The Invisible Marketing Worker Is Here

Our first article made a case. This one introduces the solution.

‍

Every task an agent performs is one of two kinds. Judgment work - pricing, reading a buyer, earning a mandate - is what only a human can do. Intelligence work - the rule-based production around every listing - is what a system should do. Marketing is the largest block of intelligence work in most agents' weeks, and it stayed manual for one reason: the industry kept selling tools, not infrastructure.

‍

PropTech spent a decade making each step faster. Faster descriptions. Faster photo editing. Faster staging. Every tool sped up one task and added one more silo. Nobody built the system. Today we are opening the one we built.

From copilot to autopilot

A copilot helps a human work faster. An autopilot does the work itself. Almost every AI product in real estate is a copilot - it speeds up one step and hands back the rest of the chain to coordinate, brand, and distribute by hand.

‍

We did not build another copilot. We did not even stop at one autopilot for ourselves. We built the infrastructure that lets anyone build their own.

‍

This is Backbone Lab, the AI engine turning raw visuals into channel-ready marketing. It is the Invisible Marketing Worker - built to plug directly into the portals, platforms, and suite of tools your business relies on. We didn't just build one AI assistant; we built the factory that lets everyone power their own.

‍

‍

‍

‍

What the Backbone Lab is

The Backbone Lab is a REST API that runs the same generative workflows our own production teams use. Upload your assets, call a workflow, get the output back - asynchronously, at scale, with the AI infrastructure managed entirely by us.

‍

Think of these workflows as the creative building blocks happening behind the scenes:

‍

• Staging - indoor and outdoor virtual staging, room transformation, and indoor and outdoor renovation.
• Enhancement - enhancer, auto-HDR, sky replacement, declutter, mess cleaner, and privacy blur.
• Inpainting - magic eraser.
• Floor plans - 3D, isometric, and video.
• Video - image to video.

‍

The point is what happens when you chain them. A broker does not just want a sky replaced. They want raw capture turned into a full marketing package, automatically, in their brand. The Lab lets them bundle these features into one custom workflow that runs end to end, then call it like any other piece of infrastructure.

‍

A concrete use case. A broker pushes raw photos from a shoot. The workflow enhances them, replaces the sky, turns stills into video and walkthroughs, applies virtual staging, and stamps the company's watermark and branded font - without a human touching a step. The output is not one asset. It is a complete marketing kit formatted for every channel: the portal gallery, the Instagram post, the brochure. This is the production tax cut by 80% - time that used to disappear into formatting going straight back to the work that grows the business.

‍

The hard part was never the individual model. It was running dozens of them reliably, at scale, behind a single call. That is the part we absorb. You design the workflow. We keep the infrastructure standing.

Why managed infrastructure is the unlock

In April we wrote that marketing is not a pure software problem. It spans physical operations and digital workflows, and connecting both is what makes the difference. The same logic applies to AI infrastructure. Running a generative pipeline that holds up across thousands of listings a day is not a model problem - it is an operations problem: orchestration, version control, quality at scale, cost, uptime. We have argued before that owning compute is usually a capital trap. The corollary: running reliable AI infrastructure is work most teams should never have to build themselves.

‍

So the Lab inverts the burden. You define what your marketing output should be; Backbone carries the part that breaks at scale. You call a workflow - you do not staff an ML platform team to keep it alive. That is the difference between buying a tool and plugging into infrastructure. One adds a silo. The other removes the chain.

Proof: we are our own first client

The Backbone Lab is not a new bet we're running on customers. It's built directly from the same staging, enhancement, floor-plan, and video workflows our teams already run to deliver marketing content to thousands of agencies across Europe. The API is engineered around those exact workflows, and we're now migrating our own service business onto it. In 2026, we expect 100,000+ property listings to run through this infrastructure.

‍

The Lab is not a launch. It is a graduation - technology proven in production, opened to the broader market. We built the system for ourselves first. It works. Now anyone can build on it.

2030, arriving early

The frontier is already in private beta: the system automatically generates a TikTok video featuring the agent's own digital avatar and voice, alongside a complete marketing brochure ready to share. These features are dropping next.

‍

We laid out a five-year view in April.

‍

• ‍2026: the gap between agencies with marketing infrastructure and those without becomes measurable.
• ‍2028: production is fully automated for standard mandates.
• ‍2030: marketing infrastructure goes invisible - embedded in the CRM, activating the moment a mandate is entered, with the agent never opening a separate tool. Marketing the way electricity works: invisible infrastructure, visible output.

‍

An API-callable, composable workflow engine is exactly the mechanism that 2030 vision requires. Infrastructure does not become invisible by being a great app. It becomes invisible by being embedded - called silently inside the systems agents already use. With the Lab open, that mechanism exists today, ahead of schedule.

What's next

The Backbone Lab is built for the organizations that turn visual volume into channel-ready output at scale: agencies, portals, CRMs, and large content producers. If you coordinate marketing across many listings, teams, or brands, the Lab lets you build the production line once and run it on infrastructure you never have to maintain.

‍

Want to build on it? Request an API key, read the docs, or reach me on LinkedIn.