DAY 2 · 17 SEPTEMBER 2026
How to be a Wolf
Day 2 at Casablanca AI Lab, 17 September 2026. A Wolf is the operator who receives a validated prototype from an Octopus and turns it into a productive, scalable system. Where the Octopus proves that an idea is worth building, the Wolf makes it something a business can depend on.

The role
Take the prototype and make it real. This is the deepest technical role of the three: a strongly technical, logical thinker who in five weeks turns the prototype into a production-ready, scalable, hardened system, covering security, compliance, and every kind of user input.
The prototype arrives having proved a central workflow with mock data and a narrow path. The Wolf’s job is everything that has to be true before thousands of users can be on it. The roles are defined by personality rather than by skill list, and each owns its project phase end to end. The Wolf phase is five weeks, the biggest block in the delivery cycle.
The mission
Take a validated prototype, build the production system underneath it, and hand the Eagle a product worth polishing rather than a prototype still pretending to be one.
Five strengths
Scale
Getting from a working demonstration to thousands of users. This is the concrete bar the role is measured against.
Hardening
Security, compliance, and every kind of user input. A prototype only has to survive the path it was demonstrated on. A production system has to survive whatever a user does to it, and has to stand up to a compliance question afterwards.
The Octopus phase ends with a delivery-ready project package, produced through the Wolf-readiness workflow in the harness and then reviewed manually by the Octopus. That manual review is the final human-in-the-loop check before delivery starts, so a Wolf inherits documentation that has already been validated against the transcripts, the prototype, and the client’s feedback. The package contains:
meeting transcripts
harness sessions and decisions
the current prototype
client feedback and recommendations
confirmed requirements
open questions and assumptions
system, integration, and data requirements
process diagrams
scope, priorities, and acceptance criteria
risks and dependencies
pricing inputs and timeline assumptions
The standard the Octopus is held to is that the Wolf can take over without repeating discovery, and that the five-week delivery timeline feels generous. If either is not true, that is a handoff failure and it should be raised as one rather than absorbed quietly.
Boundary with the Eagle
The Wolf is not expected to catch the one-in-a-thousand edge case. That is explicitly the Eagle’s territory: the click path that makes the database behave oddly, the case that only appears at the margin. Nor does the Wolf own final interface polish.
This boundary is a licence, not an excuse. A Wolf who leaves a known defect for the Eagle should leave it written down, not discovered.