According to TechFastForward, Sakana AI launched Marlin on June 15, 2026. What follows is not speculation about AI's impact on consulting. It's what's happening right now.

Marlin runs unattended for eight hours. It produces 100+ page strategy reports, executive dashboards, and board-ready slide decks without human intervention. The first 300 enterprise beta users were financial institutions, operating companies, consulting firms, and think tanks. They didn't request this capability. They adopted it because the output quality demanded they adjust how they staff research and strategy work.

The consultant who survives the next 18 months won't be the one fighting for relevance by adding AI to their toolkit. That person is still selling hours. The one who wins will sell something Marlin cannot: proprietary systems, IP-backed frameworks, and business models that exist independent of human labor.

Marlin changes what "deliverable" means

McKinsey QuantumBlack, BCG X, and Bain's analytics division built internal AI tools years ago. These tools make their consultants faster. They don't replace the need for consultants. VentureBeat covered Marlin's launch with a key distinction: "Rather than making a consultant 20% faster (Copilot-style), Marlin's proposition is that the research deliverable does not require a consultant at all."

That's the line. On one side: tools that augment. On the other: systems that substitute.

Marlin operates on pay-as-you-go pricing for corporate entities only. You don't rent an analyst. You submit a research brief, set session parameters, and come back in eight hours to a finished strategy document. The pricing model assumes no human labor in the research-to-output pipeline.

Kevin O'Leary warned about this shift publicly. "Companies first going to AI instead of consultants," he said in June 2026. The premise is simple: if an AI agent can produce the work in eight hours, the question isn't whether companies will use it. The question is what consultants do next.

The bifurcation is real

The market is splitting into two consultant archetypes. The first sells output that looks like what Marlin produces: research reports, competitive analyses, strategic recommendations, market assessments. These consultants are now competing against a machine that doesn't sleep, doesn't demand equity, and costs 80% less per engagement than a three-person team billing at standard rates.

They lose. The math is too brutal.

The second sells systems that companies implement to generate their own output. This consultant owns the IP. They license frameworks. They build proprietary scoring models, vendor assessment matrices, and market-entry playbooks. They train client teams to operate autonomously using these systems. The deliverable isn't a 100-page report. It's a repeatable machine that the client now owns.

This distinction matters because it separates those who build acquirable assets from those who build billable projects. In "The Owner's Exit Engine," the thesis is direct: consultants who own IP create a company. Consultants who sell projects own a job.

Marlin accelerates this timeline. It doesn't give consultants 18 months to transition. It gives them six.

What IP looks like in practice

I'll illustrate from experience. When I left Munich Re as an innovation scout, I didn't take the job title with me. I took the pattern-recognition framework I built. That framework helped identify which insurance products would face regulatory pressure within 36 months. It was specific to Munich Re's business model, but the method was transferable.

In the decade since, that framework has generated more revenue than the salary ever did. It's been licensed by three other insurers. It's been adapted for commercial banking. It's the asset that made the consulting practice valuable, not the hours I billed early on.

Here's what IP looks like for a consultant operating post-Marlin:

A diagnostic tool that identifies which departments a company needs to staff internally versus outsource. This isn't advice. It's a repeatable model with a scoring methodology, benchmark data, and decision logic. The consultant licenses it to 10 clients. Marlin can't produce this because it has no way to protect or monetize proprietary logic.

A vendor assessment framework that evaluates technology partners on dimensions specific to the client's industry and risk profile. Marlin generates comparisons. This consultant's system generates a risk-weighted score that informs board-level capital allocation. The difference is the intellectual property embedded in the scoring criteria.

A market-entry playbook for specific verticals. Not generic entry strategies. Built from 40 case studies in the exact sector the client is entering. Templated. Implemented. Measured against known outcomes. This is a product. Marlin can research the market. It cannot sell a proven playbook that compounds in value with each client implementation.

An internal research capability that the client team runs without the consultant present. It's a platform. It's a process. It's a culture artifact. The consultant builds it. The client owns it. The consultant gets paid for the license, training, and annual updates.

How to build IP before Marlin eats your billable hours

First, stop selling research projects disguised as strategy work. If your engagement delivers a report that could be replaced by eight hours of Marlin runtime, you're already obsolete. You just haven't admitted it yet.

Second, document the pattern-recognition work you do intuitively. The best consultants operate from mental models they've built across dozens of engagements. These models are invisible. Extract them. Systematize them. Build the diagnostic. Make it repeatable. Make it licensable.

Third, create a product that clients can implement without you present. This sounds counterintuitive to consultants trained in billable-hours economics. It's not. A framework that your client can deploy internally, update quarterly, and use across three business units generates more revenue over five years than a traditional three-person project engagement.

Fourth, build benchmarks specific to your vertical. Marlin produces analysis against public data. You produce analysis against proprietary benchmarks that took you 15 years to assemble. That's defensible IP. License it. Monetize it directly rather than embedding it in hourly consulting work.

Fifth, think of yourself as a systems vendor, not a strategy firm. This reframes how you price, how you staff, and how you measure success. You're not optimizing for utilization. You're optimizing for adoption. You're not maximizing hours per engagement. You're maximizing the number of clients running your system and the frequency with which they pay to update it.

The internal link to discovery

Many consultants struggle with the handoff between initial engagement and system implementation. AI scope documents, discovery calls, and signed SOWs happen in 24 hours. Speed matters here because once a prospect is qualified, they need to see your system in action. Your IP is what separates a slow, consultative sales process from a fast, product-oriented one.

Similarly, the ability to score prospective clients using predictive CLV models determines whether you chase low-yield projects or build a practice around clients who can sustain a system license. This is how you move upstream from hours to IP.

And when you measure success, don't use revenue attribution frameworks that conflate consulting vanity metrics with actual business outcomes. IP-backed consulting is measured in adoption, retention, and expansion revenue. Those are the metrics that matter.

Competence beats credentials

Marlin's capability doesn't exist in the consulting industry's prestige hierarchy. No MBA program teaches this. No Big Three firm has publicly documented their response strategy. The advantage goes to the operator who sees the capability before it becomes common knowledge and organizes their practice around defensible IP rather than defensive positioning.

This is not a prediction about AI. This is what's happening now with 300 enterprise users. The window to reposition is closing. Within 12 months, every strategy firm will have made a choice: compete with the machine on its terms, or build something the machine cannot. The latter requires starting now.

FAQ

Q: Isn't Marlin just a research tool? Can't a consultant still add value on top of it?

A: Yes, but that value is increasingly thin. If a consultant's job is to interpret Marlin output, they're optimizing for a supporting role in a client's research process. The valuable position is owning the questions Marlin answers and the systems that make those answers actionable. Those require IP, not interpretation.

Q: How long does it actually take to build proprietary IP as a consultant?

A: If you're documenting and systematizing work you're already doing, three to six months for a basic framework. If you're building from scratch, 12-18 months to a licensable product. The clock started when Marlin launched. Start now.

Q: Should consultants use Marlin internally to make their teams faster?

A: Yes, if your business model is still hours-based and you need a transition period. No, if you're serious about moving to IP-backed practice. Using Marlin to deliver consulting work at lower cost is a race to the bottom. It accelerates the commodification of your current service offering.

Q: What happens to junior consultants in a post-Marlin firm?

A: The ones who survive are builders, not process executors. They work on IP development, client implementation, and system refinement. The ones who disappear are the ones doing the research work that Marlin was built to automate. This is a filter. It sorts for operating talent.

Q: If every consultant pivots to IP, won't the market get oversaturated with frameworks?

A: Absolutely. The differentiation won't be in having a framework. It will be in the specificity of the framework to a particular vertical, the robustness of the benchmark data, and the adoption rate among qualified clients. This is where competence beats credentials. The operator who builds a narrow, deep system for a specific market beats the generalist who tries to license a broad framework to everyone.