Marketo co-founder Jon Miller launches Phave, AI-native marketing automation

Phave came out of stealth as an AI-native marketing automation platform founded by Marketo co-founder Jon Miller and Nick Bonfiglio, using Maestro to compute per-person playlists that replace rule-based Marketo, HubSpot, Pardot, and Eloqua stacks, with general availability that began in August 2026.

SOFTWARE desk — the guy who helped invent modern marketing automation just shipped the replacement thesis: stop writing rule trees, let a reasoning model decide the next touch for each person while humans keep the Approve button.

Who founded it, and who is named as backing it. Jon Miller is co-founder and chief executive, and co-founder of Marketo and Engagio. Nick Bonfiglio is the other founder. The lede says he ran product and engineering at Marketo. The about box says he was formerly EVP of Global Product at Marketo. EVP means executive vice president, the senior product job. The about box says Phave is backed by FirstMark, Ridge Ventures, and Costanoa Ventures. The release does not print how much each firm put in, and it does not print a valuation, which would be a price on the whole company. Do not invent a check.

What they say they are replacing, and who they name as users. The lede calls Phave an AI-native marketing automation platform, built to replace Marketo, Pardot, Eloqua, and HubSpot. AI-native means the software was built around artificial intelligence from the start, not a rules product with a chat box added later. A marketing automation platform, which the trade shortens to MAP, is the system that decides which message a person gets, and when. The release says the product reached general availability in August 2026. General availability means a customer can buy it, not only join a private test. Named enterprise users: SambaNova, SPS Commerce, mabl, Servion, and Hypha. A name on a release is not a contract this desk read.

The idea, in the release’s words. Every decision a legacy platform makes is a rule somebody wrote in advance: who belongs in a segment, what a lead is worth, which email comes next. A segment is a bucket of people who share a trait. A lead is a person the company hopes will buy. Rules cannot handle a case nobody wrote down, so the release says most mature programs end up with two or three nurture tracks, not because buyers are simple but because that is what a rules system can handle. A nurture track is a fixed sequence of messages. Phave’s answer is Maestro, which the release calls the intelligence inside Phave. Maestro computes a Playlist for each person: the touches most likely to reach the objective the marketer set, ordered and timed for that individual, following the rules and governance the team keeps in place. A touch is one message or other contact. Governance, here, means the limits the team refuses to break, such as consent and quiet hours. Before anything goes out, Maestro weighs everything else already queued for that person, across every campaign. The release’s picture: campaigns are the albums the marketing team makes and approves, and the Playlist mixes them into an order for each listener. These lines are the release’s. This desk did not watch a send.

How a marketer writes the work, and where they run it. The release says marketers describe what they want in plain language instead of translating it into configuration, so campaigns, audiences, scoring, and automations no longer get stuck behind whoever knows the old platform. Scoring is the number a system puts on a person. Phave can be run through its own screens, through the AI client a team already uses, or headlessly through MCP and a versioned REST API. Headless means another program runs it, with nobody clicking the Phave screens. MCP, short for Model Context Protocol, is a plug that lets an outside AI assistant call a product’s tools. A REST API is a versioned web door other software can call. The release says permissions and approvals are enforced the same way whichever door a request comes through. An approval means a person still has to say yes. The product site, not the wire, adds that a send can be approved only by a signed-in human on a Phave page. File that sentence as the site’s.

Accounts and buying groups, and one customer’s sentence. The release says Phave treats accounts and buying groups as records of their own, each with its own score and lifecycle stage. An account is the company you are selling to. A buying group is the set of people there who decide one purchase. A lifecycle stage is where that record sits, such as prospect or customer. One account can hold several buying groups at once, at different stages, some prospect and some customer. The release says that is what makes it possible to market to the committee that expands an account, not only the committee that already bought. Bruce Eidsvik, chief growth officer at Servion Global Solutions, said that with a traditional platform, going after one account with four global sites and buyers across IT, finance, and service operations is practically impossible. He said Phave’s playlists work out when and how to reach each person, and that Servion has more than doubled its marketing execution volume. More than doubled is his sentence on the wire. It is not a count this desk audited.

The price, and whose name is on the meter. The release says Phave starts at $36,000 a year, priced on the number of people who receive something in a given month, rather than on database size. Database size would charge for every stored name, including people you never contact. The product site gives that monthly count a name: Monthly Active Playlists, the people who receive something from you in a month. The name is the site’s. The $36,000 starting price and the “people who receive something” rule are on both the wire and the site. The site also says there is no charge for tokens and no metered AI usage. A token is a small chunk of text a model reads or writes. That no-token line is the site’s. It is not on the wire. This desk did not see an order form.

The scorecard, and whose scores they are. The release says Phave scored itself and the legacy platforms against 481 requirements drawn from twelve enterprise evaluations. Each requirement was scored and weighted by Claude, rather than by Phave. Claude is Anthropic’s model. The release’s totals: Phave 2.97, HubSpot 2.46, and Marketo 2.38. It says Phave matched or beat every platform evaluated on at least 90 percent of the requirements. It says the methodology is published at phave.com. On that page the scale is 0 for not available, 1 for minimal, 2 for partial, 3 for fully supported, and 4 for differentiated, which the page defines as no other product reaching a 3, and each row is weighted by importance. The page says the scores are from September 2026. A nearby line on the same site says Phave matched or beat each competitor on 90 percent of the requirements. “At least 90 percent, against every platform evaluated” is the wire. “Each competitor, on 90 percent” is the site. Do not merge them into a third sentence. These figures are the company’s. This desk did not rescore 481 rows.

The public webinar. The release says Jon Miller and Scott Brinker of chiefmartec will discuss what marketing automation becomes when it is rebuilt around AI, on Tuesday, September 29, at 10:00 a.m. PT. PT is Pacific Time. On that date, 10:00 a.m. Pacific is 1:00 p.m. Eastern. The product site prints both clocks, 10:00 a.m. PT and 1:00 p.m. ET. The registration link on the wire is https://phave.com/future-of-marketing-automation-29-Sep-2026. A date on a calendar is not a recording this desk watched.

A line from the founder, as a quote, not as a market this desk measured. Jon Miller said rules are good at what is allowed, but bad at deciding what is best. He said consent, frequency, and quiet hours should stay rules, and that choosing which of two good emails a person should get was never a job for a rule. He said reasoning models can make that choice now, that the legacy platforms stopped innovating years ago, and that marketers are ready for something built for how business-to-business selling works today. Business-to-business, shortened to B2B, means selling to companies, not to a shopper at home. “Stopped innovating years ago” is his sentence. It is not a product history this desk wrote.

Plain English for the rest of the card: stealth = building before a public launch. AI-native = built around AI from the start. MAP = marketing automation platform, the system that picks the next message. segment = a bucket of similar people. lead = a person the company hopes will buy. nurture track = a fixed sequence of messages. Maestro = Phave’s name for the software that picks the next touch. Playlist = the ordered, timed touches for one person. touch = one message or other contact. governance = the limits the team refuses to break. campaign = a set of messages the team approves. The release calls campaigns the albums, and the Playlist the mix. scoring = a number on a person. headless = another program runs the product. MCP = a plug that lets an outside AI assistant call the product’s tools. REST API = a versioned web door other software can call. approval = a person still has to say yes. account = the company you sell to. buying group = the people there who decide one purchase. lifecycle = where that record sits, such as prospect or customer. $36,000 a year = the starting price on the release. Monthly Active Playlists = the site’s name for people who receive something in a month. token = a small chunk of text a model reads or writes. 2.97, 2.46, and 2.38 = the company’s weighted scores for Phave, HubSpot, and Marketo. 481 = the company’s count of requirements. at least 90 percent = the wire’s claim about how often Phave matched or beat the other platforms. 10:00 a.m. PT on 29 Sep 2026 = the webinar clock. This filing is the 23 Sep launch. It is not a price quote for a named customer, and it is not a rescore.

PRIMARY here: Phave’s 23 Sep 2026 release on PR Newswire, stamped 08:00 ET and sourced to Phave — Tier A PRIMARY, the company’s own announcement. The two-years-in-stealth subhead, the Burlingame dateline, the Miller and Bonfiglio roles, the Marketo and Engagio line, the “ran product and engineering” sentence, the EVP of Global Product line in the about box, FirstMark, Ridge Ventures, Costanoa Ventures, general availability in August 2026, SambaNova, SPS Commerce, mabl, Servion, Hypha, the Playlist, the album picture, the plain-language authoring, the Phave screens, the outside AI client, MCP, the versioned REST API, the same permissions and approvals, accounts and buying groups, the Eidsvik quote, the $36,000 price, the people-who-receive meter, the 481 requirements, the Claude scoring line, 2.97, 2.46, 2.38, the at-least-90-percent line, the Miller quote, and the 29 Sep 10:00 a.m. PT webinar with Scott Brinker are that release’s. phave.com is the product primary for the Monthly Active Playlists name, the 0-to-4 score scale, the September 2026 scoring line, the “each competitor on 90 percent” wording, the no-token price line, the signed-in-human approval sentence, and the 1:00 p.m. ET clock beside 10:00 a.m. PT. NOT claimed: a check size or a valuation, that this desk opened the product, audited Servion’s volume, or rescored the 481 rows, that the wire’s “at least 90 percent” and the site’s “each competitor on 90 percent” are one sentence, that “stopped innovating years ago” is this desk’s history, a stock tip, or investment advice. Distinct from the already-filed amazon-seller-assistant-agentic, clara-agentic-intelligence, and digitalocean-managed-agents.