Marie Robin develops Fleet Forward, which helps creative agencies put AI to work through training, workflow design, and systems that combine agency knowledge with human judgment.
They are most useful for helping creative agencies adopt AI through training, workflow design, and systems.
Best for
A B2B AI consultancy targeting French creative, social and PR agency leaders.
Not for
A US developer-tooling vendor wanting English sponsored placements with guaranteed impressions and segmented demographics.
Key takeaways
An agency's advantage is an intelligence factory, not a content factory.
Agentic AI needs a framework and a person who decides.
Build on method, not on a model you do not own.
Check first: Confirm follower demographics, agency-role breakdown, and whether they offer consulting rather than sponsorship.
Main ideas
An agency's advantage is an intelligence factory, not a content factory · Agentic AI needs a framework and a person who decides · Build on method, not on a model you do not own
Audience
12,000+ LinkedIn followers (self-reported) · 450,000+ LinkedIn impressions over the year (self-reported)
Marie Robin develops Fleet Forward, which helps creative agencies adopt AI through training, workflow design, and systems built around each agency’s own knowledge.
Their LinkedIn work connects practical AI implementation with the business questions agencies face: how to protect quality, stand apart from competitors, and prove value.
Robin also says they are writing a book; ask for its title and subject to understand how it fits into their broader work.
Robin’s distinctive contribution is to frame AI adoption as an organizational and strategic challenge, not just a choice of tools.
They argue that agencies should make their methods, client history, and past results usable in AI systems while keeping people responsible for important decisions.
Their examples make the risks concrete: in one post, uncontrolled agent workflows lead to duplicated work, wasted tokens, or security problems; in another, an AI-generated imitation of a creator’s visual illustrates reputational and legal concerns.
Robin also connects these operational issues to market positioning, contrasting mass-produced content with systems that preserve an agency’s expertise.
Across these posts, they combine practical warnings, named industry examples, and a clear point of view.
Their work is useful to agency leaders and marketers weighing how AI could change their operations and offers.
Marie Robin works on how creative and independent agencies should adopt AI.
Read more
Their recurring argument is that the technology pays off only when wrapped in an agency's own method and judgement, rather than used as a volume machine.
They develop Fleet Forward, where the work runs from training and diagnostics to installing what they call an agencyOS — agentic setups with a person deciding at each key step — and an 'intelligence factory' that turns an agency's ideas, client history and past performance into something people and agents can reuse.
They also track the wider market: French social and influence launches, generative-search visibility data, Adobe's automation pivot, and agencies that take equity, drop timesheets or refuse free pitches.
Two habits run through their posts: they argue by example, often starting with a failure or analogy before stating the rule, and return to the same themes in different formats — a reflective post, an event presentation, or a written answer to a question they say they keep being asked.
An agency's advantage is an intelligence factory, not a content factory
Marie Robin argues that agencies competing by automating content production end up selling the same output as everyone else using the same models, and competing on price.
Read more
Their alternative is what they call an intelligence factory: an agency's own ideas, client history, sector expertise, past performance and production practices, made reusable by human teams and AI agents.
They say this distinguishes agencies using identical models, creates an asset that can raise the agency's value and opens service offers that are not billed by the hour.
They are explicit that it takes sustained work — challenging the status quo, testing, anchoring, pivoting — and that building a second brain for an organisation is harder than building one for an individual.
In a post answering the question 'content factory or not', Marie Robin contrasts the two models and cites Ryder & Davis on a tenfold EBITDA effect for agencies that build a transferable system. In a separate post presenting the toolkit behind agencyOS, they say every agency using the same models and training data would deliver the same results without its own self-learning layer.
Marie Robin concedes that content factories meet real short-term demand, while few producers can move fast and at volume.
Large holding companies can pursue high-volume, low-cost production without the same hit to profitability.
They say watching teams at large groups ignore the AI tools they are given convinced them that value lies in use cases, not tool rollout.
Agentic AI needs a framework and a person who decides
Marie Robin's position is that agentic AI is a real capability, but without a framework it produces chaos; human involvement is for arbitration, not rubber-stamping.
Read more
Because AI says yes to any request, they write, agencies need people who can judge and decide.
They warn that if AI makes the decisions while teams become pass-throughs, the agency stops being essential to the client.
Most AI charters state 'human in the loop' as a principle, they note, but a team with no room to think will validate whatever arrives without reading it.
Marie Robin retells Disney's Sorcerer's Apprentice as a fable about agentic AI, with three agency failure modes: an art director running an AI film workflow without knowing the optimisation settings and running up an enormous token bill; a project manager connecting a model to CRM and mail 'to test', then sending 2,000 phishing mails after a prompt injection; and three code agents running in parallel on one brief, duplicating work and delivering incoherent versions. In a later post, they describe building an agency's second brain as a puzzle assembled from the corners: make the agency's history readable, decide what the system retains, and fix who validates what.
Marie Robin argues that an agency building its AI promise on a particular model or vendor is exposed when the rules change, because access restrictions and model changes are outside its control.
Read more
For most agency work — briefs, benchmarks, editorial planning, research, asset declinations — available models are good enough, they say, so the real risk is basing your value on a tool rather than a method.
They describe protecting the agency by making the provider replaceable, rather than betting on the right one.
The durable asset, they say, is the documented prompts, methods and processes nobody can take away.
Marie Robin was interviewed by a BDM journalist after Fable was cut off for three weeks and a new model was reserved to a handful of users.
They used the interview to lay out three reflexes: build model-agnostic systems, connect them to an intelligence factory of documented methods, and keep the right human in the loop at each key step.
Fleet Forward, they say, helps agencies move from vendor dependence to systems they own.
Proof, not engagement, is what social and influence are sold on now
Marie Robin reads several French market launches as part of one shift: social and influence are no longer sold on engagement metrics but on demonstrated business impact.
Read more
Agencies built on the creator economy now face competitors with stronger measurement, CRM and consumer data, they conclude; the risk is losing on the idea while still selling time when the client wants to buy results. They also argue that generative search changes the pitch: if brands cited in AI answers convert better than search traffic, the skills PR teams already have — depth, authority and presence in the third-party sources models cite — are exactly what those brands need, provided the agency reframes how it sells them.
Marie Robin analyses three launches: HEROIKS' Saucy, promising full budget transparency and real-time data; Hopscotch's heaven with Vertical, for continuous vertical social production; and Publicis' Influential France, using Epsilon econometrics across 18,000 influencers. In a separate post, they use Similarweb's first GenAI Brand Visibility Index — ChatGPT-referred visitors converting at 7% versus 5% from Google, 11.4% versus 5.3% in e-commerce, and 1.4 external links clicked per visit versus 0.6 — to argue that brand size does not guarantee AI visibility.
Marie Robin notes that many PR agencies still sell their value in press coverage and advertising-equivalent metrics, while SEO agencies appear to be picking up the subject faster. They frame the opportunity as a service repositioning that Fleet Forward supports, rather than a tool sale.
Pirate agencies show independents what is worth testing
Rather than treating unconventional agencies as marketing postures, Marie Robin studies what each type actually breaks with and presents them as experiments independent agencies can copy.
Read more
Independence gives these agencies a freedom that holding groups lack, they argue: the liberty to break something to test where the market moves. They sort the agencies by what they reject — prudence, timesheets, the fee model, the bespoke scope, the free pitch, human execution — and note that they do not all hack the same thing.
Marie Robin's taxonomy covers cultural insurgents such as Mischief @ No Fixed Address and RIOT NYC, which select clients on appetite for risk rather than budget; operational hackers such as Anomaly, which drop timesheets and price on impact; commercial pirates such as Red Antler and Pattern Brands, which take equity in clients.
They note that Red Antler held stakes in Casper and Allbirds, and that Gin Lane went further by closing the agency to run a portfolio of DTC brands.
The taxonomy also includes subscription players Designjoy and Superside at around $5K a month for unlimited requests; ethical rebels PORTO ROCHA and Zulu Alpha Kilo, which refuse free pitches; and AI-native agencies Sequencr AI and Designity, delivering a model trained on the brand instead of a campaign.
Marie Robin argues that agencies sending thousands of AI-generated prospecting messages do not get the briefs they want.
Read more
Established agencies get meetings but little work follows; newer agencies are left on read because everyone is prospecting the same way, and dependence on pitches stays high as response deadlines shrink. Their alternative growth levers are a value proposition backed by proof, founder market fit and personal branding, communication focused on the client's problems rather than the agency's, CRM discipline with market-signal monitoring instead of cold outreach, and a client experience good enough to produce testimonials.
Marie Robin lists specific levers with named examples, pointing to Melissa Simoni and Jérémy Lacoste as agencies doing founder market fit well.
Agencies with a 5,000-email newsletter sending budget wins, or a LinkedIn page updated every three months, are not doing the communication that works, they note.
On operations, they list the recurring tasks Fleet Forward sets out to remove: monitoring, audits, meetings, benchmarks, project framing, follow-ups, budgets, recommendation and reporting slides, data extraction and compilation, CRM updates, quote requests, feedback collection, versioning and archiving.
Responsible AI use is a governance problem, and agencies are taking shortcuts
Marie Robin treats generative AI ethics as an operational question rather than a policy statement.
Read more
The grey zone around intellectual property and AI has narrowed enough, they argue, that agencies can now be expected to set a framework — flexible but safe — and educate staff, because the shortcuts are tempting. They use failure cases to make the case for that framework rather than opening with rules.
Marie Robin writes about Shopify and an agency that reproduced a creator's key visual with AI to illustrate a template.
The creator saw it, triggering a backlash; Robin notes the reputational damage and the unanswered question of compensation.
To give agencies practical answers, they interviewed lawyer Aurore Sauviat and published the conversation as a two-part Substack interview covering AI and intellectual property, the risks agencies run, and how to guard against them.
The legal analysis is Sauviat's, attributed as a guest; Robin frames their own position around governance.
Answers “What does Marie Robin actually build for agencies?” and presents their signature framework in their own words.
Read more
Robin introduces AgencyOS as an operating model for agencies: AI agents for each team, hybrid agentic systems with a human in the loop, and a self-learning ‘intelligence factory’ built from the agency’s ideas, client history and past performance. Without this proprietary layer, Robin argues, every agency using the same models would deliver the same results.
For brands planning a campaign
For an AI consultancy targeting creative agencies, this is the clearest articulation of Robin’s core offer: not just tools, but a system that captures and mobilizes agency-specific knowledge to stand out from generic content factories.
Answers “How does Robin approach the risks of AI agents?” with concrete failure modes and a memorable analogy.
Read more
Using Disney’s Sorcerer’s Apprentice, Robin illustrates common agentic AI failures in agencies: a DA racking up token bills in Weavy, a project manager connecting Claude to CRM and triggering phishing emails via prompt injection, and Code agents producing duplicated, incoherent work. Robin concludes that agentic AI without a framework is chaos.
For brands planning a campaign
Shows Robin brings risk management and governance to AI adoption, not just enthusiasm. A campaign could highlight their ability to prevent costly mistakes.
Answers “Does Robin address AI ethics and legal risk?” with a real case and expert commentary.
Read more
Robin recounts how an agency used generative AI to reproduce a creator’s key visual for a Shopify template, causing a backlash. They note the reputational damage and unresolved legal questions, then interview lawyer Aurore Sauviat on AI and intellectual property.
For brands planning a campaign
Demonstrates Robin can advise agencies on ethical and legal AI use, an important trust factor for clients in regulated or reputation-sensitive sectors.
Answers “What is the scope of Robin’s work, and who do they work with?” in a single roundup.
Read more
Robin lists 2025: 474 online meetings, 50+ agency leaders interviewed on AI transformation, two webinars and a Paris event with VML, BUZZMAN and Opella, training at EssilorLuxottica, and work with agencies like Zcomme and BUZZMAN. They also launched a newsletter, passed 1,200 Substack subscribers and 12,000 LinkedIn followers, and are writing a book.
For brands planning a campaign
Shows Robin’s scale, client list and thought leadership—useful for assessing their network and credibility for a campaign targeting agency leaders.
Answers “How does Robin see the social and influence agency market changing?” with named examples and a clear thesis.
Read more
Robin analyzes three French launches: HEROIKS’ Saucy (budget transparency), heaven/Hopscotch’s Vertical (performance-driven social production), and Publicis’ Influential France (econometric measurement with Epsilon). Robin argues social and influence are now sold on proof, not engagement, and warns agencies that built on the creator economy.
For brands planning a campaign
Shows Robin tracks competitive moves and can help agencies reposition around measurement and business impact, a key concern for agencies facing procurement pressure.
Answers “What is Robin’s alternative to mass AI content production?” and shows how they frame agency differentiation and valuation.
Read more
Robin contrasts the ‘content factory’ model (high volume, low cost) with Fleet Forward’s ‘intelligence factory’—a system that makes an agency’s methods and best talent mobilizable, building a transferable asset. They cite Ryder & Davis on x10 EBITDA potential and argue this protects against commoditization.
For brands planning a campaign
Gives a concrete value proposition for agencies worried about AI-driven price competition. Useful for positioning Robin as a strategic partner, not just a tool provider.
Answers 'How does Robin connect AI, PR, and business outcomes?' with data and a specific agency opportunity.
Read more
Robin cites Similarweb's GenAI Brand Visibility Index: ChatGPT-referred visitors convert at 7% vs.
5% from Google, and 11.4% vs.
5.3% in e-commerce.
They argue that brand size doesn't guarantee AI visibility and that PR agencies—with their earned media and authority-building skills—are well placed to optimize for LLM citations.
For brands planning a campaign
Positions Robin as a strategic advisor for PR agencies looking to reposition around AI visibility, a service line Fleet Forward supports.
AI decision-making?' and shows the philosophy behind their systems.
Robin argues that because AI can say yes to any request, agencies need human experts to decide and arbitrate.
They warn that if AI makes the decisions and teams become 'pass-throughs', agencies lose their essential role.
They describe building a second brain for an organization as harder than building one for an individual.
For brands planning a campaign
Reinforces their emphasis on human judgment and organizational design in AI adoption, relevant for clients concerned about quality and differentiation.
Intelligence factory versus automated content factories4
Governance and responsible AI use3
Agency business models, from pirate agencies to client equity3
AI search visibility and how brands get recommended2
Agency growth, prospecting, and daily operational friction2
The French social and influence market1
AI adoption inside creative agencies
Robin helps creative and independent agencies put AI to work without losing human judgement.
Read more
They describe training, diagnostics, coaching for leaders and creatives, and installing an agency operating system they call agencyOS — agentic setups with a person in the loop at each key step.
Their own 2025 recap says they interviewed more than 50 agency leaders about their AI transformation and worked with named agencies on live programmes.
Their early-autumn 2026 plan lists agentic AI training, an AI day for an independent agency, opportunity diagnostics for a startup and a training body, content-engine rebuilds, and executive coaching.
Robin starts with a concrete agency task — briefs, benchmarks, editorial planning, project framing, asset declinations — then shows where AI fits and where a person still has to decide.
For brands planning a campaign
A brand selling AI tooling, training, or services to creative agencies can work with Robin on explainer content or a co-hosted session, reaching agency owners, managers, and creatives rather than general marketers.
Intelligence factory versus automated content factories
Robin argues that agencies that only automate content production compete on price and get commoditised.
Read more
Their alternative is an intelligence factory: the agency's own ideas, client history, sector know-how, past performance, and production practices made reusable by agents.
They say this stops agencies from delivering the same output as everyone else using the same models, gives them an asset they claim can multiply agency value, and opens service offers that are not sold by the hour.
They cite Ryder & Davis on a tenfold EBITDA effect and contrast their approach with content factories that undercut prices.
Robin returns to this contrast in different formats: a reflective post, an event presentation of agencyOS, and a written answer to the recurring 'content factory or not' question.
Robin acknowledges real short-term demand for content factories and notes that large holding companies can pursue high-volume, low-cost production without the same damage to their profitability.
For brands planning a campaign
A software or platform brand can brief Robin on content about knowledge systems, retrieval, or agent orchestration for agencies. They stress method and ownership over any single tool.
Robin's governance content uses failure stories rather than policy language.
Read more
In a piece built on Disney's Sorcerer's Apprentice, they describe an art director burning through tokens on an AI film workflow, a project manager connecting a model to CRM and mail and sending 2,000 phishing mails after a prompt injection, and a contract-checking assistant that only read fragments of a reference document far larger than its context window.
They conclude that agentic AI needs a framework and a method, not just access to tools.
They also covered an agency that reproduced a creator's key visual with AI for Shopify, interviewing lawyer Aurore Sauviat on image rights, ethics, and generative AI and publishing it as a two-part Substack interview.
Analogy and anecdote first, then the operational rule: a recurring structure across Robin's posts.
For brands planning a campaign
Legal, insurance, compliance, or AI-governance brands can reach an agency audience through Robin, for example by supporting a written guide or live session on responsible AI use in agency production.
Agency business models, from pirate agencies to client equity
Robin studies agencies that break with traditional agency rules and sorts them into types: cultural insurgents such as Mischief and RIOT NYC that select clients on appetite for risk; operational hackers such as Anomaly that drop timesheets and price on value; commercial pirates such as Red Antler and Pattern Brands that take equity in clients; subscription service players such as Designjoy and Superside; ethical rebels such as PORTO ROCHA and Zulu Alpha Kilo that refuse free pitches; and AI-native agencies such as Sequencr AI and Designity that deliver a trained model rather than a campaign.
Read more
They present these as experiments independent agencies can copy, while noting that holding groups have less freedom to test them.
A numbered taxonomy with named agencies and a closing question to readers — a format Robin uses for several topics.
For brands planning a campaign
A brand seeking association with agency reinvention and new commercial models can brief Robin on the economics of agency work; the content will foreground named competitors rather than the sponsor.
AI search visibility and how brands get recommended
Robin uses Similarweb's first GenAI Brand Visibility Index to argue that brands appearing in ChatGPT answers convert around twice as well as those arriving from Google — 7% versus 5% in general, 11.4% versus 5.3% in e-commerce — and that ChatGPT users click roughly 1.4 external links per visit against 0.6 on Google.
Read more
Brand size, they add, does not guarantee AI visibility. They read this as a repositioning opportunity for PR agencies, arguing that depth of content, authority, and presence in the third-party sources models cite are exactly the skills good PR teams already have.
Robin notes that many PR agencies still sell their value through press coverage and advertising-equivalent metrics, while SEO agencies appear to be taking up the subject faster.
For brands planning a campaign
An SEO, analytics, or AI-visibility tool can commission Robin to explain generative search measurement to agencies. They frame the opportunity as a service repositioning rather than a tooling sale.
Robin tracked three French launches and read them as one shift: HEROIKS with Saucy promising full budget transparency and real-time piloting, Hopscotch's heaven with Vertical for continuous vertical social production, and Publicis with Influential France using Epsilon econometrics across 18,000 influencers.
Read more
Their point is that social and influence are no longer sold on engagement but on proof of business impact, and that agencies built on the creator economy now face competitors with stronger measurement, CRM, and business data. They framed this as the theme of a Reboot evening with Ryder & Davis and agency leaders.
For brands planning a campaign
A measurement, data, or influence platform can work with Robin on content about proving influence impact in France, reaching agency leadership rather than creator talent.
Agency growth, prospecting, and daily operational friction
On growth, Robin argues that mass AI outbound messaging does not work for agencies: established agencies get meetings but few briefs, newer ones get ignored, and proposal responses stay dominant.
Read more
They lay out five levers: a proof-backed value proposition; founder market fit and personal branding; communication focused on the client’s problems; CRM discipline with market-signal monitoring instead of cold outreach; and a client experience good enough to produce testimonials. On operations, they list the tasks Fleet Forward sets out to remove from agencies: monitoring, audits, meetings, benchmarks, project framing, follow-ups, budgets, recommendation and reporting slides, data extraction and compilation, CRM updates, quote requests, feedback collection, versioning, and archiving.
For brands planning a campaign
Agencies or agency-software brands can brief Robin on growth and operations content and expect practical, checklist-style material that names specific tasks and tools rather than making broad AI claims.
Marie Robin writes about how creative and independent agencies should take up AI.
Read more
Start with their recurring “content factory or not” argument, which contrasts automated mass production with what they call an intelligence factory built from an agency's own methods, client history and past performance.
Then read their retelling of The Sorcerer's Apprentice, featuring agency failures — a runaway token bill, a prompt-injected mail blast, three code agents duplicating campaign instructions — to see why they argue human judgement has to stay in the loop.
From there, their work splits two ways.
On the business side, they study agencies that drop timesheets, take equity in clients, refuse free pitches or sell design as a subscription.
On the market side, they track generative search visibility and French social and influence launches, arguing both are now sold on proof of business impact rather than engagement.
Follow them for the agency-economics angle as much as the AI one: they are a working consultant, not a commentator, and their posts keep returning to what an agency can actually charge for.
Both argue for practical agentic AI for a business audience rather than model talk.
Read more
Bidet describes building lightweight agents that remove daily irritants — summarising mail and filing attachments, producing meeting notes with a to-do list, consolidating e-commerce data with a CRM. Both make the point that an agent must fit an existing organisation, its processes and constraints to be useful rather than a risk.
Bidet works with small and mid-sized companies and builds the agents himself.
One of his posts looks further ahead at interfaces disappearing as agents act on a user's behalf.
Robin stays on the agency side, exploring the agency's business model, pricing and how it keeps judgement in the loop rather than the mechanics of a single build.
Both offer practical, diagnostic marketing advice rather than inspiration.
Read more
Duvergé walks through a five-step marketing audit — clarify the value proposition, map the acquisition channels, read real behaviour, find the leaks, then turn the audit into an action plan — and backs a later post with channel-level acquisition cost figures. Robin applies the same approach to agency growth, listing the value proposition, communication, CRM discipline and client experience as the levers that actually bring briefs.
Duvergé addresses owners of small and mid-sized businesses about their own acquisition and the real cost of each channel; Robin addresses agency leaders about positioning, AI adoption and the economics of the agency itself.
Both describe marketing work turning into systems rather than campaigns, and use hands-on examples to show it.
Read more
Benoliel narrates running a first ABM campaign at an AI-native company, where templates, translations and formats were generated from one spreadsheet, freeing time to challenge the messaging.
She later argues that emerging GTM engineer roles exist because content has to be wired into pipeline.
Robin's agencyOS and intelligence factory make the same move for agencies.
Benoliel writes from inside an in-house marketing team at a technology company, focusing on ABM production and turning attention into a contactable list. Robin writes from the agency side, exploring what the agency owns, how it prices and how it keeps the client relationship rather than the pipeline mechanics.
Marie Robin has roughly 13,000 LinkedIn followers (read 7 October 2026).
Read more
Their end-of-2025 review counts more than 12,000 LinkedIn followers, more than 450,000 impressions over the year, and more than 1,200 newsletter subscribers.
Their posts focus on how creative, social and PR agencies put AI to work.
Written in French, most address agency owners and senior leaders directly, with lines such as « si vous dirigez une agence créative » and a closing question to « dirigeants d'agences »—the same people the inquiring consultancy wants as clients.
For a campaign aimed at French-speaking agency decision-makers, Robin offers a route to that audience rather than to enterprise marketing or technical buyers; English campaign material would need adapting or translating.
The posts do not show who actually reads them: there is no breakdown by region, seniority, company size or job function, or indication of how many followers work inside agencies rather than in adjacent roles.
Before committing spend, ask for the platform's follower-demographics breakdown, any reader survey or client list covering agency roles and countries, and recent results for agency-AI posts comparable in depth and format to the planned campaign.
LinkedIn followers (self-reported)
12,000+
Marie Robin's year-in-review post on LinkedIn, 31 December 2025 · 2025-12-31
LinkedIn impressions over the year (self-reported)
450,000+
Marie Robin's year-in-review post on LinkedIn, 31 December 2025 · 2025-12-31
Newsletter subscribers, Substack (self-reported)
1,200+
Marie Robin's year-in-review post on LinkedIn, 31 December 2025 · 2025-12-31
Marie Robin runs Fleet Forward, which sells AI adoption work to creative agencies and the brands that buy from them.
Read more
They describe paid engagements rather than a sponsorship inventory and do not publish a media kit, rate card, or sponsorship policy.
Their posts point to consulting, not sponsored placements on their channels.
Buyers should plan for a consulting relationship with defined entry points, not a creator sponsorship.
Work starts through direct contact: Robin's posts repeatedly close with an invitation to send a LinkedIn DM.
No inquiry form or partnership page is published.
Rates, exclusivity terms, and standard engagement lengths are not disclosed.
Robin describes a waiting list for diagnostics, so buyers should expect to book weeks ahead.
They frame their differentiator as an "intelligence factory": the agency's own methods, client history, and sector expertise made usable by AI agents, with senior people kept in the loop rather than replaced.
AI training for agency teams
Buyer requirement
A named group of creatives, managers, or commercial staff who will attend and use the tools in their work.
Training covers ChatGPT, Claude, Gemini, Perplexity, and NotebookLM.
Robin notes that even large groups with paid tool access may not use the tools unless someone brings concrete use cases and follows up.
Plan for internal reinforcement after the session, not a one-off course.
AI opportunity diagnostic
Buyer requirement
A defined agency or business unit willing to open its current workflows and tooling to outside review, plus a decision-maker who can act on the findings.
Capacity is limited: the August 2026 post describes September as fully booked, with two agencies waiting for October.
Build several weeks of lead time into the plan.
AgencyOS and agentic system setup
Buyer requirement
Access to the agency's client history, sector expertise, past performance data, and production practice, because the system is trained on that material.
Setup also requires the agency to decide what the system retains and who approves which outputs.
Robin describes this scoping as the first step, not an afterthought.
The result is model-agnostic, so the agency is not locked to one AI supplier.
Intelligence factory or content engine build
Buyer requirement
Documented methods, prompts, and processes that people and AI agents can retrieve, plus senior staff involved in orientation and final decisions.
Robin positions this against fully automated content factories and treats it as long-term work spanning multiple quarters, not a tool purchase.
They cite the 2026 reboot session where Ryder & Davis put the valuation effect at 10x EBITDA.
Leadership coaching and strategic advisory
Buyer requirement
Directors or creative leads who will attend coaching sessions, or, for group-level work, an executive sponsor able to commission a communication and AI strategy across brands.
The 2025 recap lists coaching for agency directors and creatives, and group-level strategy and content engine work for three large organizations.
Expect multi-stakeholder sign-off.
Co-hosted events, webinars, and interviews
Buyer requirement
A topic at the intersection of AI and agency operations, and an audience of agency leaders.
Robin organises webinars and Paris events.
The June 2026 post names an evening with Steve, Jellyfish, Brainsonic, and CEETADEL, so co-hosts need to bring named speakers or a client-side audience, not just funding.
A B2B AI consultancy that already serves technology clients wants to move into French creative, social and PR agencies.
Read more
It needs a credible route to the people who approve training budgets and new operating models: agency owners and senior leaders.
Marie Robin is a close fit for this campaign.
They write in French for exactly this group, addressing agency leaders directly with openings such as "si vous dirigez une agence créative" and closing questions aimed at "dirigeants d'agences".
Their subject matter overlaps almost completely with what the consultancy sells.
They train agency teams on ChatGPT, Claude, Gemini, Perplexity and NotebookLM; run AI opportunity diagnostics; set up agentic systems they call AgencyOS; build what they call an intelligence factory from an agency's own methods and client history; and coach directors and creative leads.
Their 2025 review lists more than 50 agency-leader interviews on AI transformation, a Paris event with VML, BUZZMAN and Opella, training at EssilorLuxottica, and work with named agencies including Zcomme, BUZZMAN, AUDITOIRE, Agence HKM, 1969, MOUTON NOIR, Just Happiness and Grand-Mercredi.
These are checkable reference points the consultancy can ask about.
The fit rests on specifics, not generic appeal.
Robin reaches the buyer, not the end consumer: their posts assume readers run or work inside an agency and are weighing tool adoption, pricing models and client retention.
Their content is practical enough to borrow.
The list of recurring agency tasks they propose removing—benchmarks, project framing, recommendation and reporting slides, CRM updates, feedback collection, versioning and archiving—is the sort of worksheet a consultancy can hand a prospect.
They also address risk, with failure stories about runaway token spend, prompt injection sending phishing mail from a CRM, and a contract-checking assistant reading only fragments of an oversized document.
A two-part interview with lawyer Aurore Sauviat covers AI image rights after an agency reproduced a creator's key visual.
A consultancy selling cautious, governed AI adoption can stand alongside that material without contradicting it.
Before planning spend, the buyer must accept three constraints.
Robin sells consulting engagements, not sponsored placements: their posts contain no media kit, rate card, sponsorship policy or inquiry form, and the only documented contact route is a LinkedIn direct message.
Expect a paid engagement or co-hosted session, not a bought post.
Capacity is constrained: an August 2026 post says September was fully booked, with two agencies waiting for October, so build in weeks of lead time.
Language matters too.
Their posts are in French for a French-speaking agency audience, so English campaign copy must be rewritten or translated.
The audience is modest—roughly 13,000 LinkedIn followers and 1,200 newsletter subscribers, the latter self-reported at the end of 2025—and its composition is unverified.
Judge the campaign by conversations and briefs generated, not reach; ask Robin for audience data before committing if composition is a buying criterion.
When to choose a specialist
A US-based developer-tooling vendor wants an English-language sponsored placement with guaranteed impressions, a published rate card, exclusivity terms, and follower demographics broken out by country, seniority and job function.
Read more
This is a poor fit, and the mismatch is documented rather than a matter of taste.
First, the audience cannot be segmented as this campaign requires.
Robin's posts show a following total and self-reported impressions and subscribers, but no breakdown by region, seniority, company size or job function.
A buyer who needs to contract against a defined segment would be buying an unverified assumption.
Second, the language is wrong.
Robin's posts are in French for agency leaders, whom they address directly; an English-language product launch would reach readers who did not come for it and require adaptation the vendor's campaign calendar may not allow.
Third, there is no sponsorship product to buy.
Their posts list no media kit, rate card, exclusivity policy or standard format length, and give no indication that they sell placements on their channels.
Their commercial model is diagnostics, systems setup, training and coaching, alongside co-hosted webinars and Paris evenings.
Fourth, a sponsorship-style campaign would conflict with their position.
Robin argues that agencies should avoid commoditised, high-volume output and build a proprietary intelligence factory instead.
Their writing names their frameworks and competitors such as Red Antler, Anomaly, Designjoy and PORTO ROCHA, leaving little room for a sponsor's product to be the story.
There is also a format clash.
A vendor seeking fast, cheap, high-volume content at scale would be commissioning the exact model Robin argues against: they describe fully automated content factories as a short-term demand they expect to damage the brands that use them.
That does not rule them out as a supplier of agency-facing expertise, but it makes them a poor vehicle for a volume- or placement-led campaign.
The pragmatic alternative is to reframe the brief as a French-language, agency-focused thought-leadership engagement booked weeks ahead, or to look for creators whose published audience data and sponsorship terms match a media-buy brief.
A worked campaign example
A worked angle for the B2B AI consultancy's campaign to agency leaders, built from Robin's posts rather than a new concept: a 90-day French-language engagement delivering a useful artefact, not a campaign burst.
The offer: a two-part collaboration.
Part one is a 45-minute French webinar for agency leaders, framed around Robin's own terms: what an AI operating system is, what to automate and what must stay human.
The session opens with failure modes Robin already tells well: the art director burning tokens on an AI film workflow; the project manager whose CRM connection sent 2,000 phishing mails after a prompt injection; and three code agents duplicating each other's work on one campaign.
It then turns to their intelligence-factory argument: agents are only as distinctive as the agency knowledge they receive, so methods, client history and sector expertise must be made retrievable before tooling is chosen.
Part two is a worksheet agency leaders can complete in about twenty minutes.
It maps the recurring tasks Robin lists as agency irritants—monitoring, audits, project framing, follow-ups, budgets, recommendation and reporting slides, data extraction and compilation, CRM updates, quote requests, feedback collection, versioning and archiving—to a decision about whether a person, an agent or a person-in-the-loop should own each one.
The consultancy brands the worksheet, hosts the download and follows up with respondents.
Sequence and assets.
Weeks 1–2: brief Robin, agree the theme and dates, decide who speaks for the consultancy, and confirm in writing what the consultancy may reuse.
Weeks 3–5: three LinkedIn posts from their account and one newsletter mention promote the webinar, each making a single argument rather than simply announcing the event.
Week 6: run the webinar live with questions and send the worksheet to attendees.
Weeks 7–12: publish one follow-up post summarising what agencies reported, then have the consultancy contact respondents with an offer of a paid diagnostic.
Measurement.
Track registrations with an agency-side job title, completed worksheets, meetings booked and pipeline value.
Impressions and follower counts are context, not the target.
A useful secondary measure is how many attendees describe a specific task they intend to automate; that language can inform a diagnostic engagement.
Why this angle rather than a product pitch?
It draws on Robin's strengths: their signature framework in their own words, concrete failure stories and a checklist they already publish in spirit.
It asks their audience to take action inside their own agency—the action a consultancy needs—and gives the consultancy a qualified list instead of a vanity metric.
Before costing, ask Robin what they accept as a paid engagement versus a co-hosted event, whether they are available given their published backlog, whether the consultancy's name can appear in their follow-up material, and whether they can share audience data to validate the invite list.
Choose Cédric Bidet over Marie Robin when the consultancy’s route into a prospect is hands-on agent building for small businesses, not agency-wide transformation, and when the argument that lands is “less jargon, less magic.”
Read more
The same buyer task—a B2B AI consultancy reaching decision-makers with a governed-adoption pitch—is better served by him if the deliverable is one working automation inside the prospect’s own workflow.
His material speaks to dirigeants who want something they can see working: an email summary that files attachments correctly, a meeting note that produces a personalised to-do list, or an e-commerce extraction joined to the CRM and sales targets.
He sells personally deployed agents from €150 a month with no commitment, which suits a consultancy that wants a builder beside its strategist or a low-risk first step for a hesitant prospect.
The tradeoff is the buyer and the unit of work.
Bidet addresses SME owners with repetitive tasks, not agency leaders deciding how to run a creative business; his posts say nothing about agency pricing, creative teams or client retention.
His monthly product is not an engagement, so a consultancy cannot book him for a co-hosted workshop without changing what he sells.
His posts also show no governance or risk content comparable to Robin’s stories about prompt injection, token spend and AI copyright—a distinction that matters if governed deployment is the consultancy’s differentiator.
Choose Pierre Duvergé when the campaign needs a demand-generation mechanic rather than a transformation narrative, and when the first useful deliverable is a template the prospect can complete.
Read more
For the same task—a consultancy trying to open doors with a diagnosis offer—he supplies the arithmetic that makes diagnosis urgent: B2B acquisition costs up 40 to 60 per cent between 2023 and 2025, cold outbound around €1,820 per lead against €140 for referrals, and CAC assumptions from 2022 likely to be wrong by that same margin.
He also shows the shape of a working lead magnet: a free Google Sheets kit offered in exchange for attention, then left to speak for itself rather than followed by a sales call.
A consultancy selling audits to agencies, or wanting a workbook attached to the pitch, can borrow that structure directly.
The tradeoff is subject and audience.
Duvergé writes about marketing mechanics for SME owners; his two posts do not mention creative agencies, agency economics or AI transformation, so the consultancy must supply the agency context.
His figures appear without cited sources, so re-check them before building a campaign around them.
Where Robin speaks to agency leaders in their own vocabulary, Duvergé addresses them through pipeline arithmetic—a useful but different conversation with a different promise.
Choose Emmanuelle Benoliel when the audience is in-house B2B technology marketing teams rather than agency leaders, and the credible story is an operator running AI-native campaigns from the inside.
Read more
For a consultancy selling AI marketing systems to tech companies, she is the closer peer: her posts describe building an ABM campaign at ElevenLabs with Claude, Clay, a CRM and an automated design engine that produced a full campaign’s assets in hours.
They also trace marketing AI’s shift from side project to dedicated function, citing three emerging roles and 3,000 open GTM Engineer positions for January 2026.
She lays out a workflow clients would recognise—capture interactions, enrich profiles with real data, connect them to the CRM and events, score against the ICP, then hand sales a short, activatable list—turning 300 likes into 15 named people to contact.
The tradeoff is that Robin speaks as an in-house practitioner about their employer’s stack, not as an adviser to agencies.
Their two posts contain no offer, rate card or documented availability for outside work; ask about commercial terms before planning a collaboration.
The consultancy gains an inside-the-tech-company voice but loses the agency-leadership audience and operating-model content.
The role study they report is industry data, not first-party agency evidence, so ask for agency-specific evidence if that is central to the pitch.
Marie Robin's fourteen complete LinkedIn posts published across 2025 and 2026 cover Robin's AI work with agencies, market commentary on social, influence and generative search, and Robin's own accounts of Fleet Forward projects in 2025.
The posts do not provide the contents of Robin's Substack articles, the written interview with a lawyer, newsletter issues, webinar and podcast recordings, or the press article linked in the posts, so anything about those formats comes from Robin's own summaries in the posts.
Robin refers to a book Robin is writing, but no post names it or states its subject, so that detail remains open for anyone searching Robin's name.