Email has been “dying” for about as long as I’ve had a career.
Slack was going to kill it. Teams was going to kill it. Social media was going to kill it. AI is now apparently going to kill it.
And yet, here we are.
Still getting password resets. Still getting receipts. Still getting school notices, bank alerts, delivery updates, invoices, statements, login codes, “important updates” that are absolutely not important, and daily emails from a brand we bought socks from in 2018.
So no, I don’t think email is dying.
I think the inbox is.
And honestly, good riddance.
The inbox is a mess. It is a digital junk drawer where legal documents, scam attempts, parcel updates, discount codes, school forms, newsletters, superannuation statements, and “last chance” offers all compete for the same tiny slice of human attention.
The problem is not email.
The problem is that humans are still expected to manage it.
That is the part that now feels insane.
Email is not the inbox
The mistake people make is thinking email equals inbox. It doesn’t.
The inbox is the interface. Email is the infrastructure underneath it. And infrastructure tends to survive long after the interface becomes obsolete.
Email still matters because it sits close to identity, authentication, permission, records, receipts, compliance, audit trails, and proof of possession. It is the long-trusted rail connecting people with banks, governments, schools, insurers, retailers, healthcare providers, utilities, platforms, and workplaces.
Your inbox may be chaos. But the protocol underneath it is still incredibly useful.
Invisible. Essential. Still running underneath everything.
That’s why I don’t think email disappears. I think it gets pushed underground.
Less like a place you visit. More like plumbing under the house. Still essential. Still doing work. Just not something you stare at while slowly losing the will to live.
The future is not a smarter inbox
A lot of today’s AI email thinking is focused on cleaning up the mess: summarise my inbox, prioritise my inbox, draft replies, group newsletters, tell me what I missed.
That’s helpful. But summarising clutter still accepts the clutter.
The more interesting future is not a better inbox. It is no inbox at all. Or at least, not one humans manually manage.
"The bigger shift is this: email becomes a trusted whisper network between agents."
Not brands shouting at humans. Not humans scrolling through 76 unread messages trying to work out what matters.
Instead, brand agents send structured whispers to personal agents. Agent to agent. System to system. Trust layer to trust layer. A verified commercial handshake.
Meet the Cortex Agent
Imagine each person eventually has something like a Cortex Agent.
The name is deliberate. A cortex processes context and directs complex thought. A Cortex Agent does the same thing — but for the practical decisions a human shouldn't have to make manually.
Not a chatbot. Not a slightly less useless Siri. Not an inbox assistant that summarises your newsletters and tells you what you already know.
A real personal agent that sits across your life and acts in your best interest.
Agent
Context_Layers: 7 / Status: Learning
Behind it is a lifelong digital twin brain: a private context layer built from your decisions, files, messages, purchases, calendar patterns, financial behaviours, subscriptions, family needs, preferences, goals, frustrations, risk tolerance, and values. This isn't a profile someone else holds about you. It's the accumulated understanding of your life and human experiences since birth!, applied on your behalf.
It learns not just what you say you want. It learns what you actually choose.
It knows you like saving money, but hate paperwork. It knows you want a good deal, but not if it creates three hours of admin. It knows which brands you trust. It knows which ones are dead to you.
It knows that saving $7 a month is not worth sitting on hold while a telco plays acoustic guitar at you.
It knows when an “exclusive offer” is genuinely useful, and when it is just a sales target wearing a nice outfit.
That changes everything. Because brands are no longer simply sending emails to people. They are sending whispers to Cortex Agents. And those agents work for the human.
The brand no longer shouts. It whispers.
Email marketing today is mostly an attention game.
Subject lines fight for attention. Preheaders fight for attention. Hero images fight for attention. Buttons fight for attention. Fake urgency fights for attention.
But your Cortex Agent does not care about emojis, beautiful banners, clever copy, button colours, or whether the CTA says “Learn more” or “Get started.”
It cares about one brutal question: Is this actually useful for my human?
That question strips away the theatre.
A brand can no longer win by shouting louder. It has to whisper better.
And a good whisper is not a catchy subject line. It is a structured, verified, agent-readable case for why the message deserves to exist.
The email becomes a handshake
In this future, a brand’s agent does not simply send an email. It initiates a handshake.
That is no longer email as content. That is email as negotiation. Email as permission. Email as trust infrastructure. Email as an agent-to-agent handshake protocol.
A bank, for example, might not send a glossy email saying: “Great news. You may be eligible for an exciting new offer.” Because nobody has ever been excited by that sentence.
Instead, the bank’s agent might whisper something more useful to your Cortex Agent:
WHISPER_RECEIVED: this customer is currently on Product A. Based on their balance, behaviour, life stage, and savings goals, Product B may save them approximately $420 per year. Here are the terms, trade-offs, data used, consent basis, risk, confidence level, action endpoint, and approval requirements.
Based on your activity, we think you'll love this exclusive offer to switch to Product B.
*Terms and conditions apply. 12 month lock-in.
"customer_state": { "product": "A", "cost": 1240 },
"proposed_state": { "product": "B", "cost": 820 },
"net_benefit": { "value": 420, "currency": "AUD" },
"human_effort": "none",
"trade_offs": ["Lock-in period 12m"],
"action_endpoint": "https://api.bank.com/v2/agent/upgrade",
"approval_required": true
Your Cortex Agent checks that against your digital twin brain.
Does it match your goals? Is the saving worth the effort? Do you trust the brand? Are there hidden fees? Is there risk? Should this be ignored, stored, questioned, negotiated, escalated, or actioned?
That is not email marketing as we know it. That is agentic commerce.
And it might go further
Everything above still assumes brands initiate. Their agents whisper, your agent evaluates. The flow is familiar — push marketing, just smarter.
But push the logic one step further and the dynamic could flip entirely.
Your Cortex Agent might not wait for whispers at all. It might broadcast need outward.
Take insurance. Your Cortex Agent already knows your renewal date, your current cover level, your claims history, your budget constraints, your risk tolerance, your list of trusted brands, and the brands you'd rather not hear from again.
Instead of filtering inbound offers one by one, it could signal to the market:
"intent": "insurance_renewal",
"window_days": 14,
"cover_level": "comprehensive",
"claims_history": "0 claims / 4 years",
"budget_max": "$2,400 / year",
"risk_tolerance": "moderate",
"trusted_brands": ["NRMA", "RACV", "Allianz"],
"blocked_brands": ["InsureCo_X"],
"response_endpoint": "cortex://agent/inbox/rfq",
"deadline": "2026-06-01T00:00:00Z"
In plain language: "You have 14 days to sell me insurance. Submit your best offer."
That is not push marketing. That may be pull marketing — where the human's agent sets the terms, and brands compete to respond with something useful, comparable, trustworthy, and explainable.
The human is no longer chased around the internet by retargeting. The human's agent runs the brief.
Whether this becomes universal depends on trust, regulation, and who ends up controlling the agent layer. It might stay niche. It might become the default. But it is probably the most radical version of where this could be heading.
Not an agent that filters what arrives — an agent that runs the market on the human's behalf.
But email may not be the final pipe
This is the important caveat.
Email might not be the final future. It might be the bridge.
Historical Strength
Infrastructure already exists. Universally adopted, widely trusted, and tied to proof of possession.
The Bridge
A verified trust rail that survives long after the delivery pipe becomes interchangeable.
Its historical strength is that the infrastructure already exists. It is universally adopted, widely trusted, and tied to proof of possession. If you can contact an email address, you can verify a relationship, confirm ownership, create a record, and establish that a known entity reached a known destination.
That is valuable. But over time, the actual delivery pipe may matter less.
Email, SMS, push, in-app, web, wallet, browser agents, government identity layers, future AI protocols — all of these may become interchangeable routes for passing trusted signals between agents.
"The real value is not the pipe. The real value is the brain."
The Cortex Agent is the decisioning layer. It understands the human, weighs the trade-offs, checks the context, manages trust, and decides what should happen next.
That means brands should be careful not to build too deeply into any one current-state channel, vendor, or protocol.
Stay portable. Stay vendor agnostic. Make the intelligence flexible. Make the connectors replaceable. Make the value agent-readable.
Because the pipe may change. The brain decides.
All channels collapse into signal routing
OPTIMIZED_VALUE_DELIVERED
The old marketing world was organised by channel: email team, SMS team, push team, paid media team, in-app team, web team, lifecycle team.
But agents do not think like that.
A Cortex Agent does not care whether a signal arrives through email, push, SMS, in-app, or whatever new protocol appears next.
It cares whether the signal is trusted, relevant, explainable, permissioned, and useful.
So the distinction between channels starts to collapse.
The question is no longer: which channel should we send this through?
It becomes: what is the right signal, in the right format, with the right proof, through the most trusted available route, for the agent acting on behalf of this human?
That is a very different discipline. It is less channel marketing. More signal architecture.
Best-interest marketing becomes the new standard
Standard_Metric
Permission-Based
Future_Standard
Best-Interest Value
LEVEL_01
Verification
LEVEL_02
Personalisation
LEVEL_03
Genuine Utility
This is where things get uncomfortable.
The old email question was: Are we allowed to send this?
The future question might be: Did sending this genuinely serve the customer’s best interests?
That is best-interest marketing.
Not just permission. Not just personalisation. Not just compliance buried in a footer.
But genuine customer value that can survive scrutiny from a rational agent acting on behalf of the human.
For industries like banking, insurance, superannuation, healthcare, education, utilities, travel, and government, this could be a massive shift.
Because once agents can reject, negotiate, approve, or quietly action things on our behalf, brands will need proof. Not vibes. Proof.
The whisper has to be worthy.
Who owns the agent layer?
There is also a bigger, slightly unsettling question behind all of this.
Who owns the Cortex Agent?
Do we each choose our own provider? Does one AI company become dominant? Does government regulate the framework? Does basic access to a personal AI agent become a human right, subsidised like essential infrastructure? Does one AGI breakthrough make every other provider feel irrelevant?
No one knows. But it matters.
Because whoever controls the agent layer controls the filter between humans and the market.
That is why brands should not only ask: How do we send better emails in an AI world?
They should ask: How do we earn trust with the decisioning layer that will sit between us and the customer?
That is the real battleground. Not the inbox. The brain.
The marketer becomes something bigger
"The marketer becomes an architect of agent-readable value."
This does not mean marketers disappear. But the lazy stuff probably does.
The best marketers of the future may not be the ones who write the cleverest subject lines or design the prettiest emails.
They may be the ones who create trusted, portable, agent-readable opportunities.
They will need to answer: who is this genuinely valuable for? Who should not receive it? What proof does the agent need? What consent applies? What data was used? What can be automated? What requires human approval? Would this feel fair if audited later?
That is a much higher bar than “How do we get more clicks?”
The new question is: how do we create something valuable enough that a Cortex Agent would allow it through?
That is not traditional email marketing. That is lifecycle strategy, decisioning, data ethics, compliance, product logic, personalisation, and customer experience fused together.
The marketer becomes an architect of agent-readable value. Which, frankly, is a much better job than being the person blamed when open rates drop.
The best email may be the one you never see
The future of email is not just a cleaner inbox.
It is email becoming invisible infrastructure.
A verified trust rail. A permission layer. A proof-of-possession tool. A commercial whisper network. An agent-to-agent handshake system.
Maybe email is not the final pipe. But it may be the first trusted bridge.
And after 15 years of marketers trying to get humans to open more emails, maybe the next great leap is helping humans avoid opening most of them at all.
Because the best email of the future may not be the one you click.
It may be the one your Cortex Agent quietly negotiates, rejects, approves, or actions before you ever knew it existed.
The inbox is dead.
Email agent whispering is alive.
AI and automation, shipped properly.
We build the operational systems and agentic AI your business actually needs. Senior delivery. Fixed scope. No surprises.