic-perspective-rao-0gqec
67. https://www.datarobot.com/blog/agnostiq-agentic-ai-app-development/
68. https://www.linkedin.com/pulse/chapter-2-evolution-work-from-hierarchies-hybrid-teams-jas-dhillon-jgqqc
69. https://www.sprinklr.com/blog/ai-agents/
70. https://www.pwc.com/us/en/tech-effect/ai-analytics/ai-agents.html
71. https://www.insightpartners.com/ideas/state-of-the-ai-agent-ecosystem-use-cases-and-learnings-for-technology-builders-and-buyers/
72. https://www.infrastructure.gov.au/sites/default/files/documents/noetic-final-report-internet-governance-system-map-mapping-influence-and-interactions-in-the-world-of-internet-governance-october2023.pdf
73. https://www.linkedin.com/pulse/5-considerations-when-implementing-multi-layered-api-security-hodges
74. https://legalinstruments.oecd.org/api/download/?uri=%2Fprivate%2Ftemp%2Fd7bce2d4-a43b-4699-9fb0-3d7a086f6720.pdf&name=Action+Plan+on+Transforming+Public+Governance+for+Digital+Democracy.pdf
75. https://www.astera.com/type/blog/ai-agents-future-of-work/
76. https://quiq.com/blog/agentic-ai-cases/
77. https://www.sipa.columbia.edu/sites/default/files/2022-09/The%20Rising%20Geopolitics_2016.pdf
78. https://www.computerweekly.com/feature/Enterprise-strategies-for-API-management
79. https://wacsi.org/unmasking-digital-colonialism-empowering-civil-society-organisations-in-the-digital-age/
80. https://www.cgi.com/en/blog/artificial-intelligence-the-ai-agent-revolution-why-enterprises-must-act-now