Jon Miller, the guy who co-founded Marketo, launched Phave marketing automation on September 23, 2026. It's an AI-native platform that replaces rule-based sequencing with a reasoning model called Maestro, and it starts at $36,000 a year.
If you're running a handful of email sequences out of Mailchimp or a basic CRM, this changes nothing for you this week. If you're already paying enterprise marketing-automation-platform prices, it's worth twenty minutes of your attention.
What Is Phave Marketing Automation, Exactly?#
Phave marketing automation went into general availability on September 23, 2026, after roughly two years in stealth, according to Phave's launch announcement and reporting from MarTech.
Miller, who also co-founded the account-based marketing platform Engagio before it sold to Demandbase in 2020, built Phave with Nick Bonfiglio, formerly EVP of product at Marketo. The pitch is simple to state and harder to build: legacy platforms like Marketo, Pardot, Eloqua and HubSpot run on rules. If X happens, do Y. Phave's Maestro engine reasons about each contact instead, and builds what it calls a Playlist, an individualized sequence of touches ordered and timed per person rather than one static flow for everyone.
The key facts as of launch:
- Launched: September 23, 2026, in general availability
- Founders: Jon Miller (Marketo, Engagio) and Nick Bonfiglio (former Marketo EVP of Product)
- Pricing: starts at $36,000 a year, billed by monthly active recipients, not database size
- Reach: 10 live customers, including SambaNova and SPS Commerce
- Reasoning layer: 319 tools exposed through MCP, so outside AI agents can trigger actions directly
It also treats accounts and multi-person buying groups as their own scored records, not just a pile of individual contacts.
Does This Actually Change Anything for a Smaller Business?#
Not yet, and probably not soon.
Phave's $36,000 floor alone prices out most businesses under roughly 50 to 100 employees. It's also solving a problem smaller teams mostly don't have: scoring a buying group where five or six stakeholders at one account each need different messaging. If you're running one or two nurture sequences for a founder-led sales process, that's not your problem, and this platform doesn't fix one for you.
Where it does matter is upstream of the price tag. Miller put the core argument plainly in the MarTech interview: rules are good at what must be true, but they can't handle ambiguity, and they can't judge what's actually best. That's correct, and it holds at every budget level, including the WhatsApp and email flows we build for SaaS founders and D2C brands across the GCC, Europe, the US, Canada and India. You don't need a $36,000 platform to act on it. You need the same judgment layer, sized down to fit.
The Honest Trade-offs#
The reasoning pitch is real, and we've run into the exact wall it's describing. A static delay node in an automation flow can't decide "send now, unless this person just replied." It just waits, on schedule, whether that's still the right call or not. A judgment layer genuinely does something a fixed rule chain can't.
But there's a real catch. Reasoning models are non-deterministic. When a rule breaks, you read the rule and usually find the bug in five minutes. When a reasoning model picks a bad send time or the wrong channel, you're debugging a black box, and Phave hasn't published failure-rate data. The only performance number in circulation is that early adopters build campaigns two to three times faster, per MarTech's reporting.
Ten reference customers after two years in stealth is a small sample to hang an enterprise marketing stack on. And pricing by monthly active recipients is the kind of model that looks reasonable at launch and creeps as a list grows. None of that makes Phave a bad bet for the accounts it's actually built for. It does mean nobody should buy it on the AI story alone.
How We're Handling It#
We didn't wait around to see how Phave lands before acting on the idea underneath it.
At WebEpex we've already swapped the static delay-and-branch logic in a couple of client WhatsApp and email flows for an LLM decision step: the workflow, built in n8n, calls a model to decide send timing and channel fallback before a message goes out, instead of a fixed wait block deciding it blindly. It's a far smaller version of what Maestro is doing, running on infrastructure we already operate rather than a new $36,000 platform.
We build this kind of automation for SaaS founders and D2C brands across the GCC, Europe, the US, Canada and India, and the clients who actually need a full reasoning layer are the ones past their first few thousand monthly conversations. Below that volume, a well-built rule chain still wins on cost and on how fast you can debug it when something breaks. We're watching Phave's customer list for a business closer to our clients' size before recommending anyone buy it outright. The pattern it's proving out, we're already shipping in smaller form.
What I'd Tell a Client Asking About This#
Skip it, if you're not already paying enterprise marketing-automation-platform prices. This isn't a decision you need to make this week or this quarter.
If you are evaluating a platform swap, don't buy the "AI reasoning" story on the campaign-speed number alone. Ask the vendor for the failure modes, not just the wins. And if what actually appeals to you is the underlying idea, messages that adapt to a person instead of following a script, tell me what you're running now. There's a real chance we can build that judgment layer into your existing WhatsApp or email flow for a fraction of $36,000 a year.
If you want to know whether this is worth chasing for your setup, send me what you're currently running and I'll give you a straight answer. Takes two minutes, no pitch attached. cal.com/webepex/growth-review
Related reading: our take on what SaaS MVP development actually costs, how we price Meta ads management for service businesses, and the n8n security fixes worth patching if you're running workflows like the one described here.
