For more than a century, Madison Avenue has been closely tied to the advertising industry. New York agencies helped shape famous brands, launch national campaigns, influence consumer culture, and decide where billions of dollars in advertising money should go. Even as advertising moved from newspapers and television toward search engines, social media, ecommerce, and streaming platforms, New York remained one of the most important centers of the global marketing industry.
Artificial intelligence is now creating another major shift.
This change goes far beyond using ChatGPT to write advertising copy. AI is beginning to influence almost every decision inside modern marketing. It can help determine which customer should receive an offer, which advertisement should be shown, how much a company should bid for an impression, which creative idea is most likely to work, how a brand appears in AI-generated answers, and which products an AI shopping assistant recommends.
New York has become an important testing ground for many of these technologies.
For this article, NYC Tech Journal examined a hand-built group of private New York companies working across advertising, marketing technology, ecommerce, consumer research, creative production, personalization, media buying, measurement, and AI search. We analyzed their founding years, publicly reported funding, product positioning, customer use cases, and the specific part of the marketing process each company is trying to improve.
The research points to a broader conclusion. New York is not simply producing a handful of AI advertising startups. The city is helping build an entirely new marketing technology stack in which artificial intelligence increasingly sits between customer data and marketing decisions.
The Short Version: AI Is Moving Deeper Into the Marketing Process
The most important change taking place on Madison Avenue is not that AI can produce more advertisements. The larger change is that software is beginning to make more of the decisions that determine how marketing actually works.
Some New York companies are focused on how consumers discover brands. Profound and Evertune help businesses understand how their companies and products appear inside AI-generated answers. Nudge is trying to connect AI discovery directly with ecommerce conversion so that a recommendation made inside an AI system can lead to a more relevant shopping experience.
Another group is transforming creative production. Mirage, the company behind Captions, makes it possible to create short-form video advertising with AI. Persado applies machine learning to the language brands use in marketing campaigns. Clinch combines creative production, personalization, media activation, and increasingly AI-driven orchestration.
Other companies operate closer to the marketing decision itself. Cognitiv applies deep learning to advertising. Chalice AI develops custom algorithms for media buying. Black Crow AI predicts customer behavior. Wunderkind combines identity signals with automated decisioning. Attentive is pushing AI deeper into lifecycle marketing across SMS, email, personalization, and customer engagement.
A final group is changing the infrastructure surrounding advertising. Adelaide uses machine learning to measure media quality through attention signals. Rokt applies machine learning to decide which offers should appear during high-value transaction moments. Suzy is bringing AI into consumer research so companies can understand markets and make decisions more quickly.
Together, these businesses show why the term “AI advertising startup” is becoming too narrow. The more useful question is which part of the marketing process AI will automate or improve next.
Why New York Is Such a Strong Market for Marketing AI
Marketing AI companies need more than talented engineers. They also need sophisticated customers who understand advertising, branding, media buying, ecommerce, consumer behavior, measurement, research, customer acquisition, and creative operations.
New York offers those customers in unusual concentration.
The city is home to major advertising agencies, global consumer brands, financial institutions, retailers, publishers, fashion companies, entertainment businesses, ecommerce operators, media owners, and technology platforms. This creates an environment where founders can build products alongside companies that already spend heavily on marketing and understand the problems that advanced advertising technology is supposed to solve.
The concentration is visible in labor-market data as well. U.S. Bureau of Labor Statistics data for the New York-Newark-Jersey City metropolitan area estimated approximately 54,730 marketing managers in May 2025. The occupation had a location quotient of 2.27, which means marketing managers were more than twice as concentrated in the New York metropolitan economy as they were across the country as a whole.
Advertising and promotions managers were even more concentrated, with a location quotient of 2.77.
New York’s Marketing Talent Density
| Occupation | NYC Metro Employment | Location Quotient | Mean Annual Wage |
| Marketing managers | 54,730 | 2.27 | $207,770 |
| Advertising and promotions managers | 3,630 | 2.77 | $202,940 |
| Public relations managers | 7,940 | 1.74 | $208,800 |
Source: U.S. Bureau of Labor Statistics, New York-Newark-Jersey City metropolitan area, May 2025 occupational employment data.
These numbers help explain why New York is a natural environment for marketing AI.
A startup working on healthcare AI might need access to hospitals and doctors. A construction technology company might need access to contractors and developers. Marketing AI companies need access to brands, agencies, publishers, retailers, and marketing teams.
New York has all of them.
The city also gives startups access to difficult enterprise customers. That matters because software that works well for a small ecommerce store may struggle inside a global bank, national retailer, or multinational consumer company. Large organizations have complex approval systems, multiple technology platforms, regulatory rules, brand standards, customer-data restrictions, and many teams that need to work together.
When an AI startup successfully operates inside these environments, it often has to build stronger controls, better integrations, clearer reporting, and more reliable technology. That can become a long-term competitive advantage.
NYC Tech Journal Original Research: How We Studied New York’s Marketing AI Market
Rather than collecting every New York software company that mentions artificial intelligence on its website, NYC Tech Journal created a more focused dataset for this article.
Our research cut-off is September 1, 2026.
What Qualified for Our Analysis
To be included, a company needed a clear New York City headquarters or a strong claim to New York as its primary operating base. Artificial intelligence also had to play a meaningful role in the company’s current product rather than appearing as a small feature added to otherwise traditional marketing software.
The company’s main product also needed to influence marketing, advertising, customer engagement, ecommerce growth, media buying, creative production, consumer research, measurement, brand discovery, or another part of the customer acquisition and retention process.
We intentionally used a wider definition than a traditional martech list because AI is breaking down old boundaries. Search, ecommerce, creative production, advertising, customer service, research, and personalization increasingly overlap.
A company that influences how a product is recommended inside an AI assistant, for example, may affect marketing even if it does not look like a conventional advertising platform.
What We Left Out
We removed several important New York marketing technology companies because they are no longer independent startups.
Bluecore, for example, was acquired by Insider One in May 2026. Simon AI was acquired by Monetate in July 2026. Movable Ink also moved into a different ownership structure after receiving an acquisition investment from Symphony Technology Group.
These businesses remain highly relevant to the New York marketing technology ecosystem, but keeping them inside the core dataset would make it harder to compare independent private companies on a similar basis.
We also use the term “startup” broadly enough to include large private scaleups such as Attentive and Rokt. Although these businesses are much larger than early-stage startups, they still compete for many of the same enterprise marketing budgets and are actively shaping the direction of AI-powered marketing.
The 14 Companies in Our NYC Marketing AI Dataset
| Company | Founded | Primary AI Marketing Layer | Reported Funding Used in Analysis |
| Profound | 2024 | AI search and brand discovery | $155M+ |
| Attentive | 2016 | Lifecycle messaging and personalization | ~$866M |
| Rokt | 2012 | Ecommerce transaction decisioning | ~$488.5M |
| Mirage / Captions | 2021 | Generative video and AI ads | $175M+ |
| Wunderkind | 2010 | Identity and marketing decisioning | ~$163M |
| Persado | 2012 | AI-generated marketing language | ~$97.5M |
| Suzy | 2017 | AI consumer research and decision support | ~$113M |
| Black Crow AI | 2020 | Predictive ecommerce marketing | ~$40.3M |
| Evertune | 2024 | AI discovery and GEO | ~$20M |
| Clinch | 2013 | AI creative orchestration | ~$13M |
| Adelaide | 2019 | AI-powered attention measurement | ~$10.4M |
| Nudge | 2025 | AI shopping discovery and conversion | ~$1.1M |
| Cognitiv | 2015 | Deep-learning media optimization | Not consistently disclosed |
| Chalice AI | 2020 | Custom advertising algorithms | Not consistently disclosed |
Private company funding databases often disagree because some include debt, secondary transactions, older corporate entities, or financing rounds that other sources exclude. For that reason, the funding figures in this article should be treated as estimates based on publicly reported information rather than audited financial statements.
Even with that limitation, the numbers reveal several useful patterns.
Original Finding #1: Most Reported Capital Is Concentrated in a Few Companies
Among the 12 companies in our dataset for which reasonably comparable public funding totals were available, disclosed capital adds up to roughly $2.14 billion.
The total is impressive, but the distribution is even more interesting.
Attentive accounts for roughly $866 million, while Rokt represents approximately $488.5 million. Together, those two businesses account for about 63% of the reported funding in our comparable subset.
Adding Mirage and Wunderkind pushes the share controlled by the four largest companies to roughly 79%. The six most heavily funded companies account for around 91% of the total.
Reported Funding Across Our Dataset
| Company | Approx. Reported Funding |
| Attentive | $866M |
| Rokt | $488.5M |
| Mirage / Captions | $175M+ |
| Wunderkind | $163M |
| Profound | $155M+ |
| Suzy | $113M |
| Persado | $97.5M |
| Black Crow AI | $40.3M |
| Evertune | $20M |
| Clinch | $13M |
| Adelaide | $10.4M |
| Nudge | $1.1M |
This concentration reveals that New York’s marketing AI market has two very different layers.
The first contains mature private platforms that raised large amounts of venture capital during earlier software, ecommerce, and marketing technology cycles. Attentive, Rokt, Wunderkind, Persado, and Suzy fit largely into this group.
The second layer includes newer AI-native companies attacking marketing problems that either did not exist or were much less important a few years ago. Profound, Evertune, Nudge, Mirage, Black Crow AI, and Chalice AI represent this newer generation.
For marketers evaluating vendors, that distinction matters. The company with the most capital is not always the company positioned around the newest customer behavior. At the same time, the newest company may not yet have the integrations, data, reliability, or enterprise experience of a more established platform.
Original Finding #2: Younger Companies Are Targeting the Newest Marketing Interfaces
The median founding year across our 14-company dataset is approximately 2018.
Six of the companies were founded in 2020 or later: Black Crow AI, Chalice AI, Mirage, Profound, Evertune, and Nudge.
Yet these younger businesses represent only a relatively small share of total reported capital. Among companies where we could obtain comparable funding figures, businesses founded in 2020 or later account for roughly 18% of the funding represented in our dataset.
That might initially look like a disadvantage, but the product categories tell a different story.
Profound and Evertune are building around AI-generated search and discovery, a marketing channel that barely existed when companies such as Attentive or Rokt were founded.
Mirage is building AI-native video production infrastructure.
Nudge is connecting AI recommendations with ecommerce conversion.
Black Crow AI is applying prediction to customer intent.
Chalice AI is helping advertisers create custom machine-learning systems for media buying.
The older companies therefore have more capital and larger installed customer bases, while many of the younger companies are positioned around the newest customer interfaces.
This is not a market where one generation of companies has completely replaced another. Instead, the two generations are beginning to overlap and compete.
Original Finding #3: AI Search Is Becoming a New Marketing Channel
Three companies in our dataset focus primarily on what happens when customers discover brands through AI systems: Profound, Evertune, and Nudge.
Three companies out of fourteen may not sound like a large category, but the speed at which capital and product development are moving into this area is notable.
Profound was founded in 2024. By early 2026, it had raised more than $155 million and announced a $1 billion valuation. The company also reported serving hundreds of enterprise customers, including a meaningful share of Fortune 500 businesses.
Evertune was also founded in 2024. Its platform analyzes how brands and products appear across systems such as ChatGPT, Gemini, Claude, Perplexity, Microsoft Copilot, Meta AI, DeepSeek, and Google AI experiences.
Nudge is significantly smaller, but its thesis takes the idea one step further. Rather than measuring only whether a brand appears in an AI recommendation, the company is trying to connect that recommendation with product data, landing pages, merchandising, and ecommerce conversion.
These companies are responding to a fundamental change in online discovery.
Traditional search marketing assumed that customers searched for something, saw a page containing links, clicked one of those links, and continued their journey on another website.
AI-generated search can remove several of those steps.
A customer can ask an AI assistant for the best accounting platform for a 200-person company, request a comparison of three vendors, ask about pricing, identify potential problems, read summarized reviews, and receive a recommendation inside one conversation.
The customer may never visit the websites of most companies being considered.
That changes the marketing problem completely.
Brands increasingly need to understand not only where they rank, but also how AI systems describe them, which competitors are recommended instead, which external sources influence the answer, and whether the information presented by the model is accurate.
That is why AI-search visibility is likely to become a significant marketing software category.
Top Marketing and Advertising AI Startups in NYC
Profound — Building Infrastructure for AI Search
Profound is one of the clearest examples of how quickly a new marketing category can form.
The New York company was founded in 2024 by James Cadwallader and Dylan Babbs. Its early product was built around a simple problem: marketers could measure Google rankings, website traffic, social engagement, advertising performance, and many other digital channels, but they had little visibility into how their companies appeared inside AI-generated answers.
Profound built software to make that behavior measurable.
The platform helps brands analyze their visibility across AI answer engines, identify commercially important prompts, examine which sources influence answers, compare performance with competitors, and understand how frequently a brand appears in recommendations.
The company has since expanded toward AI agents that can help marketers act on those findings instead of simply reporting them.
By February 2026, Profound had announced a $96 million Series C at a $1 billion valuation, taking publicly reported funding above $155 million.
Why Profound Matters
Profound matters because it reflects a shift from search engine optimization toward machine recommendation optimization.
Under the traditional SEO model, marketers worried about questions such as whether they ranked first, fifth, or twentieth on Google.
AI discovery creates a different question.
When a customer asks an AI assistant which business software, hotel, insurance company, law firm, beauty product, healthcare provider, or ecommerce platform is best for a particular need, is the brand considered at all?
That question is harder because AI systems can draw information from many different sources. Reviews, publisher articles, Reddit discussions, company websites, product documentation, creator content, videos, comparison sites, and public databases can all influence what a model says.
The opportunity for Profound is therefore larger than building another SEO dashboard.
If AI assistants continue becoming major discovery interfaces, brands may need an intelligence layer dedicated to understanding and influencing how machines represent them.
For New York agencies, that could eventually make AI visibility a standard part of search, content, public relations, reputation management, and brand strategy.
Evertune — Treating Generative AI Discovery as a Marketing Channel
Evertune is approaching the same change from an advertising and media perspective.
The New York company was founded in 2024 by executives with deep experience in programmatic advertising, including early employees of The Trade Desk. It has reported approximately $20 million in funding, including a $15 million Series A.
The platform analyzes large numbers of prompts across major AI systems to understand which brands appear, how frequently they are mentioned, what language is used to describe them, and which sources appear to influence those answers.
Evertune is also moving beyond reporting. The company has been connecting AI visibility with paid media, content distribution, publisher relationships, affiliate marketing, and retargeting.
Why Evertune Is Strategically Interesting
One of the biggest unanswered questions surrounding generative engine optimization is where it will eventually sit inside a marketing organization.
It could remain closely tied to SEO.
However, it could also become part of media planning, brand strategy, public relations, affiliate marketing, and content distribution.
Evertune appears to be betting on the second outcome.
That makes sense because AI models do not always behave like traditional search engines. If particular publishers, creators, communities, and editorial sources repeatedly influence AI recommendations, marketers may begin treating those sources as important parts of their media strategy.
The boundaries between paid media, public relations, search, content marketing, and affiliate marketing could therefore become much less clear.
New York is especially well suited to this change because so many companies from those industries already operate in the city.
Attentive — Moving Lifecycle Marketing Toward Autonomous Decisioning
Attentive is one of the largest private marketing technology companies in New York.
Founded in 2016, the company became widely known for SMS marketing and later expanded into email, personalization, customer analytics, predictive technology, and AI-powered engagement.
Public private-market estimates place its total funding at roughly $866 million, making it the most heavily funded company in our core dataset.
Attentive belongs in this analysis because its product is increasingly moving beyond message delivery.
The company has introduced AI-powered customer journeys, predictive analytics, automated reporting, message generation, personalization, brand controls, and other capabilities designed to help marketers decide what should happen next for individual customers.
At its Thread 2026 event, Attentive described an increasingly agentic platform that could help manage orchestration, reporting, predictive analysis, and customer engagement. The company also reported that brands generated more than $6 billion through its platform during the first quarter of 2026.
Why Attentive Shows Where Marketing AI Is Heading
Generative AI receives enormous attention because people can immediately see what it produces.
Lifecycle AI could ultimately create greater financial value because it improves decisions that happen repeatedly and at enormous scale.
A retailer does not simply need another email subject line. It needs to determine which customer should receive an email, when it should arrive, which offer should appear, whether SMS would work better, which product should be recommended, and when sending no message at all is the better choice.
Large brands make versions of these decisions millions of times.
When AI improves those choices, personalization begins moving from a manually configured marketing tactic toward an automated operating system.
Rokt — Applying AI at the Moment a Customer Is Ready to Buy
Rokt was founded in 2012 and is headquartered in New York. Public private-market data suggests the company has raised approximately $488.5 million, while a 2025 secondary transaction valued the business at around $3.5 billion.
Its technology focuses on the transaction moment.
When a customer purchases a concert ticket, books travel, completes an ecommerce transaction, or reaches another high-intent stage, Rokt uses machine learning to decide which additional offer, message, or product is most relevant.
The company works with major consumer brands and platforms across ecommerce, entertainment, travel, payments, and retail.
Rokt has also made a substantial physical commitment to New York. The company announced plans to expand its Manhattan headquarters and create a larger research and development presence in the city.
Why Rokt Matters
Rokt represents a larger change in digital advertising: commerce itself is becoming media.
Advertising historically surrounded content. A television program, newspaper article, website, or social feed attracted attention, and advertisers paid to reach the audience around that content.
Retail media changed this model by allowing advertisers to reach customers closer to the point of purchase.
Rokt pushes the idea even further by focusing on the transaction itself.
The value of AI becomes clear because the number of possible combinations is enormous. The system can consider the shopper, merchant, transaction, available offers, timing, previous behavior, historical performance, and other signals before deciding what should appear.
That is exactly the kind of decision where machine learning can create meaningful value.
Mirage and Captions — Making Video Advertising Scalable
Captions started as an AI-powered video creation product and has since become part of a broader generative video business operating under the Mirage brand.
The New York company was founded in 2021. In March 2026, it announced $75 million in growth financing, bringing reported funding above $175 million.
The company also reported that more than 20 million people had used Captions and that more than 250 million videos had been created using its products.
For marketers, one of the most important developments is the ability to generate advertising video quickly.
A business can provide product information, create a script, select a digital actor, and produce UGC-style advertising without organizing a traditional production shoot. Programmatic interfaces also make it possible to create large volumes of video through software.
Why AI Video Changes Advertising Economics
The important change is not simply that video becomes cheaper.
It is that marketers can realistically test far more versions.
Traditional production encourages companies to create a limited number of polished advertisements because every additional shoot, actor, edit, location, and variation costs money.
AI changes the cost structure.
When it becomes inexpensive to create many different versions of an advertisement, marketers can test different hooks, messages, actors, product angles, offers, and formats at much greater scale.
That turns creative production into a data problem.
Once production becomes abundant, the scarce resource is no longer the ability to make another advertisement. The harder problem becomes knowing which advertisement should be produced, who should see it, and whether it actually improves performance.
That is why generative video platforms are likely to move closer to experimentation, audience data, media buying, and performance measurement over time.
Persado — One of New York’s Earlier AI Marketing Companies
Persado was applying machine learning to marketing language long before generative AI became a mainstream business topic.
Founded in 2012 and headquartered in New York, the company has raised approximately $97.5 million according to public private-market databases.
Persado helps large organizations generate and optimize marketing language using AI. Its technology is used across industries such as banking, retail, travel, telecommunications, and financial services.
The company has reported that many of the largest U.S. banks use its software and that its system generates large volumes of marketing messages every year.
Why Persado Still Matters
Persado demonstrates why enterprise marketing AI involves far more than asking a general-purpose chatbot to write copy.
Large companies must consider compliance, brand standards, performance history, legal approval, customer data, risk controls, measurement, and repeatability.
A general model may produce a clever sentence, but enterprise marketing systems need to determine whether that sentence is appropriate, approved, measurable, and likely to improve business results.
This creates room for specialized AI platforms with domain-specific data and enterprise controls.
The most useful content AI may therefore be less about producing impressive language and more about creating material that can safely move through a large organization’s marketing system.
Wunderkind — Combining Identity With AI-Powered Marketing Decisions
Wunderkind was founded in 2010 as BounceX and later became one of New York’s better-known performance marketing technology companies.
Public private-market estimates place its total funding at approximately $163 million.
The company’s core strength is identity resolution. Wunderkind helps brands recognize customers and visitors across devices, then uses behavioral information to determine when and how those people should receive marketing messages.
The company has increasingly expanded toward AI-driven personalization and autonomous marketing.
Its newer products combine identity information, customer behavior, email, SMS, product recommendations, and automated decisioning.
Why Identity Gives AI More Value
AI becomes more useful when it understands context.
A model can write an excellent discount message and still send it to a customer who would have purchased without receiving a discount.
It can recommend a product someone already owns.
It can send too many messages to a high-value customer.
It can make the right creative decision for the wrong person.
Identity and behavioral data give the decision system a better understanding of the customer.
This is why some of the strongest marketing AI businesses may ultimately be companies that already have access to valuable first-party signals.
The model itself can become widely available. Reliable customer context is much harder to reproduce.
Cognitiv — Using Deep Learning to Improve Media Buying
Cognitiv has one of the clearest AI-first advertising positions among New York companies.
Founded in 2015 and based in Manhattan, Cognitiv describes itself as a deep-learning advertising company. Its technology uses self-learning models to predict consumer behavior and optimize advertising activity across different parts of the customer journey.
Its platform can analyze user information, contextual signals, campaign history, and other inputs in real time before influencing how advertising budgets are spent.
Why Cognitiv Matters to Agencies and Brands
Advertising platforms have used algorithms for years, so the existence of automated optimization is not new.
The more important question is who controls the algorithm.
Large advertising platforms often optimize campaigns using systems the advertiser cannot fully see or change. Brands provide a campaign objective and allow the platform to decide how to achieve it.
Cognitiv represents a different model in which specialized AI can become part of an advertiser’s own competitive advantage.
That may become increasingly important as media buying becomes more automated.
If every advertiser can reach similar inventory through similar interfaces, the quality of the underlying decision system becomes one of the few remaining ways to outperform competitors.
Black Crow AI — Predicting Which Shoppers Are Most Likely to Buy
Black Crow AI was founded in 2020 and is based in New York. Public databases place its reported funding at approximately $40.3 million.
The company focuses on predictive intelligence for ecommerce brands.
Rather than simply reporting what shoppers did in the past, Black Crow tries to predict customer intent and make those predictions useful inside advertising, lifecycle messaging, analytics, and personalization.
The company’s $25 million Series A in 2022 was led by Imaginary Ventures and included investors with experience building major direct-to-consumer brands.
Why Prediction Could Be More Valuable Than Generation
This company illustrates another important distinction between visible AI and economically valuable AI.
Generated copy is easy to demonstrate. A company can show five headlines and immediately make the technology feel impressive.
Prediction is less visually exciting.
However, correctly identifying which shopper is likely to buy, leave, return, spend more, or respond to an offer can influence nearly every other marketing decision.
Accurate prediction can improve bidding, customer segmentation, product recommendations, email timing, SMS strategy, promotions, and retention.
That makes predictive intelligence one of the most important areas to watch as AI moves deeper into ecommerce marketing.
Adelaide — Using AI to Measure Whether Advertising Is Worth Buying
Adelaide is solving a different part of the marketing problem.
Instead of generating advertising or choosing customers, the company tries to measure the quality of media itself.
Founded in New York in 2019, Adelaide has raised approximately $10.4 million according to public private-market databases.
Its best-known product is AU, a metric designed to estimate media quality and the likelihood that an advertising placement will capture attention and influence business outcomes.
The platform uses machine learning and outcome data across channels that can include display advertising, online video, connected television, social media, traditional television, and podcasts.
Why Better Measurement Matters
Digital advertising has historically optimized around metrics that are easy to collect rather than metrics that perfectly represent value.
An impression does not mean someone noticed the advertisement.
A viewable impression does not mean someone paid meaningful attention.
A click does not necessarily prove the advertisement caused a purchase.
Artificial intelligence gives measurement companies more ways to model the relationship between media exposure, attention, customer behavior, and eventual outcomes.
If those models continue improving, marketers may be able to buy media based on expected quality rather than simply buying large volumes of impressions.
That may attract less public attention than AI-generated commercials, but it could have a much larger effect on advertising budgets.
Clinch — Connecting Creative Production With Media and Data
Clinch was founded in 2013 and is headquartered in New York. Public databases estimate that the company has raised approximately $13 million.
Its roots are in dynamic creative and omnichannel personalization.
Instead of showing every customer the same advertisement, Clinch helps brands create and manage multiple creative variations using information about audiences, products, location, context, campaign conditions, and other signals.
The company’s recent direction has expanded toward AI-powered orchestration and generative engine optimization.
Why Creative Orchestration Is Becoming Important
AI is creating a strange new problem for marketing teams.
Producing creative assets used to be expensive, which naturally limited how many assets a company could make.
Generative AI is removing that limit.
A company may soon be able to produce hundreds or thousands of variations without significantly increasing production costs.
That sounds useful until the organization has to decide which assets should be approved, which audience should receive them, which channels should use them, how they should be measured, when they should be changed, and when they should be retired.
The next generation of creative software will therefore need to manage creative decisions rather than simply generate creative output.
Clinch is positioned around that broader orchestration problem.
Chalice AI — Giving Advertisers Greater Control Over Their Algorithms
Chalice AI was founded in 2020 and is headquartered in New York.
The company develops custom machine-learning models that advertisers can use with their existing media infrastructure.
Rather than asking brands to move every advertising dollar into a new platform, Chalice focuses on helping them improve the algorithms that influence targeting, bidding, pricing, audience selection, and other campaign decisions.
Why Custom Algorithms Could Become More Important
Advertising platforms are becoming increasingly automated.
That makes campaigns easier to operate, but it can also make advertisers look more similar.
If two companies use the same platform, choose the same campaign objective, allow the same bidding system to optimize spending, and use similar AI-generated creative tools, they begin relying on nearly identical infrastructure.
That creates a risk of algorithmic sameness.
Custom models provide another path.
A company can optimize around its own definition of customer value, profit, retention, lifetime value, or business quality instead of relying entirely on a platform’s generic objective.
For companies with strong first-party data, that could create an important advantage.
Suzy — Bringing AI Into Consumer Research
Suzy operates earlier in the marketing process than many of the companies in this article.
The company helps brands understand consumers before large marketing decisions are made.
Headquartered in New York, Suzy has evolved from an on-demand consumer insights platform toward what it describes as a decision engine that connects customer research with business decisions.
The company has also introduced AI-supported research capabilities, including automated synthesis and AI-moderated consumer conversations.
Why AI Could Change Market Research
Traditional consumer research can require a long process.
Teams define a problem, create a research brief, recruit participants, conduct interviews or surveys, analyze the results, prepare presentations, and then turn the findings into decisions.
AI can compress several stages of this process.
It can help structure questions, moderate large numbers of conversations, identify themes, compare customer segments, summarize thousands of responses, and turn raw research into decision-ready information.
However, speed does not automatically create quality.
Poorly designed research remains poor research even when an AI system completes it quickly.
The strongest companies in this category will therefore need reliable methodology, high-quality human data, strong sampling practices, and clear validation processes.
Suzy’s existing consumer research infrastructure gives it an advantage because it can combine AI with an established research system rather than depending entirely on synthetic information.
Nudge — Connecting AI Product Recommendations With Conversion
Nudge is the smallest company in our dataset by disclosed funding, but its strategy is worth watching.
The New York startup announced approximately $1.1 million in pre-seed funding in June 2026.
Its thesis is that measuring AI visibility solves only part of the problem.
Imagine that a customer asks an AI assistant for the best skincare product for sensitive skin. The AI recommends a particular product and the customer follows the recommendation to the brand’s website.
If the customer then lands on a generic product page that ignores the original question, the experience loses context.
Nudge is trying to preserve that context by connecting AI visibility, product information, AI-focused content, merchandising, and prompt-specific shopping experiences.
Why AI Discovery and Conversion May Merge
If AI assistants become important shopping interfaces, marketers will eventually have to optimize two related but separate problems.
The first problem is getting recommended.
The second is converting the recommendation into a purchase.
A company might perform extremely well inside ChatGPT or Gemini yet still lose the customer after the click because the shopping experience fails to match the customer’s original need.
This creates a new space between generative engine optimization and conversion-rate optimization.
Nudge is betting that this space can become a meaningful software category.
Original Finding #4: The Biggest AI Opportunity Is Decision Automation
One of the clearest findings from our analysis is that New York’s AI marketing ecosystem is not primarily centered on content generation.
Most of the companies are trying to improve decisions.
Primary Workflow Across Our 14-Company Dataset
| Primary AI Workflow | Companies |
| Customer and lifecycle decisioning | 3 |
| AI discovery and search | 3 |
| Creative generation and orchestration | 3 |
| Media buying and optimization | 2 |
| Consumer research | 1 |
| Attention and media measurement | 1 |
| Transaction monetization | 1 |
The exact classification is our own, and several companies operate across more than one category. However, the broader pattern is clear.
The first mainstream wave of generative AI made marketers think about producing text, images, and videos faster.
The next wave is about deciding what should happen.
AI increasingly helps determine which customer should see an advertisement, what that person should see, when the message should arrive, whether the company should bid for the impression, how much it should spend, which customer is likely to leave, which product should be recommended, which media placement is worth buying, and which creative asset should be generated.
These decisions happen repeatedly across large marketing organizations.
Automating even a small improvement can therefore create significant financial value.
Original Finding #5: AI Is Breaking Apart the Traditional Marketing Funnel
Traditional marketing diagrams usually show a simple path from awareness to consideration, conversion, and retention.
Real customers have never behaved quite that neatly, and AI makes the model even less accurate.
An AI assistant can create awareness, explain a category, compare companies, summarize reviews, answer objections, recommend a product, and help the consumer make a decision during one conversation.
Several stages of the traditional funnel can therefore happen inside a single interface.
This is why companies such as Profound, Evertune, and Nudge matter.
It also explains why platforms such as Rokt, Attentive, Wunderkind, and Black Crow AI are expanding their ability to make decisions closer to individual customers.
The marketing system is moving away from separate channels and toward connected decisions.
For agencies and large marketing departments, this creates an organizational problem.
The SEO team, paid media team, ecommerce group, customer relationship team, creative department, public relations team, and analytics function may all influence the same AI-generated customer journey.
AI does not care how the company divided those departments internally.
Organizations that keep each function completely separate may therefore struggle to respond quickly.
What Madison Avenue Agencies Should Learn From These Companies
Production Alone Is Becoming a Weak Competitive Advantage
Advertising agencies have historically generated significant revenue from producing and adapting marketing assets.
AI will not eliminate the need for strong creative thinking, but it will reduce the value of many routine production tasks.
Banner variations, social cutdowns, basic product descriptions, translations, simple email rewrites, image resizing, and some types of UGC-style video can increasingly be produced automatically.
Agencies that depend heavily on charging clients for the hours required to create these variations will face growing pressure.
The more defensible work moves toward strategy, taste, insight, experimentation, measurement, customer understanding, original creative ideas, and the ability to connect marketing activity with business results.
The ability to make more content becomes less valuable when almost everyone can make more content.
The ability to know what should be created becomes more valuable.
Agencies Need an AI Discovery Practice
A few years ago, asking what ChatGPT thought about a brand might have sounded like a novelty.
It is increasingly becoming a real marketing question.
Agencies should begin identifying the customer questions that matter most to their clients and tracking how major AI systems answer those questions.
The goal should not be to monitor thousands of random prompts.
A hotel group might start with questions about family travel, luxury stays, business travel, specific neighborhoods, amenities, and price comparisons.
A software company might focus on product alternatives, industry use cases, integrations, implementation problems, pricing, and competitor comparisons.
Teams can then examine which brands appear, which companies are omitted, which external sources influence the answer, and whether the AI describes the client’s product accurately.
This should not become another vanity metric.
The purpose is to understand whether AI systems are beginning to influence real purchase decisions.
What Brands Should Test First
Companies do not need to replace their entire marketing stack to begin using AI effectively.
A small controlled experiment is usually more valuable than a large transformation program built around an unproven tool.
| Business Problem | AI Category to Test | Main Measurement |
| Brand rarely appears in AI recommendations | AI discovery and GEO | Qualified AI visibility |
| Creative production is slow | Generative creative | Cost and time per tested concept |
| Paid media efficiency is falling | AI media decisioning | Incremental CPA or ROAS |
| Website traffic does not convert | Predictive personalization | Conversion lift |
| Email and SMS performance is flat | Lifecycle AI | Incremental revenue per recipient |
| Media quality is difficult to judge | Attention measurement | Incremental outcome per media dollar |
| Consumer research takes too long | AI research | Decision speed and research validity |
| Transaction monetization is weak | Transaction decisioning | Incremental revenue per transaction |
The word “incremental” should be taken seriously.
A vendor can show that customers exposed to its technology spent more money. That does not automatically prove the technology caused the increase.
Better customers may have been more likely to receive the experience in the first place.
Marketers should therefore ask whether a product creates additional value compared with what would have happened without it.
A Practical 90-Day AI Marketing Plan for a New York Company
Days 1–30: Map the Marketing Decisions Before Buying Tools
Do not begin by asking employees which AI products they want to try.
Instead, identify the expensive decisions your marketing team makes repeatedly.
A company may decide which audience should receive an advertisement thousands of times every day. It may decide which customers should receive discounts, which products should be recommended, which campaigns should stop, which leads deserve additional budget, which messages should be sent, or which media placements are worth purchasing.
Estimate the economic value of improving those decisions even slightly.
This exercise usually reveals better AI use cases than starting with a list of popular tools.
The highest-value opportunities often involve decisions that happen frequently and influence significant amounts of revenue or spending.
Days 31–60: Run Controlled Experiments
Choose two or three use cases with measurable outcomes.
If you test AI-generated creative, compare the assets with human-created advertisements under similar media conditions.
If you test personalization, maintain a control group.
If you test automated bidding, compare incremental performance instead of relying entirely on conversions reported by the advertising platform.
If you test AI-assisted consumer research, compare the results with established research methods before using the findings for important strategic decisions.
The goal is not to prove that AI works.
The goal is to determine whether a particular system creates enough additional value for your business to justify adoption.
Days 61–90: Turn Successful Pilots Into Repeatable Workflows
Many AI projects fail after producing an impressive demonstration.
The tool generates useful output, but the output never becomes part of normal operations.
Successful pilots need connections with the systems people already use.
Customer data should reach the AI system when appropriate. Approved outputs should flow into the correct marketing channels. Performance data should return to the platform. Employees should understand when human review is required, and managers should know who is responsible when something goes wrong.
Without this operating structure, the AI tool remains an interesting side project.
With the right structure, it can become part of the company’s marketing infrastructure.
How to Evaluate an AI Marketing Startup
Ask What Creates the Company’s Real Advantage
Almost every marketing software company now says it uses artificial intelligence.
That claim alone is not useful.
Ask what makes the product meaningfully better than using a general-purpose model directly.
The answer might involve proprietary customer behavior data, a unique identity graph, media-performance history, attention signals, purchase data, a consumer research panel, custom models, a network effect, specialized experimentation data, or deep integrations with enterprise systems.
If the only advantage is access to the same foundation models available to everyone else, the product may be relatively easy to copy.
Ask Whether the System Recommends or Acts
There is a major difference between AI that produces advice and AI that takes action.
A dashboard that tells a marketer which customers appear likely to leave still creates work for the marketing team.
A system that identifies those customers, chooses the correct message, sends it through the right channel, measures the result, and adjusts its strategy automatically removes much more work.
The second system can create greater value, but it also introduces more risk.
Companies therefore need approval controls, audit trails, clear permissions, rollback procedures, and limits on what autonomous systems are allowed to do.
AI autonomy should expand only after the system has repeatedly demonstrated that it can operate safely.
Demand a Strong Measurement Method
Marketing AI is particularly vulnerable to impressive numbers that do not prove real value.
A company might claim that AI-generated messages produced millions of dollars in revenue.
That number tells you very little without knowing what those customers would have spent anyway.
Ask whether the vendor supports holdout groups.
Ask how incremental lift is calculated.
Ask whether results can be verified using your own analytics environment.
Ask how seasonality, discounting, customer selection, and media spending are controlled.
A strong vendor should be comfortable answering these questions.
What Could Go Wrong With AI Advertising
Cheap Content Could Produce Large Amounts of Mediocre Advertising
If every brand can create hundreds of advertisements almost instantly, consumers will see far more content.
That does not mean the content will become better.
AI can easily create a huge volume of technically acceptable but forgettable advertising.
This makes strong creative direction more important rather than less important.
When generation becomes inexpensive, judgment becomes scarce.
Brands still need people who understand culture, emotion, customer needs, timing, and what makes an idea distinctive.
Advertising Algorithms Could Become Too Similar
Another risk is convergence.
Thousands of advertisers may eventually use the same foundation models, campaign platforms, bidding systems, audience tools, and optimization goals.
When everyone uses similar technology, companies may begin making similar decisions.
Competitive advantage then moves elsewhere.
Proprietary data becomes more important. Strong brands become more important. Exclusive customer relationships become more important. Original creative strategy becomes more important. Custom algorithms and high-quality first-party information become more important.
The more standardized AI becomes, the more valuable genuinely unique inputs may become.
Brand Safety Will Expand Beyond Content
Traditional brand safety usually asks whether an advertisement appeared next to inappropriate content.
Agentic marketing creates many additional questions.
Should the AI have sent the message?
Was the discount appropriate?
Was the customer data used correctly?
Did the system describe the product accurately?
Was a legal claim approved?
Did the company accidentally treat one customer group differently from another?
Was an AI shopping assistant given accurate product information?
Marketing governance will therefore need to cover automated decisions as well as generated content.
Consolidation Has Already Started
The independent company landscape is likely to keep changing.
Bluecore was acquired in 2026. Simon AI was also acquired during 2026. Movable Ink moved into a new ownership structure before that.
These transactions point to an important trend.
Large marketing platforms want to control more of the AI marketing stack.
One possible future is dominated by broad platforms that combine customer data, identity, messaging, personalization, decisioning, analytics, content generation, and AI agents in one system.
The other future favors specialists.
A smaller company becomes exceptionally good at one difficult problem such as AI discovery, attention measurement, customer prediction, media optimization, transaction relevance, or creative intelligence and then integrates with larger platforms.
Both models can succeed.
The most difficult position may be the middle, where a company is not broad enough to become infrastructure but also not specialized enough to remain essential.
What Our Research Suggests About the Future of Madison Avenue
The phrase “Madison Avenue” once referred mainly to advertising agencies.
The next version of Madison Avenue may describe a much larger ecosystem that includes agencies, AI software companies, retailers, publishers, ecommerce platforms, consumer-data companies, model providers, creators, media networks, and autonomous software agents.
New York has a credible position in that future because the city already sits at the intersection of many of these industries.
The concentration of marketing and advertising talent supports that argument.
Our startup dataset supports it as well.
The companies we examined operate across nearly every major stage of the marketing process, beginning with consumer research and continuing through discovery, creative production, media buying, personalization, conversion, transaction monetization, and customer retention.
The market is also supported by more than one generation of companies.
Persado, Rokt, Cognitiv, Clinch, Attentive, Wunderkind, and Suzy show that New York had significant data-driven marketing businesses before the current generative AI boom.
Profound, Evertune, Mirage, Black Crow AI, Chalice AI, and Nudge represent a newer generation built around customer behaviors that have changed substantially since 2020.
That combination gives the city both existing marketing infrastructure and new AI-native experimentation.
The Biggest Opportunity May Be the Marketing Decision Layer
It is easy to misunderstand the AI advertising boom by focusing entirely on content generation.
AI writes headlines, creates images, produces videos, and generates email copy. These capabilities matter, but they represent only one part of the marketing system.
The larger economic opportunity may sit above the content.
AI can decide whether a message should be created in the first place.
It can select the customer.
It can choose the channel.
It can determine the offer.
It can decide whether an advertising impression is worth buying.
It can adjust the bid.
It can identify changes in consumer behavior.
It can measure whether media deserves additional spending.
It can choose which product fits a customer request.
It can monitor how AI systems describe a brand.
It can generate a creative asset only when performance data suggests that the asset is actually needed.
This is much bigger than content automation.
It is the beginning of marketing operations becoming computational.
Final Thoughts: Madison Avenue Is Being Rebuilt Rather Than Replaced
Artificial intelligence is unlikely to eliminate marketing, but it will change where human judgment creates the most value.
Routine work surrounding marketing will continue to become cheaper.
Creating variations, summarizing research, adjusting bids, adapting content, identifying likely customers, monitoring channels, personalizing messages, and producing basic campaign assets will increasingly be automated.
As those tasks become easier, the remaining human decisions become more important.
Companies still need to decide what their brands stand for. They need to understand why customers care, identify which problems are worth solving, create ideas people remember, make difficult strategic trade-offs, and recognize situations where an algorithmically efficient recommendation is strategically wrong.
The New York companies covered in this article are building the systems that could automate much of the work surrounding those decisions.
Some of these companies will become much larger. Others may be acquired. Certain categories will merge into broader software platforms, while others may disappear entirely as individual features become commoditized.
However, the direction of the market is already clear.
The next generation of important Madison Avenue companies may not resemble traditional advertising agencies.
They may look like AI search intelligence platforms, customer prediction systems, automated media buyers, consumer research engines, creative orchestration software, attention measurement platforms, transaction decision systems, and autonomous agents making millions of small marketing choices every day.
For New York businesses, the opportunity is not simply to adopt AI because competitors are doing it.
The real opportunity is to identify the marketing decisions where better intelligence can create measurable economic value, test those decisions carefully, and build an operating system that allows both people and machines to make them better.
