Skip to content

RevPrime® marketing OS 2.0

Not a framework. Not a deck. An operating system that works.

RevPrime® started as a method to couple marketing back to pipeline. Today it's an operating system built from your data, a knowledge base holding the revenue logic of your business and specialized AI agents that work from it every day. The decisions stay with you. Version 2.0 succeeds the RevPrime® Marketing Hub I’ve already run client projects with. It adds a knowledge base at the core and agents that execute on their own after your go.

  • Knowledge base at the core
  • 8 core agents + your own
  • Human go
  • MCP-ready
  • runs in your stack
Read the video as text

The video has no speech, only music. Scenes and on-screen text:

  1. 0:00Logo: igromo · RevPrime® Marketing OS.
  2. 0:04“AI ships in days what used to take weeks.” Time to a finished campaign set: without AI 6–8 weeks, with chat AI 1–2 weeks, with RevPrime® agents 2–3 days.
  3. 0:12“Without a system you just ship the wrong thing faster.” Disorganized assets fly around: LinkedIn post, ad creative, landing page, email sequence, webinar, blog article and more.
  4. 0:18“Marketing that runs itself.” Agentic marketing systems for sales-led B2B SaaS & AI.
  5. 0:26“8+ agents. One knowledge base.” The eight core agents around the knowledge base, each with its output and cadence: signal, angle & content, paid media, pipeline reporting, outreach routing, inbox & ops, product truth, QA & brand guard. Then: “Working in your stack.”
  6. 0:38“8 core agents are just the start.” 20+ agents fully built out, depending on the business, e.g. for events, competition, expansion, partners, security questionnaires or your own bottleneck.
  7. 0:44“Agents propose. You decide.” An agent log from 06:00 to 07:02 (QA check, content opportunity, tasks, account score, ad fatigue, weekly report) and three proposals waiting for a go that get approved. The loop: read, analyze, propose, your go, execute & log.
  8. 0:56“No agent changes anything without a go.” The permission ladder: L0 read, L1 propose, L2 write after go, L3 approved rule. Zero changes without approval, every action in the change log.
  9. 1:04“Measured on pipeline.” A chart shows marketing-sourced pipeline.
  10. 1:12Track record: 15+ years B2B SaaS & tech, €75M+ influenced revenue, 50+ clients with growth, 8 core agents plus your own. Quote from Thorsten Heilig, CEO of Paretos: marketing-sourced pipeline has more than doubled, and for the first time it's clear why.
  11. 1:22“One system. Built for you.” Product (RevPrime® OS) plus service (build & personalize) makes the solution: your marketing OS, running in your accounts and handed over with documentation.
  12. 1:34Mario Maier: “Not a consultant with AI slides. An engineer who builds the system himself.”
  13. 1:42“Find out where your marketing is leaking revenue.” Book a diagnosis call, 30 minutes, free.
RevPrime AI marketing hub – dashboards with pipeline by source, funnel and anomalies (demo data)

Example view with demo data · your hub is built in your design, with your data and for your topic.

Architecture

Five layers. One system.

Every layer has its own job, from signal to cockpit. At the center sits the knowledge base every agent works from. Together they turn marketing activity into a steerable system.

  1. 01

    Signal layer

    Data & signals

    Everything that tells you what's happening: CRM, ads, website, intent, enrichment, product docs, inbox & chat. Read for the analysis. The same tools are where the agents write when they execute.

    • CRM
    • LinkedIn Ads
    • Google Ads
    • GA4
    • Intent
    • Enrichment
    • Inbox
  2. 02

    Knowledge base

    The brain

    The revenue logic of your business in one place: ICP, buying center, message angles, priorities, product truth, rules and decision memory. Every agent works from it. Change something once and they all follow.

    • ICP
    • Angles
    • Product truth
    • Playbooks
    • Memory
  3. 03

    Agent layer

    The work

    Eight core agents for signals, content, paid media, outreach, reporting, ops, product truth and QA, plus whatever other agents your business needs. Each with a role, cadence and output.

    • 8 core agents
    • + custom
    • daily
    • weekly
    • on demand
  4. 04

    Human go

    The control

    Numbered proposals with a single go, or routines that run on rules you approved once. Then the agent writes straight into the tool. Change log with before/after.

    • Approval
    • Audit trail
    • Owner
  5. 05

    AI marketing hub

    The cockpit

    Tasks, dashboards, assets and timeline in one place, and queryable via MCP straight from Claude, ChatGPT or Cursor.

    • Tasks
    • Dashboards
    • Assets
    • MCP

Knowledge base · the core

First the knowledge base. Then the agents.

Most AI initiatives in marketing fail in the same place. The knowledge sits in prompts, slide decks and individual heads, so every person and every agent works with their own version of the truth. In RevPrime® every agent works from one shared knowledge base. A new agent is productive from day one, and when the strategy changes, every agent knows it the same day.

Agents amplify what’s there – clarity as much as chaos. Your agents will never be better than your knowledge base.

  • 1×

    change it, and every agent follows the same day

  • Day 1

    productive, because new agents read what they need from the KB

  • 0

    updates without a go, the knowledge base included

ICPBuying centerAnglesPrioritiesProduct truthBrand & legalPlaybooksMemorySIGANGADSREVOUTOPSTRUQAKnowledgebaseGo
Agents work from the KBUpdates only after a go

Eight modules. Each with an owner.

  1. 01

    ICP & segments

    Firmographics, tiers, exclusion criteria, win/loss evidence. Who you sell to, and who you don’t.

    Owner: Revenue lead

    Used by: SIG · ADS · REV · OUT

  2. 02

    Buying center

    The 3–8 roles per deal: pains, objections, the proof they need, how they buy.

    Owner: Revenue lead

    Used by: SIG · ANG · OUT

  3. 03

    Message angles

    1 angle = 1 segment: persona, use case, workaround, edge, outcome. Status: live, testing, retired.

    Owner: Marketing lead

    Used by: ANG · ADS · OUT · TRU

  4. 04

    Priorities & KPIs

    The few levers for the next 90 days, targets, score thresholds, SLAs between marketing and sales.

    Owner: Leadership

    Used by: SIG · ADS · REV

  5. 05

    Product truth

    Docs, release notes, pricing, approved claims, proof points and case studies.

    Owner: Product

    Used by: ANG · TRU · QA

  6. 06

    Brand, claims & legal

    The guardrails for everything that goes live: voice, wording rules, forbidden claims, compliance.

    Owner: Marketing lead

    Used by: ANG · OUT · TRU · QA

  7. 07

    Playbooks & rules

    Routing rules, activation thresholds, sequences and the routines you approved once.

    Owner: RevOps

    Used by: SIG · ADS · OUT · OPS

  8. 08

    Decision memory

    Every proposal, go and no is recorded with its reason, before/after and result, so the system can learn from it.

    Owner: System

    Used by: all agents

How it stays good

Rule of thumb: if an agent needs to know it to do its job well, it belongs in the knowledge base, not in its prompt.

  • One source of truth

    Agents don’t carry their own knowledge. They read it from the knowledge base.

  • Structured & versioned

    Machine-readable files (Markdown, YAML) in a repo or workspace. Every change carries a version, owner and date.

  • Agents propose, owners decide

    New feature, retired angle, new threshold: agents propose updates, the owner gives the go.

  • Decision memory

    Every no flows back with its reason, so the next proposal gets better.

  • Data isn’t instructions

    Content from emails, chats or web pages never becomes a rule unchecked.

  • Runs in your accounts

    The knowledge base lives as files in your repo or workspace, separate from the model, and is handed over with documentation. Usage rights are set out in the proposal.

The revenue logic

The method stays. It now runs on agents.

Sequence isn't cosmetic. Build cadence without accountability and you build theater. Unleash agents without prioritization and you just produce the wrong thing faster. So RevPrime® sticks to the sequence, every time.

  1. 01

    Diagnosis

    I map pipeline, activity, owners and bottlenecks straight from your data.

    What the agents take over

    Agents pull data from CRM, ads and web and condense it into a diagnosis in days, not weeks.

  2. 02

    Prioritization

    The few levers that actually move pipeline in the next 90 days. Everything else gets stopped.

    What the agents take over

    Lever scoring by pipeline impact, time-to-impact and effort, stored in the knowledge base as the rulebook for every agent.

  3. 03

    Accountability

    Clear owners between marketing, sales and leadership. Shared KPIs nobody can explain away.

    What the agents take over

    Every task has an owner, whether human or agent. Every change is logged and approved.

  4. 04

    Operating cadence

    Daily agent loop, weekly pipeline review, angle review every 8 weeks.

    What the agents take over

    Agents prepare every weekly: what works, what doesn't, what needs a decision.

Agent catalog

Eight core agents. Plus your own.

These eight core agents cover the work almost every sales-led B2B team has. On top of that, I build the agents your particular business needs. None of them is a generic chatbot, and all of them work from the same knowledge base. I start with the two or three agents with the biggest pipeline leverage; fully built out it’s 8 to 20+, depending on the business.

SIGcontinuous

Signal Agent

Reads buying and account signals and decides the next move.

  • Condense web visits, engagement, CRM activity and intent into a score (0–100)
  • Route accounts into tracks: retargeting, outreach or sales hand-over
  • Alert sales the moment a buying center heats up

Works from: ICP · Buying center · Priorities · Playbooks

→ Account priorities & next best action

ANGon demand

Angle & Content Agent

Produces assets per message angle, not by gut feeling.

  • Derive landing pages, posts, email sequences and ad creatives from one angle
  • 1 angle = 1 segment: persona, use case, workaround, edge, outcome
  • Generate test variants within brand and legal guardrails

Works from: Buying center · Angles · Product truth · Brand & legal

→ Review-ready assets

ADSdaily

Paid Media Agent

Monitors campaigns daily and proposes optimizations with a hypothesis.

  • Spot budget pacing issues, CTR/CPC drift, ad fatigue and wasted search terms
  • Separate real leads from vanity conversions
  • Every proposal with expected effect, risk and how to measure it

Works from: ICP · Angles · Priorities · Playbooks

→ Findings & hypotheses to approve

REVweekly

Pipeline Reporting Agent

Makes marketing impact visible in pipeline and revenue.

  • Weekly pipeline report with traffic lights, trends and anomalies
  • Quantify marketing's contribution to the forecast
  • Prepares the weekly: what works, what doesn't, what needs a decision

Works from: ICP · Priorities · Memory

→ Decision brief for the weekly

OUTdaily

Outreach Routing Agent

Puts the right contacts into the right track.

  • Route accounts into owner 1:1, LinkedIn-first or sequence
  • Contact hygiene: valid email, role, region, duplicates
  • Safety stop: ends sequences automatically once sales takes over (approved rule)

Works from: ICP · Buying center · Angles · Brand & legal · Playbooks

→ Clean, prioritized outreach lists

OPStwice daily

Inbox & Ops Agent

Turns emails and chats into tasks with an owner.

  • Match new info from email & chat against the project state
  • Separate “waiting on you” from “waiting on others”
  • Numbered update proposals, written only after a go

Works from: Playbooks · Memory

→ Up-to-date task list & timeline

TRUweekly

Product Truth Agent

Makes sure marketing never promises what the product can't keep.

  • Match product docs & release notes against website and campaign copy
  • Flag new features as content opportunities
  • Contradictions turned into concrete correction proposals

Works from: Product truth · Brand & legal · Angles

→ Discrepancies & content opportunities

QAtwice daily

QA & Brand Guard Agent

Checks everything that goes live before someone else does.

  • Find broken links, mobile issues, outdated data and numbers
  • Enforce brand, claim and legal wording rules
  • Findings with a proposed fix, prioritized red/yellow

Works from: Brand & legal · Product truth

→ Health report with fixes

+ Your agents

Not a SaaS package. Built for your business.

The eight core agents are the starting point. Because every agent works from the same knowledge base, a new agent is mostly a new role: define it, connect it to your stack, take it live with a human go. It has your knowledge from day one.

Examples of custom agents

  • Webinar & event follow-up
  • Competitor & pricing intel
  • Expansion & churn signals
  • SEO & GEO visibility
  • Partner & channel
  • RFP & security questionnaires
  • Localization
  • Review platforms
  • 2–3agents to start, the ones with the biggest lever
  • 8core agents as the common base
  • 20+fully built out, depending on the business

Which agents you get is decided by the diagnosis alone.

How every agent works

Read. Propose. Go. Execute. Log.

Every agent follows the same pattern. That's why RevPrime® is fast and still controllable.

  1. 01

    Read

    The agent reads signals and state from your tools. Every analysis starts there.

  2. 02

    Analyze

    Matched against the knowledge base: revenue logic, goals, history and earlier decisions.

  3. 03

    Propose

    Numbered proposals: what, why, expected effect, risk.

  4. 04

    Go

    You approve item by item or reject. Rejections go into the decision memory of the knowledge base.

  5. 05

    Execute & log

    The agent writes the change straight into the tool, whether that's a landing page, a workflow or a budget. Before/after in the change log, measurement starts.

Governance by design

Full speed. Full control.

Agentic marketing only works if leadership trusts it. That's why control is architecture, not a feature.

  • Single go or approved rule

    Actions with external effect, such as outreach, publishing or budgets, need a single go. Recurring routines like audience upkeep or CRM notes run on rules you approve once and can revoke anytime. Then the agents write straight into the tool.

  • Complete change log

    Every change with before/after, reason, approval and timestamp. Traceable for leadership and audits.

  • Read to analyze, write after go

    Agents read what they need for the analysis and hold write access per tool and agent. That way approved changes land directly in the system: the landing page goes live, the HubSpot workflow fires, the opportunity moves.

  • Data is data, not instructions

    Content from emails, chats or web pages is treated as information, never as a command. Together with limited tool rights and approvals, this is part of the safeguards against prompt injection.

  • Aggregated data only

    The hub stores metrics and status, not unnecessary personal data. Built with GDPR in mind.

  • Every open task has an owner

    Human or agent, but never nobody. Accountability is part of the architecture.

Security & governance pack for IT and data protection

AI marketing hub

One cockpit for everything marketing is doing.

The hub is where humans and agents work together. For marketing, sales and leadership, it's the shared source of truth.

  • 01

    Work & status

    What's due, what's waiting on whom and what changed, with owner, timeline and “new since your last visit”.

  • 02

    Dashboards & analysis

    Pipeline, campaigns and accounts live, with traffic lights, trends, anomalies and a weekly report.

  • 03

    Knowledge base, docs & assets

    The knowledge base (ICP, buying center, message angles, product truth, rules) plus landing pages, content, ads and emails. Versioned in one place as the source for every agent.

  • 04

    Ask your hub

    Ask via MCP straight from Claude, ChatGPT or Cursor: “What's open this week?” The answer comes with sources from the system.

RevPrime AI marketing hub – agent overview with cadence, proposals and change log (demo data)

Example view with demo data · your hub is built in your design, with your data and for your topic.

A service, not a software subscription

Built for you. Not off the shelf.

RevPrime® isn't a SaaS package you subscribe to. I build and personalize your system: on a tested architecture, but tailored to your business, your stack and your brand. The three hubs here are the same system, built for three different companies.

  1. 01

    Diagnosis

    Which levers, which agents, which rules.

  2. 02

    Build & personalize

    Hub, agents and knowledge base in your stack and your design.

  3. 03

    Hand-over

    Documented for your team. The system runs on your side.

Vektorly Revenue Hub: AI marketing hub in the brand of a fictional company, Work & status view (demo data)
Vektorly· Revenue HubWork & status
Quorvia Growth Cockpit: AI marketing hub in the brand of a fictional company, Dashboards view (demo data)
Quorvia· Growth CockpitDashboards
Pexalo Marketing OS: AI marketing hub in the brand of a fictional company, Agents view (demo data)
Pexalo· Marketing OSAgents

Always the same: the tested core

  • Security & permissions per tool and agent
  • Human go, change log & audit trail
  • Agent management: roles, cadence, approval levels
  • Knowledge base structure with owners
  • Safeguards against prompt injection
  • MCP & integrations

Built for you: everything else

  • Your hub in your design

    Logo, colors, type and name of your brand.

  • Your agents

    Built for your bottleneck, not picked from a catalog.

  • Your revenue logic

    ICP, angles, playbooks and KPIs of your business.

  • Your rules

    What needs a single go and what may run as a routine.

  • Your stack

    Connected to the tools you use today.

  • Your accounts

    Runs in your repo or workspace and is handed over with documentation.

Fictional companies and logos, demo data.

Integrations

Works inside your stack. Reads and writes.

  • HubSpot
  • Clay
  • LinkedIn Sales Navigator
  • HeyReach
  • Instantly
  • LinkedIn Ads
  • Google Ads
  • Google Analytics
  • Search Console
  • Fibbler
  • Webflow
  • Lovable
  • Vercel
  • ClickUp
  • Slack
  • Notion

Selection. The agents read these tools for the analysis and write approved changes straight into them, from landing pages to audiences. New tools get added only when the revenue logic demands it.

Model-agnostic by design

Built on Claude by Anthropic, designed to be model-agnostic: knowledge base, playbooks and approvals are kept separate from the model. Other models and harnesses are possible, and every combination is tested for the specific deployment.

Models

  • ClaudeDefault
  • OpenAI
  • Gemini
  • Grok

Agent harnesses

  • Claude Code
  • Codex
  • Cursor
  • Hermes
  • OpenClaw
  • Grok Bot

What the system covers

Everything that touches pipeline logic.

  • ICP & segment logic
  • Message angles & positioning
  • Account signals & scoring
  • Pipeline steering
  • Marketing–sales alignment
  • KPI & reporting model
  • Channel mix & prioritization
  • Paid media monitoring
  • Content & demand logic
  • Outreach routing
  • Role & owner design (human + agent)
  • Marketing's forecast contribution

What RevPrime® is not

Not an agency. Not a tool. Not a fractional CMO.

vs.

Classic agency

Delivers output against a retainer. Scales with hours, not with a system.

RevPrime® builds a system you own that generates output itself.

vs.

Pure AI tools

Speed up single tasks. Solve neither prioritization nor accountability.

RevPrime® connects agents with revenue logic and governance.

vs.

Strategy advisory

Delivers slide decks. Leaves the room before execution starts.

RevPrime® runs every day, measurably, in your stack.

vs.

Fractional CMO

Takes over team leadership part-time. Strongly person-dependent.

RevPrime® is an operating system your own lead can run.

Trusted by 50+ B2B SaaS & AI teams

FAQ · RevPrime®

Frequently asked questions about the OS.

Is RevPrime® software I buy?
No. RevPrime® is an operating system built inside your environment: knowledge base with the revenue logic, agent prompts and playbooks, integrations, hub and governance. You don't get a subscription, you get a running system you own.
Which AI do the agents run on?
By default on Claude by Anthropic, as scheduled routines and sessions with firmly defined rules. The system is designed to be model-agnostic: knowledge base, playbooks and approvals are kept separate from the model, so it can also run with other models (e.g. OpenAI, Gemini, Grok) and agent harnesses (e.g. Claude Code, Codex, Cursor, Hermes, OpenClaw, Grok Bot). A switch is still not a single click: tool support, permissions, context limits and reliability differ, so every combination is tested and signed off for the specific deployment. Via MCP the hub can also be queried from Claude, ChatGPT or Cursor.
Do the agents execute changes themselves?
Yes. That's what sets RevPrime® apart from dashboards and consulting. After your go, the agents write directly into your tools: publish a landing page on Vercel or HubSpot, trigger HubSpot workflows, move opportunities, update audiences in LinkedIn and Google, start sequences in HeyReach or Instantly. Nobody has to copy anything or open a ticket, and every change shows up with before/after in the change log.
What happens when an agent is wrong?
You reject the proposal, and the agent remembers. Because nothing is executed without approval and every change is logged with before/after, every step stays traceable. Where it's technically possible, you can roll back. Irreversible actions like sent messages or spent budget always need a single go.
What about data protection and security?
Agents read for the analysis and only write after approval, using write access granted deliberately per tool and agent. The hub stores aggregated metrics rather than unnecessary personal data. Content from emails or web pages is treated as data, never as an instruction. That is a key part of the safeguards against prompt injection.
In what order is RevPrime® built?
Always the same: diagnosis → prioritization → accountability → operating cadence. First the revenue logic in the knowledge base, then the data connections, then the first agents, starting with the biggest pipeline lever.
Is it always eight agents?
No. The eight core agents cover the work almost every sales-led B2B team has and make it tangible how the system works. On top of that, I build the agents your business needs, for events, competition, expansion, partners or security questionnaires, for example. It usually starts with two or three; fully built out it’s 8 to 20+. Because all of them work from the same knowledge base, the system grows without steering getting more complicated.
What is the knowledge base, and why does it come first?
The knowledge base is the shared brain of all agents: ICP, buying center, message angles, priorities and KPIs, product truth, brand and legal rules, playbooks and the decision memory of every go and every no. Every agent reads its knowledge there. That’s why a new agent is productive from day one, and a change in strategy reaches every agent the same day. It lives as versioned files in your accounts, and every module has a human owner. Agents only propose updates; decisions are made with a go.
Does RevPrime® work with our existing team and stack?
Yes, that's what it's built for. RevPrime® plugs into CRM, ads, analytics and communication. Your team stays the owner of decisions; the agents take over routine.
Won't AI soon do all of this on its own anyway?
AI accelerates execution, but it doesn't replace revenue logic. Which ICP carries your plan? Which two levers decide this quarter? Which activity should be stopped? Those decisions stay with humans. RevPrime® makes sure agents do the right things faster.

Next step

Ready for marketing that runs itself?

A 30-minute diagnosis call: I look at your setup and show you where pipeline leaks and which agents would have the biggest leverage first. You get that even if RevPrime® isn't a fit.

Free · directly with Mario · reply within 24h

Rather send a quick message? Message me on WhatsApp