Advertising's dirty secret is not that the industry lacks intelligence. It is that the intelligence arrives in six dashboards, three agency decks, an inbox full of approvals and one spreadsheet named something like FINAL_media_plan_v27_USE_THIS.xlsx. The campaign can be brilliant while the machinery beneath it wheezes. MINT, a New York-headquartered enterprise software company with Italian roots, has spent the past decade turning that machinery into a product.
Its name for the category is Advertising Resource Management, or ARM. The analogy is deliberately plain: finance has enterprise resource planning, sales has customer relationship management, and advertising should have a system of record too. MINT connects planning, buying, trafficking, performance data, approvals, optimization, reporting and financial reconciliation. It does not ask a global advertiser to throw out every ad platform or agency system. It sits across them, harmonizes what they produce and gives marketers, procurement teams and finance leaders a shared view of what is happening.
The product is the handoff
Most adtech products improve a slice of the job: bid more efficiently, make more creative versions, attribute a conversion, or draw a prettier chart. MINT's difference is that it treats the gaps between slices as the product. A brief becomes a plan; a plan becomes a buy; a buy becomes live placements; platform results become a report; the report must eventually agree with an invoice. Every handoff creates delay, duplicate entry and an opportunity for numbers to diverge.
One campaign, fewer loose ends
The platform's three layers explain its current pitch. Workflow automation handles repeatable steps. A unified data layer standardizes inputs from channels and enterprise systems. Specialized AI agents then work inside that governed environment. One agent might draft a media plan, another forecast key performance indicators, another explain why a budget should move from one platform to another. Decisions can be questioned, refined and logged. This is less magical than a chatbot running the department, which is precisely why it has a chance inside an enterprise.
The newest visible expression is MINT IQ, a mobile web app aimed at executives. It lets a leader ask about revenue impact, regional performance or campaign health, then assemble a deck or spreadsheet from the answer. Voice mode turns the phone into a data query interface. The promise is not more analytics; it is escaping the wait for somebody else to assemble the analytics before the meeting starts.
That makes the buyer unusually broad. A media planner wants fewer copied rows. An ad-operations lead wants clean trafficking. Procurement wants to compare agency commitments with delivery. Finance wants the authorized plan to match the invoice. The CMO wants one answer before the CFO asks for it. Publicly named customers include Mastercard, Air France and General Assembly; the expanded group also points to relationships with consumer giants such as Danone, Reckitt and Nestlé. The platform earns its keep only when those constituencies stop maintaining separate versions of reality.
“The industry doesn't need another point solution. It needs unified intelligence.”Louisa Wong, CEO of MINT
A pivot disguised as a new category
Founder Andrea Pezzi started Myntelligence in 2014. His route into enterprise software was not standard: before becoming a technology entrepreneur, he was an Italian radio and television presenter, including years as an MTV video jockey. That history makes MINT's obsession with the backstage of media oddly fitting. The company is concerned with what happens after the bright idea and before the result can be trusted.
The first model did not remain the model. In 2021, as Myntelligence became MINT, the company stopped directly managing clients' media budgets and committed to pure software sold as a service. That was the consequential change of mind: owning the workflow could scale better than operating every campaign. Tikehau Capital agreed that year to acquire 25 percent of the company, with the transaction price undisclosed, to support European growth and product development.
The switch also clarifies what “cost” means here. MINT is quote-based enterprise software, not a $49-a-month browser tab. The bill depends on markets, modules, users and integrations, but the expensive ingredient is organizational: documenting approvals, cleaning taxonomies, connecting systems and persuading agencies and regional teams to use the same workflow. If a company skips that work, it can purchase a single source of truth and still maintain several private sources of comfort.
The Air France proof
Air France began implementing MINT in January 2024 around four practical problems: cross-country reporting, flexible budget allocation, collaboration between internal teams and agencies, and visibility into global performance. The rollout started with four countries and automated reporting. It then expanded to ten countries and more of the campaign lifecycle, with capacity designed for as many as 100 countries. MINT says automated reports saved the airline's teams hundreds of hours.
Air France rollout: prove, expand, prepare
This is the useful operator lesson. Start where the waste is visible and the output is easy to inspect. Automated reporting is less politically threatening than autonomous buying. Once teams trust the shared data, the system can absorb planning, activation and optimization. The hard metric is not how many agents exist; it is whether campaign cycles get shorter, errors fall and finance can reconcile the spend without an archaeological dig.
Partners, purchases and the moat problem
MINT has filled capability gaps through partnerships and acquisitions. Celonis named it the process-intelligence partner for marketing and media, bringing process mining to approval delays, handoffs and compliance. IBM is collaborating on advertising-specific multi-agent systems built with watsonx. Early MINT customers using agentic workflows reported up to a 30 percent reduction in monthly hours spent on media operations and overhead. Snowflake underpins a newer data warehouse architecture for cross-channel campaign data and faster reporting.
Two acquisitions made the product broader. In February 2025, MINT acquired info³, adding deeper media planning, SAP connections and financial workflow automation. Redmill Solutions added media investment intelligence and a formidable operating map: it worked with 360 agencies across 172 countries, serving brands including Nestlé, Coty, Danone, PepsiCo, AB InBev and Mondelēz. Together, the deals move MINT closer to an end-to-end system from planning through reconciliation.
That breadth is also the defense against a familiar question: why not assemble this with a data warehouse, a work-management tool and the dashboards already included with ad platforms? Many companies will. MINT competes with Mediaocean, Skai, Smartly.io, Adobe, Salesforce and Workfront, but its stickiest alternative may be the homegrown stack nobody officially chose. Its answer is domain structure: more than 900 integrations and tracked media sources, mapped advertising workflows, proprietary approval and outcome data, and agents designed around narrow media roles.
Where the idea breaks
ARM is not automatically useful because advertising is complicated. A small brand buying on two channels with one approver may turn a simple operation into an implementation project. A business with inconsistent campaign naming, unclear ownership or hostile agencies will feed confusion into the new system faster. And an organization unwilling to let recommendations change budgets will get an expensive reporting layer instead of an operating model.
Good fit
Many markets, channels, agencies and approvers; recurring reporting and reconciliation work; leaders willing to standardize the process.
Bad fit
Few campaigns, clean native dashboards, simple approvals, low integration burden or no appetite for workflow change.
Agentic AI adds its own condition: permission must track evidence. Let an agent summarize before it publishes, recommend before it reallocates, and reconcile before it pays. MINT's “human-led, agent-operated” phrase works only if humans can see what an agent changed, why it changed it and how to reverse it. The company's Optimization Agent is a good example on paper because users can interrogate a proposed budget shift, add constraints and record the final decision.
What MINT has built is not a cure for messy organizations. It is a bet that messy organizations will pay for a faithful map of their mess, then use that map to remove steps. The category will stick if buyers can connect deployment costs to hours saved, errors avoided, faster launches and better media decisions. If it becomes another dashboard demanding manual updates, the spreadsheet wins again.
The stealable playbook
There are five moves worth borrowing. Name a neglected system problem instead of chasing a feature. Use a familiar analogy so buyers can place the budget. Integrate with what customers already own. Pilot on a measurable, low-drama workflow. Give automation more authority only after the data and audit trail earn trust. MINT did not invent media planning, reporting or reconciliation; it made the seams between them the main event.
That may be the most practical AI story in advertising. Creative demos get applause. A campaign launched on time, a budget moved with an explanation and an invoice that matches the plan get renewed.