Jon Miller

An AI-native system does not need every possibility written in advance; it can work out the right action when the decision comes up. Miller frames this as a shift from rules to reasoning, not as adding AI to existing automation.

In his keynote post, he gives examples: if title contains operations, put them in the operations nurture; if they click an email, add three points. He says the fix is not another rule but a system that reasons about the decision at the moment it arises.

The argument appears first as the keynote 'The 7 Ways AI Changes Marketing Automation' and then becomes the premise for Phave. Miller says Phave followed two years of conversations with more than 200 marketers who were frustrated with slow innovation and wanted AI-native capability rather than add-ons.

Read original · Original 2

Journey optimization picks among journeys a marketer has already mapped by hand.

• Next-best-action decides one move at a time, which Miller compares to shuffle: each touch may be fine, but nobody in B2B buys because of a single touch.
• A Playlist plans a sequence of approved actions for each person, looking many steps ahead, about thirty days out, and adjusts as the buyer engages.
• Miller argues hand-mapped journeys break down combinatorially: three actions can run in six orders, four actions in 24, and each addition multiplies the possible orders until there are more journeys than a team could draw.
• With Playlists, the team keeps the offers, content, goals, frequency rules, and eligibility constraints, while the AI assembles the order and should show why it chose what it chose.
• The Playlists idea is also the product Miller is selling through Phave, so the explanation is analysis and positioning at once.

A record shows what that person is scheduled to receive next month. Miller contrasts this with next-best-action as shuffle and with hand-mapped journey optimization, which works for a few journeys but collapses as the library of offers, channels, and content grows.

Read original · Original 2

Summarizing Lena Waters, formerly CMO at Notion and a marketing leader at DocuSign and Grammarly, he endorses her framing that AI adoption in marketing is a permissions problem before it is a talent problem: whether marketers have read and write access to the CRM, how many days and departments it takes to get product usage data, and whether someone can have an idea in the morning and put it into market that afternoon.

• In his conversation with Scott Brinker, Miller adds the context layer: pulling raw data out of a CRM or data lake is only part of context.
• The harder part is knowing the facts and memories accumulated about a person, their account, and their buying group.
• Brinker calls the platform version context-as-a-service; Miller calls the middle layer between model and marketing a 'marketing harness.'

Waters's test questions are concrete: do marketers have read and write access to the CRM, and how long does product usage data take to obtain? Miller's example of the context problem is that raw data alone does not tell you what has been learned about a person, account, and buying group, which is what makes AI useful rather than generic.

Read original · Original 2

The priority is what a leader actually does, not a general theme. The guest views are attributed separately from Miller's own arguments.

From Denise Persson, CMO at Snowflake for ten years, the post relays that there are no magic campaigns to rescue a soft quarter; any campaign that could fix the quarter would already be running.

• Her team uses Snowflake's data to see a soft quarter two quarters out, watches micro-territories of about eight sales reps because higher-level views only allow guesses at the cause, and looks for red when any region's top-of-funnel or conversion target falls behind.
• From Lena Waters, the post relays that teams should stop organizing by channel, make every team metric roll up to one shared department number, and prefer putting twenty variations into market over ten more meetings to pick one.

Read original · Original 2

Miller names Phave's CTO and CPO, Nick Bonfiglio, who the post says was EVP of Global Product at Marketo from 2009 to 2016 and later founded Aptrinsic and Syncari, and cites more than ten enterprise users including SambaNova, SPS Commerce, mabl, Servion, and Hypha.

• A follow-up post adds customer commentary: Nicole Fuselier, described as VP of Growth and Demand at SambaNova Systems and a Phave development partner, says marketers considering a new platform should start now because waiting widens the gap; Ali Jawin, described as Chief Revenue Officer at Hypha, contrasts a traditional campaign, which specifies who gets what over which channel and when, with Playlists, where the team builds the library and the AI decides who gets what and when inside guidelines.
• These are launch claims and customer testimonials gathered for the launch, so they evidence how Miller positions the product rather than independent proof that migrations succeed.

Miller uses two shirts twenty years apart, a Marketo shirt from 2006 and a Phave shirt, to argue that B2B buying changed while marketing automation did not. He says the old playbook was content to forms, nurturing, scoring, and MQLs, while buying now involves committees and anonymous research, so marketing must reach deal acceleration, expansion, and customer success as well as pipeline creation.

Read original · Original 2

Miller uses two shirts 20 years apart to date the start of the category and to announce the company they built.

Miller holds up a Marketo shirt from 2006 and a new Phave shirt, and uses the gap to argue that B2B buying changed while marketing automation did not.

• The post says the old playbook was content to forms, nurturing, scoring and MQLs, and that buying now involves committees and anonymous research, so marketing has to reach deal acceleration, expansion and customer success as well as pipeline creation.
• Miller writes that over two years the team spoke with more than 200 marketers frustrated by slow innovation and wanting AI-native capabilities rather than add-ons, and that this is why Phave was built.
• The post describes Phave as AI-native marketing automation that uses context about a person, their account and their buying group to decide personalized journeys, segmentation, scoring and buying-group mapping, with marketing operations still setting the rules.
• It names CTO/CPO Nick Bonfiglio, who the post says was EVP of Global Product at Marketo from 2009 to 2016 and then founded Aptrinsic and Syncari, and lists more than ten enterprise users including SambaNova, SPS Commerce, mabl, Servion and Hypha.
• Miller says the product was already generally available and was built in stealth because the goal was a platform an enterprise could switch to without losing anything.

For brands planning a campaign

Useful for an interview or advisory brief on platform migration and on what separates AI-native from AI bolted on. The named users and the co-founder's background give concrete reference points to probe.

Miller calls legacy marketing automation platforms "glorified if-this-then-that rules engines" and gives examples: if the title contains operations, put the person in the operations nurture; if they click an email, add three points.

• The argument is that rules are good at what is allowed but bad at what is best, because each rule only handles the situation it was written for.
• When something else comes up you add another rule, and another, until the system becomes a mess nobody wants to touch.
• An AI-native system that reasons, Miller writes, does not need every possibility written in advance and can work out the right action when the decision comes up.
• Miller frames the change as a shift from rules to reasoning rather than adding AI to existing automation.
• The keynote itself is linked in the post but not included in the material reviewed here.

For brands planning a campaign

Gives a clean, debatable thesis for a thought-leadership interview: which parts of marketing operations still belong in rules, and where reasoning should take over.

Miller summarizes a conversation with Scott Brinker, editor of chiefmartec and curator of the martech landscape, about what a marketing platform becomes when AI agents do more of the work.

• Brinker's point, as relayed here, is that platforms used to be judged by what people could do in the interface and now must also be judged by what agents can do, including agents from other companies; he calls this context-as-a-service rather than software-as-a-service.
• Miller pushes further: pulling raw data out of a CRM or data lake is only part of context, and the harder part is knowing the facts and memories learned about a person, their account and their buying group.
• That knowledge, Miller writes, is what makes AI useful instead of generic.
• Miller adds that models will keep getting smarter, so marketers need a layer between the model and the marketing, which engineers call a harness and Miller calls a marketing harness.
• The full video is linked in the post but not included in the material reviewed here.

For brands planning a campaign

Supports interview questions on data access and context infrastructure, and on what marketing operations should own versus what the model provider supplies.

Miller separates "decisioning" into three things: journey optimization picks which pre-built journey a person follows; next-best-action picks the single best move for that person right now; and a Playlist plans the best sequence of moves for each person, looking many steps ahead.

• The post argues hand-mapped journeys break down on scale, because three actions can run in six different orders and adding a fourth makes 24, with the count growing factorially as the library of offers, channels and content grows.
• Miller compares next-best-action to putting music on shuffle, where each song might be fine but nobody in B2B buys because of one touch, and contrasts it with a language model that writes better by looking several words ahead.
• A Playlist, in this description, takes the actions a team has approved and assembles a sequence for each person, planned about thirty days ahead and adjusted as the person engages, with any record showing what that person is scheduled to receive next month.
• Miller says the team keeps the offers, content and experiences in the library, the goals each Playlist works toward, and the rules on who must or must not receive what and how often, and that the AI should show its reasoning for each choice and the alternatives it considered.

For brands planning a campaign

Good source of specific interview questions about where human judgment is required, how much explanation buyers need from an AI decision, and how a marketing team would audit a planned sequence.

Miller relays advice from Nicole Fuselier, described as VP of Growth and Demand at SambaNova Systems and one of Phave's development partners, that marketers considering a new marketing automation platform should start now because waiting puts them further behind.

• Miller writes that AI is driving the biggest change in B2B marketing since marketing automation itself.
• The post also relays a reaction from Ali Jawin, described as Chief Revenue Officer at Hypha, to Playlists: a traditional campaign specifies who gets what, over which channel and when, while with Playlists the team builds the library of content and offers and the AI decides who gets what and when, based on behavior, signals and buying-group context, inside guidelines the team sets.
• Josh Kessell, described as Digital Marketing Manager at Telarus, is quoted on having the AI build programs rather than doing it piece by piece.
• Miller thanks development partners by name, including Nicole Siegal Fuselier, Anna Poulakis, Bruce Eidsvik, Josh Kessel, Jessica Meyers, Ali Jawin and Leah Itzkovich.
• The testimonial video and longer versions are linked in the post but not included in the material reviewed here.

For brands planning a campaign

Useful for a debate about migration timing, and for asking who is actually running the tool day to day rather than who endorses it.

Miller relays a conversation with Denise Persson, CMO at Snowflake for ten years, during which the post says revenue grew from about one million dollars to more than four billion.

• Her argument is that there are no magic campaigns to fix a soft pipeline inside a quarter: any campaign that could rescue the quarter would already be running, and at most B2B enterprises the sales cycle is long enough that even a working campaign will not close deals in that quarter.
• The post describes Snowflake predicting two quarters out whether a quarter will be soft, while there is still time to build pipeline, and says the team was focused on first-quarter pipeline while recording in the third quarter.
• It also describes marketing owning new business pipeline, with every region holding targets for top of funnel and conversion, and "looking for the red" when any of those numbers is behind.
• Miller cites former CEO Frank Slootman's principle that you are only succeeding when every theater is succeeding, so a company can beat its number and still not be succeeding if one region overachieves while others fall behind.
• The team drills down to micro-territories of eight sales reps, because at a higher level you can only guess at the cause.
• The episode also covers a daily brief that replaced dashboards, moving SDRs back into sales, and growing without adding headcount.
• The full episode is linked in the post but not included in the material reviewed here.

For brands planning a campaign

Offers a pipeline-planning interview angle that does not depend on talking about Miller's own product, and a concrete example of early-warning measurement.

Miller relays a conversation with Lena Waters, described as having built marketing teams at DocuSign and Grammarly and most recently CMO at Notion, about CMOs asking what an AI-native org chart should look like.

• Her answer, as reported, is that the org chart is the wrong place to start and that AI is a permissions problem rather than a talent problem: marketers are asked to rebuild everything from first principles while not being given enough permission to reach the data and tools.
• Miller relays the questions she raises, including whether marketers have read and write access to the CRM and how many days and departments it takes to get product usage data, and writes that hiring AI orchestrators and upskilling the team are downstream of that access question.
• The post says someone should be able to have an idea in the morning and put it into market that afternoon, and that if they cannot, the obstacle is permission rather than capability.
• On team structure, Waters argues that organizing by channel teaches people to optimize their own lane and defend it, and that every team metric should roll up to one shared department metric.
• On speed, she would rather put twenty variations in market and let the audience decide than lose most of them to consensus meetings, and her check on whether a change is real is whether it scares her ten percent.
• The episode also covers giving the next dollar to product experience before brand advertising, questions to ask a CEO before taking a CMO job, and running MQA alongside MQL for a couple of quarters.
• The full episode is linked in the post but not included in the material reviewed here.

For brands planning a campaign

Pairs with the context-and-harness thread and supports interview topics on marketing operations permissions, shared measurement, and how teams decide what ships.

Each new exception adds another rule until the system becomes something nobody wants to touch.

• The alternative is a system that reasons about the decision when it comes up, rather than needing every possibility written in advance.
• Miller developed this as a keynote, "The 7 Ways AI Changes Marketing Automation," and built Phave around it, describing the company as a bet that AI should not be bolted onto a legacy architecture but used to reimagine what marketing automation is.

Framing argument first delivered as a keynote, then carried into a product. Miller says the product followed two years of conversations with more than 200 marketers who were frustrated with the pace of innovation in their existing platforms and wanted AI-native capability rather than add-ons.

For brands planning a campaign

This is the strongest single interview spine: rules versus reasoning is a claim Miller has staked a company on, so a host can push on where rules still beat reasoning, what an AI-native system cannot yet do, and what breaks in year two of a migration.

Examples: Source 1 · Source 2

Journey optimization picks among journeys a marketer has already mapped by hand, which works for two or three journeys and collapses beyond that: three actions can run in six orders, four in 24, and the count grows factorially until there are more possible journeys than any team could draw.

• Next-best-action decides one move at a time, which Miller compares to shuffle — each touch may be fine, but nobody in B2B buys because of a single touch.
• A Playlist plans a sequence of approved actions several steps ahead, thirty days out, and adjusts as the buyer engages.
• The team keeps the offers, goals, frequency rules, and eligibility constraints; the AI assembles the order and should show why it chose what it chose.

Miller sells Playlists through Phave, so this is analysis and product positioning at once. The distinction is still concrete enough to be tested against a buyer's actual journey inventory.

For brands planning a campaign

A good interview move is to ask Miller to apply the three-way distinction to a real campaign the audience would recognize, and to say what evidence would show that a Playlist outperformed a well-mapped journey set.

Examples: Source 1 · Source 2

Summarizing an episode with Lena Waters, formerly CMO at Notion and a marketing leader at DocuSign and Grammarly, Miller endorses her framing that AI adoption in marketing is a permissions problem before it is a talent problem: whether marketers have read and write access to the CRM, how many days and departments it takes to get product usage data, and whether someone can have an idea in the morning and put it into market that afternoon.

• Miller's conversation with Scott Brinker adds the context layer: pulling raw data out of a CRM or data lake is only part of it, and the harder part is knowing the facts and memories accumulated about a person, their account, and their buying group.
• Brinker calls the platform version of this context-as-a-service; Miller calls the middle layer between model and marketing a "marketing harness."

For brands planning a campaign

Directly addresses the brief's data-access angle. Ask what a marketing harness actually contains, who owns it, and how a team would know the context it feeds a model is accurate rather than merely available.

Examples: Source 1 · Source 2

From Denise Persson, who has run marketing at Snowflake for ten years as revenue grew from roughly $1 million to more than $4 billion: there are no magic campaigns to rescue a soft quarter, her team uses Snowflake's own data to see a soft quarter two quarters out, and marketing watches micro-territories of about eight reps because at a higher level you can only guess at the cause. From Lena Waters: stop organizing the team by channel, make every team metric roll up to one shared department number, and prefer putting twenty variations into market over ten more meetings to pick one.

A recurring format: a named practitioner's own quotes, a small number of concrete operating details, a short list of additional episode topics, and a closing question aimed at the reader's own experience.

For brands planning a campaign

Evidence that Miller can run a thought-leadership interview, not just speak in one. Useful as a template question set: how does a guest's answer change what the audience should do on Monday.

Examples: Source 1 · Source 2

Miller names Phave's CTO and CPO, Nick Bonfiglio, who was EVP of Global Product at Marketo from 2009 to 2016 and later founded Aptrinsic and Syncari, and cites more than ten enterprise users including SambaNova, SPS Commerce, mabl, Servion, and Hypha. A follow-up post adds customer commentary and the argument from SambaNova's Nicole Fuselier that marketers considering a new platform should start now because waiting widens the gap.

These are the company's own launch claims and customer testimonials gathered for the launch, so they evidence how Miller positions the product, not independent proof that migrations succeed.

For brands planning a campaign

Strong interview topic on migration reality: what an enterprise actually has to unwind, how long a switch takes, and what Miller would tell a team that decides to stay put for another year.

Examples: Source 1 · Source 2

He describes the product as using context about a person, their account, and their buying group to decide on personalized journeys, segmentation, scoring, and buying group mapping, and he names Phave's CTO/CPO, Nick Bonfiglio. He says the platform is generally available with more than ten enterprise customers and that he has spoken with more than 200 marketers frustrated with legacy platforms.

View example

Nicole Fuselier, VP of Growth and Demand at SambaNova Systems, is described as one of Phave's ten-plus development partners who have been using the product and helping shape what gets built.

View example

Hypha's chief revenue officer, Ali Jawin, is quoted reacting to Phave's Playlists, which let a team build a library of content and offers while AI decides who receives what and when.

View example

The posts do not state a contract or commercial relationship with Telarus.

View example

The posts describe his hosting role but do not state whether he owns the podcast or is paid for it.

View example

The post does not state any employment or commercial relationship with Marketo.

View example

Both treat the buyer's research path as something marketing systems should read and act on.

Radde's posts focus on identifying companies behind anonymous traffic and combining ChatGPT Ads with GEO/AEO; Miller's focus is the automation platform itself, including decisioning, Playlists, and the data and permissions layer.

Read original · Original 2

Both are aimed at marketing leaders who need to fix or run the demand engine.

Basu's material is checklist- and playbook-oriented for diagnosing a marketing system and equipping buying committees; Miller's is more architectural, arguing about reasoning systems, decisioning, and context layers.

Read original · Original 2

Both connect AI adoption to the underlying data and operating model rather than treating it as a tooling add-on.

Wueschinski's posts center on product catalogs, agent-readable data, and revenue-team agent governance; Miller's center on marketing automation journeys, personalized decisioning, and buying-group context.

Read original · Original 2

The reviewed posts cover two things a marketing technology vendor's audience asks about directly: identifying the companies behind anonymous B2B website traffic and enriching them with company and intent data so sales can prioritise on real behaviour, and splitting AI search spend between paid placement inside ChatGPT and longer-term generative engine optimisation so a brand is mentioned inside AI answers rather than only clicked past. If the vendor wants an episode that ends with a channel plan its audience can act on, Radde's material is closer to that than Miller's platform-architecture argument.

The tradeoff is depth and durability.

• Both reviewed Radde posts are short, close by asking for comments, and one is a spotlight on another company's product rather than an argument Radde owns.
• There is no extended thesis to test, no record of interviewing named marketing leaders, and no stated position on how marketing automation should be rebuilt — so a host gets tactical advice that ages quickly instead of a defensible point of view.
• Miller is the stronger choice when the conversation has to survive a technical audience; Radde is the stronger choice when it has to produce immediate demand tactics.

Examples: Source 1 · Source 2

The reviewed posts supply a 100-day sequence — audit closed-won deals, map channels against pipeline produced, check the real MQL-to-SQL rate, audit attribution, interview the last five losses, then align on ICP and a single North Star pipeline metric — and a breakdown of how six buying stakeholders (champion, executive, finance, IT and security, operations, legal) each need different content, with deals stalling because teams build for only one of them. Those are usable in a workshop or an advisory session within the same week.

The tradeoff is that the material stays at process level.

• It does not argue about what marketing automation should become, how AI reasoning differs from rules, or where a context layer sits between the model and the marketing — the questions this vendor's brief is built around.
• A session with Basu produces a remediation plan for a marketing system; a session with Miller produces a position on AI-native automation and data access.
• Choose Basu when the audience needs the first and accept that the platform-strategy angle will be thin.

Examples: Source 1 · Source 2

The reviewed posts argue that agents now decide on structured inputs rather than browsing, that an audit found 70 percent of leading brands' product pages fall short of that standard, and that structured product data should be treated as a revenue line with an owner, a budget, and a board metric. The second post makes a governance argument for AI agents in the revenue engine — one named owner per agent, one workflow per agent, a business metric set before launch, and an exportable rollback path — using the gap between AI scaling and reported EBIT impact as the hook.

The tradeoff is scope.

• The reviewed material concerns product-catalogue legibility and revenue-agent governance, not marketing automation platforms, buying-group context, or the shift from rules to reasoning.
• A vendor wanting an episode about AI-native marketing automation would have to steer Wueschinski's material toward that topic, whereas Miller arrives with the argument already made.
• Choose Wueschinski when the vendor wants a CFO-facing or commerce-facing episode; choose Miller when the topic is the marketing stack itself.

Examples: Source 1 · Source 2