Book a call
monbi
MONBI = MMO + NBI

Artificial Intelligence for
Marketing Mix Optimization
& Next Best Interactions

Marketing has lived in two separate compartments: aggregate advertising investment on one side, one-to-one customer relationships on the other. MONBI joins them — eight AI services that understand your market, decide where every euro goes, and optimise every individual interaction, inside one architecture that learns as it runs.

Start a pilotExplore the services
Test before you spend
Every decision is rehearsed on a digital twin of your market before it reaches the real one.
One optimisation logic
Paid media, pricing, creative, sales and affiliation optimised as one system, not silos.
Adopt service by service
Start with the service that solves today's problem. Grow into the full ecosystem.
Services

Eight services, one ecosystem.

Each one solves a problem on its own, and every one of them gets sharper when the others are running alongside it.

MarketClone console
MarketCloneMarket simulation

A digital twin of your market you can experiment on

Synthetic populations and multi-agent simulation turn market research into a permanent laboratory — test products, prices and campaigns before they reach the real market.

MarketClone builds an agent-based model of your market: thousands of synthetic individuals and companies whose demographics, psychographics, price sensitivity and funnel position respect the aggregate constraints you actually know. MCMC methods disaggregate those constraints into coherent individuals; Bayesian networks encode the causal chains between them; VAEs and self-organising maps turn the result into a navigable geography of the market.

Because the population is granular, you can interview digital clones of specific customers, run synthetic focus groups and surveys, and watch how a campaign, a price change or a new product propagates through the social network — including the segments you have never reached. It is also the sandbox where every other service trains before touching a real customer.

Simfonio console
SimfonioMMO

Where the next euro should go

Answers marketing's hardest question — where does the next euro go? — modelling carryover, saturation and word-of-mouth across every channel at once, for revenue today and brand tomorrow.

Simfonio is the only service with full visibility of every investment lever — paid media, direct marketing, affiliation, owned and earned — and the authority to decide how much budget each one gets, when, and why. It works on investment units rather than campaigns, so temporal coordination and per-lever allocation are optimised together.

Its objective function holds both horizons at once: conversion now and brand equity later. Adstock, fatigue and saturation mean today's impression still works weeks from now, and word-of-mouth means it reaches people it never touched directly. The plan is simulated on MarketClone first, executed on the real platforms, then recalibrated against what actually happened.

AcordIA console
AcordIADirect marketing

The next best interaction for every customer

The NBI layer: it decides the next best interaction for each customer — which touchpoint, offer, tactic and moment — across conversations and direct marketing, optimising conversion, satisfaction and lifetime value at once.

AcordIA treats the relationship with each customer as a sequence of decisions rather than a script. At every step it reads the state — profile, sentiment, inferred needs, objections, funnel position — and chooses the next best interaction: the touchpoint, the offer, the tactic, the tone and the timing that maximise conversion, satisfaction and future value together.

That covers live conversation and the direct-marketing lifecycle alike: which customers to approach, through which channel, with what, and when to stay quiet. It rehearses against MarketClone's digital clones before it ever reaches a real lead, and it works with any LLM because what it optimises is the decision and the context, not the model.

CatervIA console
CatervIAPoint-of-sale analytics

What visitors are really trying to do

Reconstructs the motivation and intent sequence behind every session, so you know what a visitor truly wants and how much their session is worth — in real time.

Two visitors can perform identical actions for opposite reasons: one rechecks a category because they are comparing before buying, another because they cannot find anything. CatervIA models each session as observable events generated by hidden states, separated into three levels — the visitor’s stable motivation, the intent that shifts as they browse, and the events themselves.

Forward-Backward and Viterbi reconstruct the most coherent explanation of the whole journey, not each click in isolation, and the model is regularised so every state keeps a meaning an analyst can read. The result feeds AcordIA with context before the first message, and reveals where interest is actually lost. The same logic applies to physical spaces once movement is captured.

ArtBait console
ArtBaitPaid media

Which creative will work, and for whom

Predicts which creative will work, and for whom, before you spend on media — then generates new variants aligned with brand and business objectives.

ArtBait reads an ad the way a strategist would: a multimodal model turns image and copy into a vector of style, mood, information content and brand positioning, and per-microsegment models estimate the response each audience will give. You get an expected impact before a euro is committed, plus a reading of how the piece moves perception across the five brand-equity axes.

Beyond scoring what exists, a planner generates new variants aimed at a specific objective, segment and channel — so creativity becomes a measurable, improvable process without losing the tactical judgement that makes it work.

KairosUnda console
KairosUndaDynamic pricing

Price, promotion and assortment as one decision

Optimises price, promotion and assortment across the whole portfolio — capturing substitution and cannibalisation instead of pricing product by product.

Price is not an isolated variable. KairosUnda models the portfolio as a graph where edges carry substitution, cannibalisation and transferred demand, calibrates the elasticities — own, competitor and internal — against real sales, and then sets discounts and priorities under uncertain demand.

Objectives are adjustable without retraining: lean towards revenue, occupancy or penetration as the moment requires. It reads real bookings from AcordIA, returns the pricing signal to Simfonio's global plan, and tells the sales layer what to offer, at what price, and in which order.

Legatus console
LegatusOwned media

Your brand present on every channel

Virtual ambassadors that hold a consistent identity across web, WhatsApp and social — autonomous where fitting, with a human in the loop where it matters.

Legatus —from the Latin for delegate, ambassador— creates digital representatives of a person or an organisation. Each delegate is a structured character: knowledge, values, style, channels and privacy rules, versioned so every correction consolidates into a new release rather than drifting.

Supervision is graduated: autonomous where that is appropriate, human approval where brand or regulation demand it, with a prioritised queue so only what genuinely needs judgement reaches you. One identity stays coherent across web chat, WhatsApp, X, LinkedIn, Instagram and TikTok — which is how you operate at scale without cloning your team.

KarmaNet console
KarmaNetAffiliate marketing

Community turned into growth

Rewards every contribution beyond the purchase — referrals, content, acquisition — turning a passive audience into an active network that grows with you.

Not everyone who interacts will buy, but many of them know someone who will. KarmaNet replaces the sales commission with Karma: points earned for referrals that convert, content, acquisition and quality of participation, with permanent attribution and double-generation rewards.

Because Karma is redeemable for access to the ecosystem's own tools rather than paid in cash, it incentivises without the legal and tax friction of an affiliate payroll. Simfonio decides how much Karma is worth investing per affiliate, based on the expected value of the leads they bring — so community becomes an investment line, not a cost.

Adoption

Three ways to bring monbi in.

Most companies already run ERPs, CRMs, martech and legacy systems, with decisions split across departments. Adoption enriches that reality instead of replacing it — you incorporate only the services that add value, at your own pace.

Strategy 1 · Embedded
In-house technical team
  • For organisations with teams able to maintain their own developments.
  • Services are partially embedded — only the relevant modules.
  • monbi contributes the technical know-how of models and services.
Strategy 2 · Integration
Integrating consultancy
  • For large corporations with a trusted technology consultancy.
  • The consultancy leads as prime integrator.
  • monbi provides the core models, algorithms and decision services.
Strategy 3 · SaaS
Managed service
  • For companies that prefer to focus on their core business.
  • monbi Labs builds the adaptation and integrates the models.
  • monbi Labs maintains and evolves the system on demand.
Blog

Notes on marketing, decisions and AI.

MONBI — Marketing Mix Optimization & Next Best Interactions
The bookOut now

MONBI: Marketing Mix Optimization & Next Best Interactions

The whole methodology, written down: why marketing split into aggregate investment and one-to-one relationships, why that split is now a technical accident rather than a necessity, and how AI lets a single architecture hold both. Available on Amazon.

Get it on Amazon →

The book builds the argument from the ground up. It starts with the division that shaped the discipline — mass advertising delegated to media agencies, customer relationships kept in house — and shows that the two evolved apart for a purely technical reason: no single model could hold an aggregate view and an individual one at once.

From there it develops the methodology service by service: simulating a market with synthetic populations and agent-based models, inferring the motivation behind observed behaviour, optimising the marketing mix as one system, pricing a portfolio rather than a product, treating creativity as a measurable process, deciding the next best interaction for each customer, extending brand presence through digital delegates, and turning a community into an affiliation network.

It closes on the practical part most methodology books skip: how any of this actually enters an organisation that already runs ERPs, CRMs and years of legacy — the three adoption routes, and the industry verticals the same core supports.

Simulation8 min

How to build a market you can experiment on

Companies know a small, biased slice of their market, but decades of research on human behaviour is stable. How MCMC, Bayesian networks, VAEs and self-organising maps turn known constraints into a plausible synthetic population.

Read article →
About

Built by people who ran the budgets they are optimising.

monbi Labs comes out of two decades inside marketing and commercial teams — the ones who had to defend a media plan, justify a discount, and explain last quarter's brand investment with the tools available at the time.

The services exist because those tools optimised fragments. We build the decision layer we wanted: rigorous enough for the model, plain enough for the meeting where the budget is decided.

Jesús Figueres Cañadas
Jesús Figueres Cañadas
Founder · monbi Labs

Twenty years across marketing, revenue management and analytics in consumer and B2B organisations. Now building the MMO + NBI ecosystem — from the market simulation layer to the conversation that closes the sale.

LinkedInjesusfigueres.com
Get in touch

Write something, I'm listening.

A question, a disagreement, a problem you are trying to solve, or just a thought on any of the above. Leave your name only if you feel like it.

Identifying yourself is entirely voluntary — anonymous notes are just as welcome.

Thank you — it's been received. If you left an email, you'll hear back.
¡habla conmigo!