Head of Strategy

Every commerce shift eventually produces a toll-booth where consumer demand is concentrated. The Industrial Revolution had the railways and container shipping ports, retail had supermarkets and out-of-town shopping centres, the internet had the platform. The question hanging over the emergent agentic economy is whether the next choke-point has already been created, and whether it's Google, Anthropic, OpenAI or even open source agent project, OpenClaw, who will own it.

Future winners aside, agentic commerce looks set to disrupt one of the most persistent characteristics in consumer propositions: bundling.

If today's consumer bundles are buffets - set menus of content, software or services priced as a single line item for us to graze on - then the agentic future looks more like a picnic: a basket assembled piece-by-piece by an AI working on your behalf.

Bundling & unbundling: a familiar cycle

Jim Barksdale's old line - that there are only two ways to make money in business, bundling and unbundling - has survived decades of economic cycles because it keeps proving itself. From the McDonald's meal deal to Amazon Prime, Thomas Cook's package holidays to Vitality's exercise-meets-health-insurance offer and influencer touted subscription boxes, break-out bundles have a knack for being the most influential propositions of their era.

And with every great bundle breakthrough, there are those who claim the inevitable unbundling will be permanent. But predictions of their demise have, so far, been greatly exaggerated.

The early-internet hype would have had you believe the infinite shelf space of ecommerce alongside search engines would unbundle everything, leaving each consumer to assemble their own optimum. In practice, this assumed a level of patience that consumers do not often possess.

The unbundled holiday - flights, transfers, accommodation, insurance, lunch arrangements, drinks destinations - is technically cheaper for many travellers, and potentially a better fit for their preferences. But the buffet Jet2 and Royal Caribbean offer is bigger than ever. According to ABTA, package holidays are around 62% of UK holiday trips abroad, rebounding from their pre-COVID lows. And while Fintechs have spent the past decade or more attempting to unbundle different products (loans, cards, savings, investments) from incumbent banks, many customers tend to keep their savings accounts and credit cards with their primary bank even when they switch (47% and 35% respectively).

Bundling has survived the internet age largely because of the residual faff: the cognitive load of DIY is real (even with Compare the Market, Google, Skyscanner, Airbnb, Booking.com and the others to help), and many people will pay a premium to not have to think about it.

Agents break the faff impasse

Agents change the calculus. An AI that can fetch, compare, decide and, increasingly, purchase on a user's behalf is, in Ben Thompson's terms, a new “demand aggregation” layer. His aggregation theory says that value flows to whoever owns the consumer relationship. So what happens when the consumer relationship is intermediated by an agent the consumer trusts more than any individual provider?

The implication is uncomfortable for incumbents. Across industries, the biggest media, finance, retail and leisure brands have built moats by lowering the cost of curation for consumers: choose us, and we'll do the work. The leaders today have refurbished and reshaped those moats as connected commerce - search, platforms, digital transactions and easy or entirely digitised fulfilment - happened and upended industries. Or newly digital native competitors,like Netflix, Spotify, Monzo, Booking.com and Jet2, have popped up and done it first.

Agents promise to upend these equilibria again, pooling individual consumer’s demands and driving new patterns. Early evidence suggests that conversion rates from AI queries are almost 1.8x higher than organic search, and while still niche, AI traffic to ecommerce sites is growing rapidly.

Wasps at the picnic: Agents make bundling economics harder to get right, while accentuating transparency and bundle broadening pressure

The economics of bundled propositions can be complex. Fortunately, Shishir Mehrotra, formerly of YouTube and a Spotify board member has done some of the clearest thinking on how to make bundles work. His four bundling myths and corresponding theses could all be disrupted or accelerated in future agentic commerce scenarios.

The first is that bundles generate value by capturing what Mehrotra calls casual fans: people who would never pay full price for a thing, but happily consume it as part of a package. In some visions of agentic commerce, this seems like a tough sell. Most consumers will surely task their agents to procure things they are super-fans of. Agents could be ruthless casual-fan optimisers on our behalf, telling us which buffet(s) gives the best access to the long tail of things we kind-of want, sometimes. But this feels like a hard sell to the lay person, being asked to entrust their preferences to an autonomous agent with their bank details.

The second focuses on marginal churn contribution (how much does each bundle element drive cancellation if removed) as the best way to value different bundle components over usage. But a naive implementation of agentic commerce feels likely to focus on usage: agents proactively retiring subscriptions and products that are unused. Again, smarter agentic behaviour might find ways to measure marginal churn contribution persistently for each of us, but it’s a tougher task to operate in this future counter-factual over historic usage data.

The third, that bundles must offer an a-la-carte price or feel like a rip-off, becomes an unavoidable fact: agents will always be running the comparison, forcing (or smartly inferring) transparency on top of suppliers whether they like it or not.

The fourth, that the best bundles are broad, and that increasingly bundling will happen across categories to unite different groups of superfans + casualfans. This feels ready-made for an agent optimisation loop. Effective agents will spot the non-obvious overlaps (a cooking app, a wine subscription and a Pyrenees holiday package; a news sub, a flight tracker and a sleep supplement optimisation product), and assemble personalised bundles no single provider could have conceived.

Why the bundle buffets won't disappear

It would be a mistake, though, to assume agents will dissolve bundles entirely. Several challenges stand in the way of our future robot friends.

A first is how many of our choices are driven by intangible, unmeasurable motivations. Agentic commerce will move quickly where features are quantifiable: kilowatt-hours, gigabytes, feeds, speeds, delivery slots. But many things aren’t so tallyable. Content is an awkward case. Our enjoyment of a film, an article or an album is highly contextual, and notoriously hard to compress into a vector an agent can optimise against. An agent can probably tell you the cheapest way to insure your car; it will struggle to tell you which novel will move you on your summer holiday this year. Or quite how important the Champions League final is to you (Claude, if Arsenal make it, it’s really, really important).

Demand forecasting is the second problem. Bundles work in part because providers can pool elastic and inelastic users and smooth their purchasing accordingly. An agentic world in which every household's demand is reconciled month-to-month into atomic purchases may simply push volatility back onto suppliers, who will then price for it. This dynamic, combined with Shishir’s fundamental insight around bundles typically serve us well by giving us access to things we are less fanatical about (but casually interested in) could be their critical saving grace. Bundles survive as the best bargain available, agent-approved.

Finally, there is trust - the pervasive challenge for any agentic commerce maximalists to address. Will users let agents transact at scale on their behalf, especially when mistakes are expensive? And will providers trust agents not to flood them with returns, cancelled contracts, chargebacks or fraud claims?

OpenAI’s recent abandonment of Instant Checkout - which allowed users to buy directly within ChatGPT from Etsy, Doordash and Walmart among others - suggests that retailers aren’t ready to hand the entire purchase funnel to a single intermediary.

The toll-booth question in an uncertain agentic future

Which brings us back to where we started. The interesting strategic question is not whether bundles survive in some form - they will - but who owns or operates the agent layer. If a small number of platforms (Apple, Google, OpenAI, Anthropic, perhaps Meta or Amazon) end up running the trusted agents that consumers actually use, they may well have built a new toll-booth between every supplier and every customer. Visibility inside those agents will become the equivalent of shelf space at Tesco or a top-ten slot in the App Store. And each of the agent platforms might have different mechanisms and approaches to that new form of merchandising.

OpenAI’s ad business is nascent but growing and their cost-per-click model feels charmingly anachronistic. It’s inevitably triggered a partial backlash, and Anthropic has strongly counter-positioned against it with a ‘no ads’ philosophy:

“Whether someone asks Claude to research running shoes, compare mortgage rates, or recommend a restaurant for a special occasion, Claude’s only incentive is to give a helpful answer. We’d like to preserve that.” (Anthropic, Feb 2026)

Perhaps more intriguing still is the potential for agents to operate independently of any centrally hosted and controlled entity. The rapid growth of OpenClaw - an open source project that is enabling users to self-host agents to operate on their behalf - amongst hobbyists suggests that there is a possible future that is more decentralised and harder for any single platform to control.

Whatever the future of Agentic Commerce, smart incumbents are already preparing. Sky's new mega-bundle - of Disney+, Hayu, HBO Max and Netflix - alongside their core Sky entertainment is a striking alliance of competitors. This announcement now sits alongside their well established NOW TV brand: a relatively unbundled streaming-first product that allows consumers to choose individual elements for short periods. In the future, this means Sky’s individual content products can be marketed individually to agents, while the overall bundle is likely to stand-up to scrutiny for those agents looking for access to the widest range of content. A smart strategic hedge ahead of an uncertain agentic future.

Expect more of this - and more uncomfortable partnerships with frenemies, on the basis that being inside someone else's aggregation that an agent is considering is better than being invisible to it.

From buffets to picnic packers

The agentic economy won’t end bundling; it will shift the power of bundling from the incumbent providers - the buffet manager - to the user's agent - the picnic packer.

For most, the strategic implication is the same: Be ready to reconsider, and repack your bundles in the face of this new scrutiny. Form new alliances and collaborations (even with prior frenemies) to market new bundles that can stand up to the rigour of agentic assessment. At the same time, make your core offerings atomic, transparently priced and ready to be picked by agents for those who are optimising for a-la-carte.

The picnics are increasingly being packed by agents. Whose basket will you end up in?

Get the latest news and views from Loomery directly to your inbox

Stay ahead of the curve with our monthly newsletter, The Weave.

Newsletter sign up

Can Bundling Survive the Agentic Economy?

May 2026

The agentic economy won’t end bundling, but it will shift power to everyone’s agent, and whoever’s holding your agent’s basket.

Get the latest news and views from Loomery directly to your inbox

Technology

Company news

AI

Let's talk →

Can Bundling Survive the Agentic Economy?

May 2026

The agentic economy won’t end bundling, but it will shift power to everyone’s agent, and whoever’s holding your agent’s basket.

Get the latest news and views from Loomery directly to your inbox

Score your team against the 8Cs

Technology

Company news

AI

Let's talk →