Frequently asked questions

Phave is an AI-native marketing automation platform, built to replace legacy platforms like Marketo, HubSpot, Pardot, and Eloqua. Those platforms decide by rule. Every segment, every score, and every email is something a specialist had to specify in advance.

Phave uses AI to reason instead, about every person, account, and buying group. Maestro, the intelligence inside Phave, computes a Playlist for each person: the touches most likely to reach your objective, ordered and timed for that individual, following the rules and governance you set.

You describe what you want in plain language rather than building it as configuration. Do that in Phave, in the AI your team already uses, such as Claude, ChatGPT or Gemini, or headlessly through your own code.

Phave was founded by Jon Miller, co-founder of Marketo and Engagio, and Nick Bonfiglio, formerly EVP of Global Product at Marketo.

Phave starts at $36,000 a year. Pricing is based on Monthly Active Playlists: the number of people who receive something from you in a given month. Legacy platforms price on database size, so you pay for every record whether you contact it or not, which is why teams end up deleting people to stay under a cap.

Every subscription includes unlimited accounts, buying groups and opportunities, ten users, AI scoring, and a set of Maestro fields that refresh automatically. Everyone in your database gets a playlist, and Air Traffic Control resolves the conflicts across your campaigns, so nobody receives three emails on the same morning.

Additional Maestro fields, additional scoring models, and Maestro Playlists cost extra. With Maestro Playlists, Maestro fills the open gaps between your scheduled campaigns with the touches most likely to move each person, and runs the whole sequence itself where nothing is scheduled. Think of what’s included as adaptive cruise control, and Maestro Playlists as self-driving.

There is no charge for tokens and no metered AI usage. The price does not move with how much reasoning Phave does for you. You commit to a band for the year, and running consistently above it is a conversation about the right band rather than a surprise invoice.

Phave was founded by Jon Miller and Nick Bonfiglio. Together, they helped build the marketing automation category they are now replacing.

Jon Miller co-founded Marketo and served as its CMO, where he helped define modern marketing automation and the demand generation playbook. He then founded Engagio, an account-based marketing platform, and led it as CEO until Demandbase acquired it in 2020, after which he served as Demandbase’s Chief Product Officer and CMO.

Nick Bonfiglio was EVP of Global Product at Marketo from 2009 to 2016, running engineering, operations, product, and support as the company grew from a few million in revenue to several hundred million. He went on to found Aptrinsic, now Gainsight PX, and to co-found Syncari.

They have built marketing automation software before, and they know from the inside which parts of a platform a marketing team actually depends on.

Phave works by reasoning about each person, account, and buying group rather than just applying rules. That’s because rules are good at what must be true, but bad at ambiguity and judgment. In Phave, customer journeys, scoring, segmentation, routing, even persona and buying group mapping are all decided using AI reasoning.

Data. Phave connects to whatever data and context you need: two-way CRM sync for people, accounts, and opportunities, your website, your forms and email engagement, intent data, third-party enrichment, and more. People, accounts, opportunities, and buying groups are records in their own right, each with a score, a lifecycle stage, plus context and memory that build over time from how they behave.

Campaigns. You describe what you want in plain language and Phave builds it: the audience, the email, the schedule, the campaign structure. Your standards, your definitions, and the way your team works are written by Marketing Operations as Skills, and Phave enforces them from then on.

Playlists. Every person has a playlist, the touches planned for them across the next 30 days. Air Traffic Control reviews each one against your frequency limits, quiet hours, and delivery calendars, then approves, reschedules, or suppresses it. Maestro Playlists go further, choosing the touches themselves against an optimization objective.

Every AI decision in Phave writes a trace showing what was considered, what was chosen, and why, so your team can audit the reasoning rather than trust it.

AI-native marketing automation is a platform where the AI makes the decisions, rather than a rules platform with AI features added to it.

Every platform has AI now. The test is whether the AI decides or only assists. Drafting a subject line, predicting a score or recommending content is assistance. Underneath, a person still drew the flowchart, wrote the rules that act on the score, and set the conditions on every branch.

Two things change with an AI-native MAP:

Deciding: audiences and segmentation, scoring, routing and persona mapping, data quality, and what each person gets and when are reasoned from evidence at the moment they are needed, including the cases nobody anticipated.

Authoring: you describe what you want in plain language instead of translating it into configuration, so work stops queuing behind whoever knows the platform.

Reasoning AI needs context. This includes how your business works: your standards, your definitions, the way your team does things, written once in plain language as skills and applied from then on. It also includes what the platform knows about each person, account, and buying group, built from how they behave over time. Without both, the reasoning produces generic output. That is the thing to test when a vendor calls itself AI-native.

AI-native marketing automation is new because the capability is. Reasoning models became reliable enough to decide only after the current generation of platforms was designed.

Phave is an AI-native marketing automation platform, built to replace a legacy MAP.

The MQL, or marketing qualified lead, is dead as most companies practice it. It started as a contract between marketing and sales. Marketing passed a lead only when it had strong evidence that this person was worth sales attention now, and sales committed to following up quickly. Marketing earned credibility by being selective, and sales got better conversations.

But pipeline pressure turned the scoring into a lever for volume rather than a filter for quality. Any campaign responder became an MQL. Sales cherry-picked the obvious hand-raisers, ignored the rest, and the contract broke.

Leads don’t buy B2B solutions. Companies and buying committees do. One person on a pricing page might be noise. Three people from the same account in the same week is evidence, even though no individual among them would have triggered a score.

That’s why the MQL is giving way to the Marketing Qualified Account (MQA), an account showing evidence of buying activity across more than one person, and the Qualified Buying Group (QBG), the buying committee for a specific purchase, showing evidence of active buying.

Phave co-founder Jon Miller helped popularize the MQL at Marketo and has published what should replace it. Phave maintains accounts and buying groups as real records, each carrying its own score, its own stage, and, for a buying group, which committee roles are filled and which are missing.

A revenue marketing platform is one system for both person-based marketing automation and account-based marketing. Most B2B teams run two: a marketing automation platform for people and an ABM platform for accounts, with data moving between them. The buying group falls between the two, which is why neither system has ever owned it well.

The category is Forrester’s, formalized in a Wave published in January 2026 that evaluated ten providers. Adobe and Salesforce came to it from marketing automation, Demandbase and 6sense from account intent data and advertising, and all four were named Leaders. Vendors arriving from opposite directions is what tells you the two disciplines are merging.

To qualify, a platform needs:

Data unification across accounts, people, and buying groups.

Adaptive journeys that change with what each person does.

Account orchestration combining signals with touches.

Full lifecycle support, from anonymous through pre-sale and post-sale.

The practical question is whether you want one system for people and another for accounts, or a single system that does both. That choice decides what else you have to buy, and how much work goes into keeping two databases agreeing with each other.

Phave is a revenue marketing platform. People, accounts, and buying groups share one database, journeys adapt per person, orchestration runs on signals, and coverage runs from anonymous visitors through post-sale. Forrester published its Wave in January 2026; Phave reached general availability in August 2026 and was not evaluated.

AI campaign orchestration is deciding what each person receives — and when — by reasoning about that person at the moment of the decision, rather than routing them through campaign flows drawn in advance.

In a legacy marketing automation platform, the campaign is the unit of decision. A person enters a flow, waits, branches on a rule, and receives what that branch specifies. Two campaigns aimed at the same person know nothing about each other, so coordination becomes manual: exclusion lists, a shared send calendar, and someone checking that three teams are not all mailing the same person on Tuesday.

AI campaign orchestration moves the decision to the person. The system reasons over what it knows about that individual, their account, and their buying group, then computes the sequence most likely to reach the objective. It arbitrates across every campaign at once rather than inside one, recomputes as evidence changes, and records why it chose what it chose. A content recommendation feature does none of that.

Frequency limits, quiet hours, and send timing stay deterministic, because they are rules rather than judgments, and a system that guesses at them will mail someone at 3am.

In Phave, Air Traffic Control reviews every planned touch across a person’s next 30 days and approves, reschedules, or suppresses each one before it sends.

A B2B buying group, also called a buying committee, is the set of people at one account who research, evaluate, and approve a single purchase together. Gartner’s 2024 B2B Buying Survey puts a complex B2B purchase at six to ten of them.

A lead is too narrow. It tells you one person is engaged and nothing about the other nine. When three people from a six-person committee engage, you get three MQLs, often routed to different reps, and nobody sees the committee.

An account is too broad. An account has one status, so one closed deal makes everyone at the company a customer, including people who had nothing to do with that purchase and may be buyers for something else.

Forrester has argued for years that the right unit is the buying group: broader than a lead, narrower than an account, and matched to how companies actually buy. Few teams work that way, because maintaining a buying group has meant maintaining it by hand. Someone guesses at the personas and reps hunt for matching contacts. Which of eleven contacts at an account belong to this opportunity rather than the renewal is a judgment call rather than a title match.

Reasoning AI is what makes buying groups practical. Give a model titles, engagement patterns, and the committees that closed before, and it works out who belongs to which opportunity, what role each plays, and who is missing, continuously, for every account.

A group becomes usable through coverage, meaning which committee roles are filled and which are missing, and engagement across the filled seats. A group with a champion and no decision maker is a specific, fixable problem.

In Phave, buying groups are derived from persona classification against a committee model you define, with coverage, engagement, and a score of their own.

An AI agent in marketing automation takes actions in the platform rather than only drafting content for someone to paste in. It builds the list, adjusts a scoring model, creates an automation, manages a workflow, writes the email, and schedules the send. These agents can be built into the platform, as well as agents connecting from outside.

AI agents can run marketing automation; how safely depends on what the agent acts on, whose permissions it acts with, and what it cannot do without a person.

Identity: an agent connecting through a shared service credential appears in the audit trail as a machine, and afterward nobody can say who authorized what. Acting as a named person is what makes the record answerable.

Permissions: an agent should never do something the person behind it could not do in the interface. Anything else turns an assistant into a way around access control.

Irreversibility: reading data, drafting an email, and building a list are all recoverable. Sending to your whole database is not. Those two categories need different rules, and a platform that treats every action as one class is asking you to trust the model instead of governing it.

In Phave, your own assistant connects over MCP as the operator who consented, with exactly that person’s permissions and no more. Every write is classified reversible, destructive, or send, and a send can only be approved by a signed-in person on a Phave page.

Marketing automation is a category of software that streamlines, automates, and measures marketing tasks and workflows, so companies can run marketing more efficiently and grow revenue faster.

Marketing automation is the technology underneath most B2B marketing practices: lead generation, segmentation, nurturing and scoring, relationship marketing, cross-sell and upsell, retention, and measuring marketing ROI. Some of these are possible by hand at small volumes. At any scale the technology becomes essential.

Marketing automation is not another name for email marketing. An email tool sends to a list. Marketing automation maintains the database, tracks what every person does across your website and your campaigns, connects to the CRM so sales works from the same record, measures the result, and serves as the system of record for every campaign and program you run.

With legacy marketing automation platforms, a specialist specifies the programs in advance, defining each segment, writing each scoring rule, and drawing each branch, and the platform executes what it was given.

Two things are changing that. Reasoning models can now decide rather than execute what a person specified. And B2B teams are organizing around buying groups rather than individual leads, which the lead-shaped databases underneath legacy platforms were never built to hold.

Phave is an AI-native marketing automation platform, built to replace legacy platforms like Marketo, HubSpot, Pardot, and Eloqua.

There is no single best marketing automation platform for B2B. The right choice depends on which constraint is binding: what your stack looks like, whether you need a website and CMS alongside marketing automation, and how much of the work you want the platform to decide rather than you specify.

HubSpot: strongest if you want website, CMS, blog, and social publishing in one system alongside marketing automation, and your team is small enough that one platform beats several.

Marketo: the default for large enterprises in the Adobe stack, with the largest partner and certification ecosystem in the category. That ecosystem is also the cost, because most teams need a certified specialist to operate it.

Salesforce Marketing Cloud Account Engagement, still widely called Pardot: the easiest to get approved if you already buy Salesforce, since it arrives on a contract and a security review your company has already done.

Oracle Eloqua: the fit for large enterprises with complex compliance and approval requirements, particularly inside the Oracle stack.

Inflection.io and Conversion.ai: newer AI-native entrants, suited to smaller B2B teams that want something newer than a legacy platform and do not need more advanced functionality.

Phave: the choice if you are replacing a legacy platform and want the AI to make the decisions, reasoning about every person, account, and buying group instead of following rules a specialist wrote in advance.

The Phave marketing automation requirements analysis scored 481 requirements drawn from twelve enterprise evaluations. Phave’s weighted score is 2.97, against 2.46 for HubSpot and 2.38 for Marketo, and Phave matches or beats every vendor evaluated on at least 90% of the requirements. Scored September 2026.

The best Marketo alternative depends on why you are leaving Marketo.

Cost. HubSpot Marketing Hub and Salesforce Marketing Cloud Account Engagement are the common lower-cost choices, and Account Engagement is easiest to approve if you already buy Salesforce. Both price on database size, the same model as Marketo. Phave prices on Monthly Active Playlists, the people who receive something in a given month, so the bill follows what you send rather than what you store.

Complexity. Marketo’s power comes from configurability, which requires certified specialists and leads to bottlenecks. Some companies switch to HubSpot for simplicity, at the cost of flexibility. In Phave you describe what you want in plain language and the platform builds it, which keeps the flexibility without the bottleneck.

Support and innovation. Companies switch because they are frustrated by the support they get and by a product that has not moved. New capability now lands in Adobe Journey Optimizer B2B Edition, a separate product you buy on top of Marketo.

Leads, not buying groups. Marketo was built around emails and leads. B2B teams now need the buying group and the full lifecycle through retention, upsell and cross-sell. Adobe’s answer to buying groups is Journey Optimizer B2B Edition, which is again the upsell rather than the platform.

AI transformation. Marketo has made some investments in AI and shipped an MCP server, and both make its regular objects faster to build while the underlying system remains a legacy rules-based platform. Companies are seeking out AI-native alternatives, including Inflection.io, Conversion.ai, and Phave. Phave is the one that reasons about each person, computing what that individual should receive and when rather than executing rules someone wrote in advance.

Phave was founded by Marketo co-founder Jon Miller to replace it. People, accounts, and buying groups are records in their own right, each scored and staged, so one platform covers the whole lifecycle.

The best Pardot alternative depends on whether you stay inside Salesforce. Teams look outside because Pardot was built for simpler B2B programs, and they outgrow its automation depth, its reporting, and its handling of anything beyond the lead.

Staying inside Salesforce. Salesforce has moved new development to Marketing Cloud Next, built on Data 360 with Agentforce, and every Account Engagement tier now bundles a Marketing Cloud Next edition and Data 360 alongside classic Pardot. Account Engagement Advanced+ is a real improvement, scoring 2.45 against 2.03 for classic Account Engagement in Phave’s marketing automation requirements analysis of 481 requirements from twelve enterprise evaluations, September 2026. The cost is coexistence and complexity. Salesforce documents no automated converter, so Engagement Studio programs, automation rules, dynamic lists, completion actions, and scoring categories have no carry-over path. For most of the first year you run two builders, two data models, and two reporting stacks.

Leaving for another legacy marketing automation platform. HubSpot Marketing Hub suits teams that want website and CMS in one place, though it remains, at its core, a rules-based system. Marketo suits teams needing more configurability than Pardot offered, but it increases the need for a specialist rather than removing it.

Leaving for an AI-native platform. Common choices are Phave, Inflection.io, and Conversion.ai. Phave scored 2.97 in the 481 requirements, against roughly 2.2 for Conversion and below 2.0 for Inflection. Phave replaces the legacy platform outright, with its own database, campaigns, forms, scoring, buying groups, and CRM sync. The AI reasons across all of it: scoring, segmentation, routing, persona and buying group mapping, and what each person receives and when.

Phave was founded by Jon Miller, co-founder of Marketo, and Nick Bonfiglio, formerly EVP of Global Product at Marketo.

Phave syncs bidirectionally with Salesforce, so the CRM stays and only the marketing platform changes.

The best Eloqua alternative depends on why a company uses Eloqua in the first place. Do they need things like multi-region requirements, deep approval and compliance controls, and campaign logic complex enough to justify a specialist team? Teams look for Eloqua alternatives because of cost, a shrinking pool of people who know the platform, and an interface and roadmap that have not kept pace.

If the enterprise requirements are real, the list of possibilities is short. Marketo is the most common legacy choice, with the largest specialist pool but the same lack of innovation, complexity, and dependency on that specialist. Salesforce Account Engagement Advanced+ qualifies for Salesforce shops, but it means running two platforms and a customer data platform side by side. Phave is also an enterprise alternative. It is an AI-native MAP that scores 2.97 against Eloqua’s 2.32 in the marketing automation requirements analysis, across 481 requirements from twelve enterprise evaluations, September 2026.

For companies with less sophisticated needs, the field opens to HubSpot for breadth and to other newer AI-native platforms as well.

Three places Eloqua is ahead of Phave: it serves multiple data regions, it mails scheduled reports to a recipient list, and it calls an external API while composing a message, so the content reflects data at the moment of render.

Phave was founded by Jon Miller, co-founder of Marketo, and Nick Bonfiglio, formerly EVP of Global Product at Marketo, to replace legacy platforms like Eloqua.

Phave was founded by Jon Miller, the co-founder of Marketo, to be what comes next. Marketo is a rules-based automation system. Phave uses AI to reason about each person at the moment of decision.

Authoring: using Marketo means a specialist building things in the user interface. In Phave you describe what you want in plain language and Phave drafts it for you, following the skills and guidelines set by marketing operations.

Engagement: a Marketo engagement program is a stream, with content in a fixed order at a set cadence. Phave builds a playlist for each person, taking the full context of that person, their account, and their buying group, and crafting the sequence of actions most likely to reach the optimization objective. It recomputes as evidence changes and arbitrates across every campaign at once, so a person receives one coherent sequence without a separate manual process to keep campaigns from colliding.

Architecture: Marketo was built around people and leads, with accounts added later through a separate product. In Phave, people, accounts, and buying groups are records in their own right, each with its own score and lifecycle stage.

AI: Adobe has hosted a Marketo MCP server since May 2026, with more than 130 operations across forms, programs, smart campaigns, leads, and emails. It lets an AI assistant build Marketo’s objects faster, and the smart campaign still executes the rules it was given. It authenticates with an instance-level API credential, so the assistant acts as that service rather than as a person. In contrast, Phave registers 319 tools. Every connection acts as a named person with exactly their permissions and no more, and a send can only be approved by a signed-in human on a Phave page.

Nothing lost: across 481 requirements drawn from twelve enterprise evaluations in the marketing automation requirements analysis, Phave matches or beats Marketo on 90% of them, so a mature instance loses no critical functionality. Overall, Phave scores 2.97 against Marketo’s 2.38. Several legacy headaches are gone: duplicate sends are prevented by a database constraint rather than by careful campaign design, and CRM sync sets direction per object and per field, parks incoming changes for review, and records why any record was skipped. Source: Phave’s analysis, September 2026.

Marketo is stronger in places: twenty years of LaunchPoint integrations, deep connection to the rest of the Adobe stack, and a large pool of people who already know how to run it.

Migration takes as little as four weeks, and every customer gets a forward deployed engineer. You keep running in Marketo until you go live. Most of the time in a typical migration goes into rebuilding scoring models, nurture streams, and routing rules, and Phave needs none of them.

Choosing between Phave and Pardot means deciding whether to stay on Pardot and buy the rest of the Salesforce stack around it, or replace it. Pardot is now called Salesforce Marketing Cloud Account Engagement.

What you buy. New development goes into Marketing Cloud Next, built on Data 360 with Agentforce. Every Account Engagement tier now bundles a Marketing Cloud Next edition and Data 360 alongside classic Pardot, so a B2B buyer operates two marketing platforms and a customer data platform. Salesforce documents no automated converter between them, so Engagement Studio programs, automation rules, dynamic lists, completion actions, and scoring categories have to be rebuilt.

Complexity. Two builders, two data models, and two reporting stacks. Data 360 identity resolution runs at most once a day, standard segment publishing takes 12 or 24 hours, and only 20 segments per org get the faster 1 to 4 hour window.

Arbitration. Salesforce documents no global frequency cap, and nothing that decides which message a person should receive across every campaign at once. Einstein Engagement Frequency produces a recommendation a marketer wires into a split by hand, and its own model card places capping send frequency out of scope. Phave built Air Traffic Control for this.

Deciding. Marketing flows branch on conditions set in advance, the same model Marketo and Eloqua use. Phave computes each person’s next touch and its order from what that person and their buying group have done.

Coverage. In the marketing automation requirements analysis, Phave scores 2.97 against Account Engagement Advanced+ at 2.45, matching or beating Salesforce on 424 of 481 requirements, September 2026. The bundle is a real improvement over classic Pardot at 2.03.

Where Salesforce is stronger. Salesforce has data residency across 18 regions, SCIM provisioning and single logout, rehearsed disaster recovery, and malware scanning on uploads. Salesforce is also ahead on the warehouse, channels beyond email, and AppExchange connector breadth.

Replacing Pardot leaves Salesforce CRM untouched, and Phave keeps people, accounts, and opportunities in sync with it.

Phave and HubSpot are both marketing automation platforms, and they solve different problems. HubSpot bundles marketing automation with your website, blog, SEO, social publishing, and chat. Phave is an AI-native marketing automation platform built to replace a legacy platform outright: it reasons about every person, account, and buying group, arbitrates across every campaign at once, and gives marketing operations governance over what the AI may do. Across 481 requirements drawn from twelve enterprise evaluations in the marketing automation requirements analysis, Phave scores 2.97 against HubSpot’s 2.46, September 2026.

Where HubSpot is ahead. HubSpot does much more than marketing automation, and Phave does not compete with that. Website, CMS, SEO, and social publishing are outside what Phave treats as marketing automation, and Phave integrates with what you already run for them. Inside marketing automation, HubSpot leads on data residency, with hosting in five regions against Phave’s United States hosting, and on multi-touch attribution, identifying companies from anonymous website traffic, and pushing audiences to advertising platforms. HubSpot is the better choice when your team is small enough that one platform beats several, and your marketing automation needs are not sophisticated.

Where Phave is ahead:

• Reasoning instead of rules. HubSpot moves people through workflows someone drew in advance, and that stays true even when an AI agent drafts the workflow. Phave takes all the context it knows about a person, their account, and their buying group, and reasons from it: what they score, which segment they belong to, which rep should get them, which touch they should receive next, and when. New evidence changes the answer without anyone editing a workflow.
• Arbitration. Two campaigns both want the same person on Tuesday. HubSpot’s frequency cap, on Marketing Hub Enterprise, counts the emails and blocks the overflow, so the campaign that happens to send first wins and the other message is dropped rather than rescheduled. Phave’s Air Traffic Control reviews everything planned for that person across every campaign, then decides which touch goes now, which moves to a better day, and which should not send at all.
• Buying groups marketing can act on. HubSpot’s Buying Groups is a chart on the company record for a sales rep to read. In Phave, AI works out who belongs to the committee for a deal, what role each person plays, and which roles are missing, and keeps that current as people engage. The group gets its own score and lifecycle stage, campaigns can target it, and the Playlist optimizer decides for the whole committee at once rather than one person at a time.
• Governance. HubSpot’s agent logs record what an agent did. Phave also records why: every decision writes what was considered, what was chosen, and what was rejected, so your team can audit the reasoning. An AI assistant connects as the person who consented and can do only what that person can do, and a send needs a signed-in human to approve it. Phave scores 3.51 against HubSpot’s 1.46 on AI assistant, agents, and governance.

Companies pick HubSpot for the breadth around marketing automation. They pick Phave when the marketing automation itself has to be better.

Phave and Inflection.io are both AI-native B2B marketing automation platforms, but with different primary focuses. Inflection was built primarily around product usage, for product-led growth (PLG) companies. Phave was built as a complete legacy marketing automation platform alternative, with AI reasoning across all of it.

Both have AI agents to help streamline the work. They let anyone in marketing describe a campaign in plain language and have AI draft it, and both connect to Claude, ChatGPT, and Slack through an MCP server.

Inflection is ahead on product data. Users, organizations, and accounts are distinct objects in their model rather than an integration bolted onto contacts and companies, and onboarding, adoption, and expansion are primary motions rather than an extension of demand generation. They also sync directly with a data warehouse in both directions.

Phave is ahead on AI reasoning. Inflection’s agents generate the message, but a human still specifies which journey someone enters and what happens at each step. Phave reasons about scoring, segmentation, routing, persona and buying group mapping, and what each person receives and when. Inflection’s data model has contact, account, product user, and organization, but no buying committees. In Phave, the buying group is a record with its own score and lifecycle stage, showing which committee roles are filled and which are missing.

Inflection does enforce per-contact daily and weekly frequency caps, but nothing ranks two touches competing for the same person. In contrast, Phave’s Air Traffic Control reviews everything planned for a person across every campaign and decides which touch goes now, which moves to a better day, and which should not send at all.

Inflection has no landing page builder, and forms are embedded on your own site. It connects to Salesforce and no other CRM, with accounts, campaigns, and opportunities read-only and no custom objects. Governance is three roles, with no custom roles, no field-level permissions, and no approval workflow. Phave ships pages and forms, custom objects, more than one CRM in a single subscription, and an approval workflow on every content type across 112 named permissions.

Across 481 requirements drawn from twelve enterprise evaluations in the marketing automation requirements analysis, Phave scores 2.97 and Inflection scores below 2.0. Inflection was scored against its published product documentation in August 2026 rather than a live instance.

Inflection is the better choice for a product-led company where product usage is the signal that matters most. Phave is the choice when you are replacing a legacy platform and want the AI making the decisions across everything that platform used to do.

Phave and Conversion.ai are both AI-native B2B marketing automation platforms built to replace Marketo, Pardot, and HubSpot. Conversion has been in market longer and has more live customers. Phave was founded by Jon Miller, who co-founded Marketo and Engagio, and takes AI reasoning further: rather than making marketing automation just faster to operate, Phave reimagines how the software works with AI.

Both platforms have AI agents, an MCP server, and bidirectional Salesforce sync, and both let anyone in marketing describe a campaign in plain language and have AI build it.

As of September 2026, Conversion is ahead in four places. Conversion syncs directly with Snowflake, BigQuery, Databricks, and Redshift, bringing in contacts, custom objects, relationships, and events that can then drive audiences and triggers. Conversion sends audiences to Google, Meta, and LinkedIn for advertising. Conversion sends SMS, WhatsApp, and push messages. Conversion’s email templates support conditional logic and loops, so a message can list every line item on an order. Phave expects to ship warehouse sync, advertising audiences, and conditional templating shortly.

Phave is ahead in four places:

• Deciding what each person gets. In Conversion, a person builds the workflow and AI runs steps inside it. In Phave, AI reasons about the person, their account, and their buying group to work out their score, their segment, which rep should get them, and which message they should receive next and when.
• Handling two campaigns that want the same person. In Conversion, workflows run independently, and its documentation suggests adding a manual delay between sends to avoid overloading someone. In Phave, Air Traffic Control reviews everything planned for a person across every campaign and decides which message goes now, which moves to a better day, and which should not send at all.
• Marketing to the buying committee. Conversion reports on engagement by role across the contacts at an account. In Phave, the buying group is a record of its own, with a score and a lifecycle stage, showing which committee roles are filled and which are missing, and campaigns can target it directly.
• Controlling what AI is allowed to do. In Conversion, a connected AI assistant writes to the platform as soon as it is asked to. In Phave, every action an AI takes is classified by how risky it is, and anything that sends email needs a signed-in person to approve it.

Across 481 requirements drawn from twelve enterprise evaluations in the marketing automation requirements analysis, September 2026, Phave scored 2.97 and Conversion scored roughly 2.2. That places Conversion below Eloqua, Marketo, Salesforce, and HubSpot on the same analysis.

Conversion is a good choice if you want an AI platform with a longer track record and a data warehouse at the center of your stack. Phave is the choice if you want a platform that reimagines how marketing automation works with AI reasoning.

Phave compared itself to other marketing automation platforms by scoring 481 product requirements, for Phave and for each major platform, and published the result rather than describing it.

The requirements came from five completed enterprise RFP responses, one live RFI, five industry RFP templates, and the Forrester Wave evaluation criteria for B2B revenue marketing platforms, twelve sources in all.

Phave ran the analysis, and the scoring was done by Claude, reading Phave’s own source code and each competitor’s public documentation. Each requirement was scored from 0 to 4 for every platform: 4 meant fully supported and ahead of the field, 3 fully supported, 2 partial or delivered through a partner, 1 minimal or on a roadmap, and 0 absent. A capability a vendor described on its website but documented nowhere was credited at 2.5 rather than counted as absent. Each requirement also carried a weight from 0 to 4 for how much it affects which platform a buyer chooses, and every average was weighted by it.

Phave scored 2.97, ahead of HubSpot at 2.46, Salesforce at 2.45, Marketo at 2.38, Eloqua at 2.32, and Pardot at 2.03. Among the AI-native platforms, Conversion.ai scored roughly 2.2 and Inflection.io below 2.0. On the 214 requirements weighted 3 or 4, the ones that most affect which platform a buyer chooses, Phave scored 3.16 against 2.56 for the nearest.

Incumbents scored higher than Phave in two areas. The security and compliance envelope is one, where data residency across multiple regions is the clearest example. The breadth of the connector catalog is the other, since a platform that has been in market for fifteen years has more vendor-built integrations than one launched in 2026.

Eighteen requirements were weighted zero and excluded from every average, because they are outside what Phave treats as marketing automation: content management and blogging, SEO tooling, social publishing and listening, video hosting, and in-platform image editing among them. HubSpot scored well above Phave on all of them. Counting them as failures would misrepresent both platforms.

The individual comparisons on this page, covering Marketo, Pardot, HubSpot, Inflection.io, and Conversion.ai, each draw on the same analysis. The scores, the sources, and the method are published in the marketing automation requirements analysis. The full row-level assessment is available under NDA on request.

In Phave, a marketer launches a campaign by describing it to Maestro, the intelligence inside Phave, which builds the audience, the email, and the campaign itself.

What that replaces is a back and forth. A request arrives, marketing operations builds it, demand generation reviews it, and it returns for changes. Days pass, sometimes weeks. Across dozens of requests a month, most of a team’s capacity goes to execution rather than to strategic work.

Maestro has 319 governed tools over your own data and writes real campaigns, tactics, emails, lists, and segmentations, not drafts for someone else to assemble. It follows Skills, which are stored instructions marketing operations writes for how this team does things: how campaigns are named, which audiences are off limits, what a nurture email looks like. A campaign a demand generation manager builds alone comes out the way marketing operations would have built it. Maestro is reachable inside Phave and, through the MCP server, from Claude, ChatGPT, and Slack, so the request can start where the marketer already works.

Governance is what makes that safe. Maestro acts as the person using it and can do only what that person is permitted to do. Every kind of write is set to ask first, run unattended, or be refused outright. Drafts land editable, nothing is saved until someone acts on it, and sending requires approval.

Marketing operations owns the context and the governance: scoring models, the ideal customer profile, buying group definitions, playlist strategy, communication limits, and the Skills themselves. Setting how the platform behaves is the job, rather than building each campaign that runs on it.

Campaigns launch in hours in Phave, and the strategic work gets the time instead.

Phave integrates with Salesforce bidirectionally. People sync as both Contacts and Leads, along with accounts, opportunities, opportunity contact roles, campaigns and campaign membership, custom objects, record owners, tasks and events, and deletions.

More than one CRM at once. Multiple CRMs can run in a single Phave subscription; legacy platforms need a separate instance per org.

Bulk sync. A large update will not burn your Salesforce API or back up the queue. Phave moves volume through the Salesforce Bulk API and paces itself against the org’s hourly call budget, pausing a large pull and resuming it rather than pushing through, with nothing to configure.

Mapping you set field by field. Direction and conflict handling are per field, so a field Salesforce owns stays owned by Salesforce, and a field marketing owns is not overwritten from the other side.

Phave also syncs with Salesforce Campaigns, including creation when needed, pushing membership as people move through campaign steps and stamping UTM values on each new member, so Campaign Influence runs over Phave-sourced touches. Sync runs every five minutes, and a change in Salesforce can wake one immediately. A change Phave cannot apply is held for review with the record preserved, and every skipped record names the reason.

In the marketing automation requirements analysis, Phave scores 2.91 on CRM and sales alignment, against 2.53 for the strongest incumbent, September 2026. Every marketing automation platform syncs with Salesforce. Phave is the one built for what happens under load, across more than one org, and when something goes wrong.

Phave does not use your data to train AI models. Nothing in the product trains, fine-tunes, or evaluates a model on customer data, and there is no code in the product that could. No model that serves another customer is trained or tuned on your data.

Phave’s reasoning runs on models from Anthropic, OpenAI, and Google. Most of the time those calls go through Phave’s own accounts, on each vendor’s standard commercial terms, which do not use what is sent through the API to train their models. You can also use your own vendor keys, point Phave at a model you host yourself, or connect no model provider at all, in which case the AI features fall back to deterministic behavior and nothing leaves for inference.

What Phave records about an AI decision is the model, the numbers, and a hash of what was sent, not the text. Phave does build memory: facts and Maestro Memories accumulate on your people and accounts so later decisions are better informed. That memory is your data, in your database, and it makes Phave smarter about your market rather than about anyone else’s.

Phave reasons from your own context at the moment of the decision rather than from a model trained on anybody’s data.

Phave has a SOC 2 Type 2 report covering the Security Trust Services Criterion, issued by Sensiba LLP for the period 23 March 2026 to 22 June 2026, with the auditor’s opinion dated 15 July 2026. SOC 2 is an attestation from an independent auditor rather than a certification.

A Type 2 report tests whether controls actually operated across a period of months, where a Type 1 describes them as they stood on a single day. Sensiba noted no exceptions in its testing.

The full report is available under NDA.

Setting up and warming the sending domain. Phave helps you set up your SendGrid account and manages the provider relationship. Warming a new domain takes real sends from your side and records published in your own DNS, and Phave plans the ramp with you and watches the results as volume increases. A sending domain verifies only when both halves agree: the records in your zone and the provider’s own scan. Phave names the record that is missing or wrong and the value it expects.

Nothing sends that should not. Consent and the suppression list are checked when a touch is planned and again at the moment of sending, so an unsubscribe between those two points stops the send. Consent follows the recipient’s own jurisdiction rather than one rule applied everywhere. A bounce or a complaint clears only by operator action, never by anything a recipient does. Every send includes one-click unsubscribe and a preference center showing the recipient their own stored choices, which is what Gmail and Yahoo require of bulk senders.

Engagement numbers you can trust. Phave turns the provider’s click and open tracking off and serves its own. Machine opens and bot clicks are classified rather than counted, so Apple Mail Privacy Protection does not inflate your open rate and a bot cannot move a score or fire a trigger.

Gmail’s own view of your domain. Connect Google Postmaster Tools and Phave reads your spam complaint rate, authentication pass rates, and whether Gmail’s bulk-sender requirements are being met, with an industry comparison. The data covers Gmail recipients and arrives on a two to three day lag, so it shows a trend rather than the present moment. Inbox reputation problems also surface on the Phave home from your own sending data, each with a matching remediation path in Maestro.

Air Traffic Control reviews everything planned for a person across every campaign, so volume does not accumulate from sends that were each reasonable alone.

Your domain’s reputation is built over months, and Phave is built to protect it: authentication verified rather than assumed, hygiene enforced at the moment of sending, engagement measured first-party, and volume arbitrated across every campaign at once.

Phave keeps consent rules in one Settings object called a Consent Profile, which a campaign tactic uses. Inside it, you write a rule for a set of places: the EEA, Quebec, Germany. The narrowest rule that applies wins, so someone in Quebec gets the Quebec rule even when a broader Canada rule exists. Most teams run one profile with two forks, EMEA and Quebec.

One profile covers one form for every market. The form knows all of its rules, so when a visitor picks their location, the consent boxes, the wording, and the legal footer change on the page. No second form, no geo-IP lookup, no custom JavaScript, and opt-in boxes are never pre-ticked. The server rechecks the answer against the same rule on submit and keeps a copy of what the visitor agreed to, in a log keyed to the email address, so the record survives merges and deletions.

At send time, a default is not a choice. Where a country needs explicit opt-in, someone who never answered the question does not count as having opted in. If the profile a tactic uses is missing or unapproved, the send stops instead of falling back to something permissive.

The rules arrive already set. Add Germany to a sending rule and marketing email starts as explicit opt-in with a confirmed address required. Everything is editable, and the presets are starting points rather than legal advice.

What you end up with is one page your counsel can read, instead of a framework built from custom fields, segmentations, and campaign logic that someone maintains by hand.

Migrating from Marketo to Phave takes 4 to 12 weeks, depending on the complexity of your setup. Most of that range depends on you: how fast your security and IT review clears, and how quickly your team gets into Phave and starts building.

Migration is AI-assisted, with a forward deployed engineer working alongside your team. You keep using Marketo until the day you go live.

People, accounts, opportunities, and custom fields come across through your CRM connection, and Marketo-only records import directly. Unsubscribes, suppressions, and invalid addresses migrate with them, reconciled against the source list you name as definitive. Maestro imports your email templates or rebuilds them, and builds your forms from a description in chat. Landing page content moves to your own site behind a redirect map built while Marketo is still running.

There is less to rebuild than in a move between two legacy platforms. Scoring models, nurture streams, and routing rules are where most of a migration’s time goes, and Phave can compute all three rather than asking you to specify them. Data standardization programs become validation rules.

Two things do not migrate, and both are extracted before cutover: Data Value Change history, and analytics for campaigns you are retiring.

Phave scopes the migration during contracting, so the work is agreed before kickoff rather than discovered in week three.