The Engine Log
We publish the actual work.
Operational artifacts. Real protocols, real numbers, downloadable templates. The same playbooks we run for paying clients, published in public.
How to Buy Marketing in the Agent Era
Every buyer of marketing runs a checklist they never wrote down: how big is the team on my account, how many deliverables a month, what's the turnaround, how much of the senior person's time do I get. Those questions all meter production, which is the one input that just stopped being scarce. The claim here is stronger than "they're outdated." A proxy that loses its scarce input does not decay to noise — it reverses, and starts pointing at the partner who kept the most of the old cost structure. This is the Proxy Inversion, and it comes with four replacement questions a buyer can run in a single call.
Read the article →The Velocity Dividend
Every operator running this model says the same sentence: we ship in days what used to take months. Buyers hear a convenience claim — the same work, sooner — and a convenience claim can be matched by anyone willing to hire more people. That reading hides the actual mechanism. Speed is not how fast the work arrives; it is how fast you find out whether you were right. Shipping is a question put to reality, so the rate at which you ship is the rate at which you get answers, and answers compound where output only adds. This is the Velocity Dividend, and it is paid in calibration.
Read the article →When Execution Is Free, Taste Is the Product
For a century, marketing's binding constraint was production: making the thing cost people and hours, and that cost gated everything. That constraint just disappeared. When the cost of execution collapses to near-zero, the value does not vanish — it relocates to whatever is still scarce, and what is still scarce is judgment. The danger is that agents do not flood you with bad work; they flood you with competent work, which is exactly what used to be sellable. When competent becomes free and infinite, the product is no longer the output. It is the taste that chose it.
Read the article →The Pitch-Delivery Gap
In almost every agency, the person who sells you is not the person who does your work. The senior who wins the account moves on to the next pitch; the delivery drops to juniors you never met. That distance between who is sold and who serves is the Pitch-Delivery Gap, and it is not a bad-actor problem — the economics of a human agency force it. The operator model does not manage the gap. It removes the seam entirely.
Read the article →The Recommendation Gap
The last two issues taught you to be cited. I need to take it further, because citation is no longer the win. An AI engine can footnote you as a source and still recommend your competitor in the sentence your buyer actually reads. That distance — between being cited and being recommended — is the Recommendation Gap, and it is the part almost no one is optimizing. The research says the brands that close it are not the biggest ones.
Read the article →The Authority Leak
Last issue named the method — building claims to be quoted, not pages to rank. This issue aims it at the most authoritative content operation in marketing and pulls the trigger. HubSpot built the biggest corporate blog ever and still gets passed over by the machines answering its buyers' questions. This is the gap between authority and citation — the Authority Leak — with receipts.
Read the article →Writing to Be Quoted, Not Ranked
Search stopped being a list of links you click and became an answer a machine reads back to you. That changes the job. You are no longer optimizing a page to rank; you are engineering a claim to be quoted. This article names the atomic object of that work, the Citable Unit, and shows the exact construction we use on the article you are reading.
Read the article →The Four Attribution Models Founders Confuse
"Attribution" is one word doing four jobs. Founders use it to mean credit, cause, allocation, and story, then accept the answer to one question as if it settled another. This article separates the four models, names the costly error of crossing them (the Attribution Mismatch), and gives the one question to ask before you trust any attribution number.
Read the article →Not a Freelancer, Not a SaaS: The Agent-Staffed Function
One person running a full marketing function with a department of specialized AI agents underneath gets filed as either a freelancer with ChatGPT or a SaaS product. Both labels mispredict it. This article names the third structure, the Agent-Staffed Function, gives the three-layer test for whether a function qualifies, and shows the receipt that proves the shape transfers.
Read the article →The 27-Agent Architecture: Why Role Specialization Beats One Big General AI
Most products marketed as 'AI marketing' are one generalist model wearing different hats. The 27-agent architecture is the opposite bet. Three real receipts from the last two weeks show why role specialization is a quality lever, not an organizational nicety.
Read the article →The 4-Hour Competitive Teardown: A Protocol for Operators Who Don't Have Six Weeks
Competitive teardowns used to take a strategy consultant six weeks of calendar time and $30,000. The work itself was always under a working day. This is the protocol for compressing it back to that working day, with output a senior operator can act on Monday morning.
Read the article →The Queue Problem: Why Every Agency Pathology Has the Same Root Cause
Every frustration with a marketing agency — slow turnaround, fragmented teams, opaque pricing — is a downstream symptom of one structural cause. Once you can name it, every conversation with a current or prospective agency gets shorter.
Read the article →Get the next one in your inbox.
New artifacts go out when they are worth sending. No filler, no schedule for the schedule's sake.
What gets published here
Operational protocols
The actual playbooks the engine runs - which agents, in what order, with what prompts. Reader can copy and run it.
Operational case studies
Deeper than a portfolio entry. Focused on protocol and orchestration, not testimonials.
Engineered teardowns
A public brand’s recent marketing run through the engine. Concrete findings with screenshots.
Flagship frameworks
Long-form canonical references. The complete-guide-to-X pieces.
Failure post-mortems
Public analysis of what didn’t work. Numbers, diagnosis, fix. The build-in-public tax content.
Industry analysis
When something big shifts, the engine analyzes it within forty-eight hours.
For Linara’s broader operator-education writing, see Built Not Hired.
Visit Built Not Hired →