Agentic Web Executive Briefing 

AI

Agentic Web Executive Briefing 

Prepared for: Board and C-Suite Level Executives 

Prepared by: Danny Denhard, Coach, Consultant, Advisor 

Date: Released 28/05/2025 - Updated November 16th 2026

Read Time: 5 minutes - Internal Q&A will be needed.


Critical Industry Shift Overview

Before We Start: Agentic, Agents & Agent Swarms Explained

Every board is hearing these three words, often used as if they mean the same thing. They don’t. Getting the difference right changes where you invest, what you govern and how fast you move.

Agentic: AI That Gets Things Done

“Agentic” describes AI that acts, not just answers. A chatbot tells you how to do something. Agentic AI is given a goal, works out the steps, uses tools such as browsers, apps, payment systems and your company data, and completes the task. It does this with some autonomy and checks back with a human only when it needs approval. The shift is from AI as an advisor to AI as an operator.

In November 2026 OpenAI launched ChatGPT Wallet with Visa, Mastercard and American Express. A user asks for something, the agent finds it and builds the basket, the user approves with Face ID, and the agent pays using a stand-in card number. Search, decision and payment happen in one flow, without the customer ever visiting a website.

Agents: Your “New Digital Workforce”

An agent is a single AI worker built to do a specific job. It has a role (such as shopping, finance or scheduling), access to certain tools and data, and rules about what it can and can’t do without permission. The best agents specialise: like a good hire, an agent that goes deep in one area beats a generalist that knows a little about everything. Agents can work for your customers, your employees or your business.

Meta’s Muse agent handles email, bills, reservations, groceries and purchases for its users, and it passed 10 million US downloads in its first 40 days on the App Store. Instinct is a popular tool that has been adopted by millions of early adopters.
In the UK, Alexa for Shopping now builds a personalised shortlist of about seven products for each shopper. Your next customer may never visit your site. Their agent will, or it will decide not to.

Agent Swarms: Teams Of Agents Working Together

A swarm is a group of specialist agents working together on one larger goal, usually coordinated by a lead agent. They split the work, share information, check each other’s output and run around the clock. Swarms are where agent-to-agent (A2A) coordination pays off: they can take on the back-and-forth, hand-offs and status updates that slow organisations down. They also multiply risk. Each agent needs clear permissions, and the swarm as a whole needs a human accountable for what it does.

A potential scenario: A customer complains at 11pm that a delivery is late. A service agent logs the problem. A logistics agent tracks the parcel and books a new delivery slot. A finance agent applies a refund within pre-approved limits. A CRM agent updates the customer record and sends a personalised apology. A reporting agent flags the courier’s failure to the operations lead for the 9am stand-up. Five agents, one outcome, no human involved until the morning review.

Agentic is the shift, agents are the workers, and swarms are how the work will scale. For brands, customers’ agents will increasingly decide what gets found, compared and bought. For organisations, agents and swarms will change how work gets done. The rest of this briefing covers what that means for your business, and what to do in the next 6 months.

Current Web To Agentic Web

Timeline: The original 18-month window (May 2025 to November 2026) has now run its course. The groundwork (protocols, payments and native checkout) is largely in place. The next 12 months are about adoption, specialist agents and agent-to-agent coordination. Airbnb has said it will roll out a travel agent in Q1 2027.

Magnitude: High impact. Industry-wide change is required, from opening up bot access (most bots are currently blocked by tools like Cloudflare) to amending web protocols (APIs, etc.) and wider technology changes. Once the agentic web starts, it will not stop.

What’s Happening:

  • A permission-based web protocol is being enabled as part of the AI technology shift.

  • Microsoft is leading the charge on letting agents (bots) complete tasks such as researching, downloading and transacting on websites and apps, including agent-to-agent (A2A). Microsoft will use Anthropic’s (Claude) MCP protocol and build on top of other protocols to enable agentic transactions.

  • Adoption is low to non-existent so far. This is new technology, but it will accelerate with the huge investments being made in AI, led by Microsoft.

  • Microsoft appears to want to police the agentic web in a similar way to how Google polices search today.

  • Google has released its AP2 framework, which it describes as “a trusted and secure foundation for the future of agentic commerce”. Google has 60 partners, from banks and BNPL platforms to financial institutions, providing the guardrails and new infrastructure to enable transactions and protect users.

  • OpenAI (ChatGPT) and Stripe have partnered to enable in-chat commerce, so Stripe will power purchases inside ChatGPT.

    • Etsy has also partnered with Stripe to enable agentic commerce through instant checkout via Stripe’s Agentic Commerce Protocol (ACP).

  • PayPal and Mastercard have partnered to power agentic payments and provide infrastructure for handling mistakes and payment errors.

  • PayPal and ChatGPT have also partnered to enable direct purchases in ChatGPT sessions. PayPal could power commerce straight from prompts: searching in ChatGPT on the web, in the app or in the Atlas browser, through to buying books, entertainment and equipment. Looking further ahead, it could include paying ChatGPT to build apps on your behalf.

  • In January 2026 Google and Shopify announced new native checkouts and dynamic pricing options

From One Assistant To An Army Of Agents

AI agents are moving from one “general” assistant to a connected set of specialised agents, each responsible for one area. Models are now good enough to take real action, not just answer in chat. Agents will join email threads, propose times, book meetings and complete multi-step tasks from start to finish.

The five categories of agent to plan for:

  1. Generalist assistants: the broad, consumer-facing assistants

  2. Work-oriented agents: built into Teams, Slack and work suites

  3. Multiplayer agents: groups of agents working around the clock across teams to complete tasks

  4. News & network agents: internal listening agents that track the flow of information, bottlenecks and friction between colleagues and teams

  5. Travel agents: the biggest use case, likely controlled by the largest travel companies

Why this matters to the board:

  • Work beats home. The clearest ROI is inside organisations, where agents remove manual tasks, unblock teams and cut coordination overhead.

  • A2A is the real unlock. Each person’s agent can negotiate the details, swap the right information and come back with a decision to approve. Long email chains and large status meetings are the first targets. A2A will not remove office politics (PQ), and people will try to use agents to keep control.

  • Specialists win over generalists. Agents that go deep in one area build richer context than a general assistant can. Expect agents in specific industries, built for specific jobs.

  • Discovery will be centralised. Agents will spread through referrals, recommendations and product-led growth, and most likely through an agent app store. Apple and Google are best placed to run one. We need a plan for how our brand, or our own agent, gets found there.

Keep the hype in check: Airbnb’s latest earnings call reported “very non-material” traffic from AI assistants. Its users look at dozens of listings and 25+ photos, 90% message the host, and 80% of travellers say they like doing their own research. Agentic behaviour is coming, but how quickly customers adopt it will differ by category. We should plan for it without over-investing ahead of real demand.

Strategic Implications For Our Business

Opportunities

  • The market will grow over time, so a long-term focus is vital. More productive people mean a bigger market and more brand and technology opportunities.

    • Being seen as the go-to brand for core products in our space matters more than ever for revenue growth, market expansion and protecting any moats we think we have.

  • Customers will follow the usual crossing-the-chasm cycle. Innovators and early adopters will invest a lot of time and money in automating tasks and building the best agents. The early majority may be pushed into adopting early so they aren’t left behind on productivity and cost savings. Customers will expect the brands they know, like and trust to lead the market.

Short term:

  • Improve our technology and tech stack.

  • Build our brand: how it is considered in product searches and how it shows up in chat-based answers (known as AEO, AI answer engine optimisation, or GEO, generative engine optimisation).

  • Invest in hiring the right full-time staff, and use external consultancies to increase our capabilities. Many team members will struggle to adapt, and some may leave.

  • Find ways to build AI into our planning and develop new revenue models. While competitors struggle with how the web is changing, it’s vital to win the early battle.

  • Decide whether we will opt into chat-based commerce, or keep our own website and app as the preferred place to shop.

Mid term:

  • Build a high-trust brand, grow revenue and repeat customers (both agents and humans), and get early ROI on AI investments and automation.

  • Expand our scope. There will be chances to scale, whether in supply chain, national and international ordering and fulfilment, or in some cases a fully translated global experience.

Long term:

  • A bigger SAM and SOM. Market growth and higher productivity are likely to mean more orders and more effective marketing.

  • Faster revenue growth through 24/7 autonomous operations and a better customer experience.

The Buyer Journey Is Breaking, And Trust Is The Prize

AI is changing how customers decide, not just where they buy. The old journey ran Educate → Want → Need → Action, over one week to six months. Generative AI skips the Want and Need stages: customers prompt, get a confident answer and feel the problem is solved.

That answer usually turns out to be “good enough” and no more. Progress stalls, advice conflicts and trust in AI drops. The buying cycle stretches by another 2–12 weeks. When buyers come back to the market, they are:

  • Better informed: they already know the language

  • More precise: they know what hasn’t worked

  • Higher intent: they want the right solution quickly, not cheap answers, and they go straight to trusted brands, referrals and recommendations

What this means for us:

  • Upside: better-qualified customers, less basic education needed, stronger pricing power for real expertise and trusted brands

  • Downside: longer sales cycles, time spent undoing bad AI advice, and lumpier revenue from delayed decisions

  • Action: Stop leading with content an AI model can generate. Build content around what happens after generic advice fails. Invest in high-trust channels: referrals, case studies and a distinctive point of view. Decide deliberately which content we let AI use and which we keep for ourselves.

This answers one of our open questions, who do we market to: humans, agents, or both? We need to be visible and machine-readable for agents. We also need to be the trusted human brand customers come back to when agent answers fall short.

Threats

Short term:

  • Poor technical implementation and low trust (including negative press) will damage trust and hit transactions.

  • Customer support must be fast and effective. Agents will expect real-time error handling and solutions, or they will go to another site or agent to buy.

  • Hacking attempts will increase and become more sophisticated.

  • We could over-invest early on the back of the audit and strategic review.

  • A big increase in tech and infosec spending may be premature while the market matures.

  • Under-investing will lead to poor results and lost revenue.

  • No verification or vetting system is planned yet. One will need to be built, or verification will be controlled by a big tech player like Google, Apple or Microsoft.

  • New human and agent behaviours will cause more problems: card rejections, chargebacks (unwanted actions, or the wrong tasks given by a human to an agent), incorrect attribution, and new CRM setups (agents connected to customer accounts and approved).

Mid term:

  • Poor risk calculation and implementation will mean high fraud, or claims from agents and from humans who didn’t know about a transaction.

  • We will need to answer: who do we market to? Humans, agents, or agents that handle and filter for humans?

  • Heavy discounting and a race to the lowest price, as automated systems are very good at finding savings, the lowest price and the most efficient way to buy items or plan trips.

  • If we are slow to market or slow down agent actions, we will lose sales and market share. It will also affect how LLMs see us and how they guide agents, and customer satisfaction will drop quickly.

  • Internal agent risk: Network and listening agents that scan Slack and Teams raise questions about privacy, data ownership and who can access what. Some people will use agents to improve how they are seen, which is politics by another means. HR, IT and Legal need to set the rules before these tools arrive.

  • Rushing straight to agents: Deploying agents on broken processes leads to stop-start projects, wasted budget and lower confidence in AI across the business.

Required Response: Run It As An AI Transformation, Not A Tool Rollout

Most AI programmes fail for the same reasons. They are led by tool choice, run in departmental silos, have no senior owner, and the business never defined what “better” looks like. The response has to be a business transformation that happens to involve AI.

The 6-step transformation framework:

  1. Map your workflows. Document how work actually gets done, including the “unknown knowns”: the informal processes and the one person everyone goes to. Turn this into a central company context document.

  2. Identify opportunities. Sort tasks into four buckets: Accelerate (works but too slow, e.g. reforecasting and proposals), Optimise (bottlenecks, e.g. lead handovers and onboarding), Cure/Fix (inconsistent quality, e.g. customer service responses) and Kill (manual work to automate or remove, e.g. CRM data entry).

  3. Plan training and risks. Run AI workshops, training and hackathons. Map the risks before the benefits, then model the revenue gain or loss over time so the CFO can sign it off.

  4. Assign ownership. Set up a cross-functional AI SWAT team with a named senior leader and a RACI. Data and context requirements come out of this step.

  5. Apply governance. Use a risk framework. Low risk: drafting, summarising and brainstorming. Medium: analysis and recommendations. High: agents that publish, email customers, transact or commit funds, which always need senior human sign-off. This is where agent permissions and the “consented actions” of agentic commerce sit.

  6. Execute in sprints. Now (0–60 days): form the team, map workflows, agree guardrails and pick 2–3 use cases. Next (60–90 days): run pilots and hackathons and measure them. Future (90–180 days): scale, connect systems across teams, and run the AI plan alongside the main roadmap rather than inside it.

On top of the framework, for the agentic web specifically:

  • Specify the new tech architecture and assign a new budget. Reforecast and bring spend forward where pilots prove value.

  • Run a full strategic review and reset priorities once the tech audit is done. Expect many mid-to-high effort, mid-to-high impact tasks to be reassigned.

  • Work with your fintech and payment providers to understand AP2, UCP and ACP, and the role browsers will play. Decide where you opt into chat-based commerce and where you protect your own website and app as the preferred place to buy.

  • Prepare Marketing, Growth and CRM flows for both human and agent customers. Train customer support for real-time, agent-speed problem handling, and update any AI already in use to cope with new kinds of questions and interactions.

  • Plan for workforce change. Experienced leaders with transformation experience are critical, and new hires may be needed to handle changes in projects and scope.

  • Put AI and agent governance frameworks and compliance in place now. Regulation is uncertain, so engage early and stay adaptable.

  • Start smaller, in areas that scale: customer service, CRM flows and data, security, sales.

  • Keep training colleagues and departments together on a regular basis, such as every Friday morning. This is what will set you apart from competitors.

The golden rules:

  • Don’t start with tools or agents. Start by fixing processes and centralising company context.

  • Don’t start with job cuts. That isn’t transformation, and it holds the company back.

  • Keep senior leadership actively involved throughout.

  • Phase one saves time and cost. Phase two is doing what you never could before, and that is when agents pay off.

Benefits and Strategic Advantages

  • Early brand wins

    • Measurable ROI: Early adopters are seeing real value. For example, Reebok ad creatives enhanced by AI agents delivered a 60% higher return on ad spend (ROAS).

    • More personalisation: AI agents such as Jo Malone London’s “AI Scent Advisor” can map natural-language descriptions to complex data, moving shoppers from curiosity to a confident purchase.

  • Operational efficiency: Tools like “Authentic Intelligence” help employees speed up workflows dramatically, producing brand-aligned content as fast as trends change. This needs constant management, with full brand guidelines and processes documented.

  • Less friction: Native checkout and proactive digital concierges help capture sales at the moment of intent, significantly reducing abandoned carts. This is early technology with low usage, so bear that in mind in product roadmap and ROI discussions.

Risks and Considerations

  • Still experimental: Gen AI technologies are still experimental, so retailers must keep human supervision over (autonomous) agents. This applies from SKU-level management to warehouse and fintech management.

  • Integration complexity: UCP simplifies connections, but moving to an “AI-first”, connected retail model means significantly modernising data stacks (e.g. Vertex AI and BigQuery). Many businesses are not set up for this and will need major structural and headcount investment.

  • Security and trust: Features like native checkout are built with security at the core (using Google Wallet and PayPal), but retailers must manage the “consented actions” of agents carefully to keep customer trust. AI and agentic commerce are moving quickly while consumer behaviour takes a long time to change, and poor early experiences will damage the brand.

Existing Agentic Commerce in the Real World

  1. The Home Depot: Your In-Store Guide. The “Magic Apron” agent is a conversational assistant for home improvement projects. It helps with project planning and gives precise, aisle-level directions in the physical store, via chat or voice, in hundreds of languages, so you know exactly where to find what you need.

  2. Jo Malone London: Your Personal Scent Advisor. The “AI Scent Advisor” recreates the brand’s signature in-store consultation online. You describe the scents you like in your own words, and the agent maps your preferences to its complex scent data to give bespoke recommendations that move you from “curiosity to a confident purchase”. It solves the problem of not being able to buy perfume confidently online.

  3. Papa John’s: Your Smart Food Order Taker. Its agent lets you order in natural language across mobile, kiosks and in-car systems. It handles intelligent upselling (like suggesting a drink with your pizza) and syncs with the menu in real time, which has historically been a problem in food ordering.

  4. Reebok: Your AI-Powered Ad Designer. Authentic Brands Group, Reebok’s owner, uses specialised AI agents to quickly generate brand-aligned ad creatives, so Reebok can match changing consumer trends instantly. In early tests, these AI-enhanced creatives delivered a 60% higher ROAS than more traditional ads.

These examples show how agentic commerce is reshaping the whole retail value chain, from back-end ad creation and online consultation to in-store navigation and point of sale.

Go Deeper

If you or your team are struggling with AI, I offer dedicated, bespoke AI workshops, AI training, AI hackathons and AI transformation projects. Get in touch at ai@dannydenhard.com with any questions.


Recent News Stories Informing Agents & Agentic

1. ChatGPT Wallet lets AI agents pay for things without seeing your card details

Importance: Critical

What happened (13 Nov): OpenAI launched ChatGPT Wallet with Visa, Mastercard and American Express. Each purchase uses a stand-in card number (a payment token), so the agent and the merchant never see the real one. Before paying, the user reviews the merchant, items, total and card, then approves with Face ID or Touch ID. It’s live for ChatGPT Pro subscribers in the US. OpenAI says it expects users to eventually hand over whole purchases to agents, with spending limits and category rules.

Why it matters: Your briefing listed payments and trust as an open problem. This is the clearest sign yet that the card networks are building the checkout layer for agents. Once purchases are pre-approved within limits, agents start buying without a human checking each one.

2. “Businesses in ChatGPT” gives brands a profile, a dashboard and their own agent inside ChatGPT

Importance: Critical for any consumer brand

What happened (10 Nov): Businesses can now claim and verify their profile in ChatGPT and upload product feeds with pricing and stock. They also get data on how often they’re mentioned, how many website clicks they get, and which topics they appear in. ChatGPT has 1.2 billion weekly users. In the US, eligible ecommerce and online-service businesses can also launch a “Business Agent” that chats with customers and can be sponsored as the landing page for ChatGPT ads. openai.com

Why it matters: This gives a direct answer to your briefing’s question of how a brand shows up in chat-based answers. ChatGPT is becoming a business directory, a storefront and an ad channel at once. Claiming your profile now plays the same role that claiming a Google Business Profile once did.

3. Amazon’s plan to own the world’s product data and keep outside AI shopping agents in check

Importance: High

What happened (13 Nov, Reuters exclusive): An internal Amazon document describes “World Product Graph”: collecting data on almost every product for sale anywhere, including items Amazon doesn’t stock, partly by crawling other retailers’ websites. Amazon blocked Meta’s Muse agent from its site in September and is suing Perplexity over similar access. The document also says Amazon’s own AI search agent was giving inaccurate information. This is a proposal, not a launched product, and it could change or be dropped.

Why it matters: The platforms are now competing over who holds the definitive record of each product, not just who owns the customer. An agent can only recommend your product as well as that record describes it. One analyst put it this way: product feeds tell a platform what exists, but the links between products decide which one an agent recommends.

4. Alexa for Shopping launches in the UK

Importance: High for UK brands

What happened (11 Nov): Amazon replaced Rufus with Alexa for Shopping for every signed-in UK customer, with no Prime membership needed. On Amazon’s 29 October earnings call, Andy Jassy said US customers who use it spend 50% more per order. Data from Azoma, which sells AI shopping optimisation to brands, shows the new assistant personalises 42.7% of its answers, against 33.6% for Rufus. Each answer shows about seven products, and 99.8% of answers mention ratings and reviews.

Why it matters: UK brands now have to compete to make a shortlist of about seven products, chosen for each shopper individually. Reviews, price and clear use cases decide who gets onto it.

5. The UK sets out how it will handle agents that act out of control

Importance: High for UK businesses

What happened: On 3 November the government updated its plans for “agentic loss of control” incidents under the Cyber Security and Resilience Bill. After consultation, it intends to introduce resilience requirements, a duty to tell other affected organisations, rules on keeping evidence (such as logs of agent actions), and powers for ministers to review incidents. gov.uk

Separately, the ICO’s call for evidence on agentic AI closes on 20 November. On 30 October the CMA, FCA, ICO and Ofcom said in a joint statement: “An AI agent’s autonomy does not diminish the responsibility of organisations that deploy it.”

Why it matters: This backs up the governance step in your briefing. UK regulators expect any business using agents to keep a list of every agent, write down what each one is allowed to do, log its actions, and be able to switch it off. None of this is law yet, but it shows where scrutiny is heading.

6. IBM and Microsoft launch “Imagine Enterprise” with more than 6,000 prebuilt agents

Importance: Medium–High for enterprise readers

What happened (10 Nov): IBM and Microsoft launched a platform for building, running and governing industry-specific AI agents. IBM supplies more than 6,000 prebuilt agents and assets, and Microsoft supplies the cloud, identity and security tools. It starts with financial services, insurance and retail, and both companies’ engineers will help customers move from pilots into live use.

Why it matters: This supports your briefing’s point that specialist agents will beat generalist ones. The big vendors are now selling agents built for specific industries, with governance included, rather than general assistants.

7. Meta’s Muse agent moves onto wrists and into video chat

Importance: Medium

What happened (5–9 Nov): Meta added video chat with a Muse avatar for US users and launched “Spaces”, focused conversations where Muse works towards a goal for the user. It also brought Muse to Apple Watch and Wear OS. Muse handles email, bills, reservations, groceries and purchases. It passed 10 million US downloads within its first 40 days on the App Store. techcrunch.com

Why it matters: Muse is the consumer agent with the most momentum, which is why Amazon is blocking it. More of your customers’ shopping may start with an assistant like this before they ever reach your website.

8. Google brings agentic checkout to Canada

Importance: Medium

What happened (4 Nov): Google launched checkout built on the Universal Commerce Protocol (UCP) in Canada with lululemon, Shopify, Shoppers Drug Mart and Simons. Its research says 59% of Canadian holiday shoppers plan to use an AI tool to help them shop this season. blog.google

Why it matters: UCP, which your January update covered, is now expanding country by country. A UK rollout is the next one for UK brands to watch.

9. Anthropic expands the Claude Marketplace with partner agents

Importance: Medium

What happened (4 Nov): The Claude Marketplace now offers more than 25 partner agents and products across security, developer tools, creative work and productivity. unite.ai

Why it matters: This is an early version of the agent app store your “Future of Agents” post predicted. AI labs are building their own stores rather than waiting for Apple or Google to do it.

10. Coming up: Microsoft Ignite, 17–20 November

Importance: Watch

What’s happening: Satya Nadella gives the opening keynote on Tuesday. Sessions cover running agents safely on Windows and Windows 365. PCMag reports that last year’s agent features for Windows (a registry for connecting agents to desktop apps, and a separate agent workspace) never reached the stable version of Windows, so Microsoft has a lot to prove. uk.pcmag.com

Why it matters: Your briefing calls Microsoft the leader of the agentic web. Ignite will show whether that still holds for business customers.

What ties these together: These two weeks were about who controls the agent’s checkout, the product data and the customer conversation. OpenAI, Amazon and Google are each trying to become the place where products get found and paid for, while UK regulators make clear that the business deploying an agent is responsible for what it does.


If you find yourself or your team struggling with AI I offer dedicated and bespoke AI workshops or an AI enablement project, get in touch to ask any specific questions.

 

Supporting Presentations & Insights

Agentic Report From McKinsey

Summary:

  • How big it is: By 2030, agents could orchestrate $900B–$1T of US B2C retail revenue, and $3–5T globally. Those figures cover goods only, not services or B2B. McKinsey expects this to move faster than the web and mobile shifts did, because agents can use the same online checkout routes humans already use.

  • The technology behind it: Several new standards let agents connect to tools, talk to each other and pay securely. These include Anthropic’s MCP, Agent-to-Agent (A2A), Google’s AP2 payments protocol and OpenAI and Stripe’s Agentic Commerce Protocol. Visa, Mastercard, PayPal, Shopify and startups like Skyfire are building payment systems that agents can use.

  • What businesses need to change: Retailers need to decide where to build new things and where to upgrade old ones, across six areas: product discovery, loyalty, commerce platforms, payments and fraud, in-store systems, and fulfilment and returns. Product catalogues have to be readable by agents. The “agent experience” may soon matter as much as the customer experience, and traditional SEO will matter less.

  • Revenue at risk, and new ways to earn: Ad-based income, especially retail media networks, is under threat as agents skip past traditional ads. New ways to earn include:

    • bundling products from several brands and sharing the revenue

    • fees for negotiating deals

    • paid premium agents

    • selling data insights

    • sponsored suggestions that still serve the user’s intent

    Companies that move first get to set these pricing models.

  • Trust and risk decide how far it goes: The report names three main risks:

    • Systemic: one bad instruction can set off a chain of errors.

    • Accountability: it’s legally unclear who is to blame when an agent gets something wrong.

    • Data sovereignty: rules on where data is held and processed differ by country.

    Trust also varies by culture. For example, German shoppers prefer bank transfers and buy-now-pay-later over credit cards. To earn trust, agents need clear records of their decisions, ways to undo actions, and a human override. McKinsey’s message is to act now rather than wait and see.

Earlier Agentic Overview


Need To Shape Your Thinking More Deeply?

Here Are The Agentic Risks And Benefits

Previous
Previous

The AI Leadership Gap: Why Most CMOs Need To Assign An Owner

Next
Next

The Real State Of Content Marketing - Now Is The Time To Break Bland Content