The AI Infrastructure Playbook is the productized version of our audit practice: an editable Operating Canvas that maps every department to the right model placement (local SLM, private LLM, cloud frontier, or human) with the cost math and the compliance flags attached. One image to brief your board. Refreshed quarterly as the model landscape moves.
This is a live slice of the canvas: a 200-person company, eighteen workloads, two ways to run them. Flip the toggle and watch what placement does to the monthly bill. Click any department to see the reasoning and the compliance flags.
Illustrative numbers for a 200-person company at typical usage. Your canvas is built on your headcount, workloads, and quotes. That's the point of it being editable. High-risk items (hiring, anything customer-affecting) carry an EU AI Act flag and keep a human gate regardless of placement.
An editable department-by-department map, in FigJam and PDF, of where every AI workload should run and what it costs. The board briefing is one export away.
60+ workload placements at launch: the task, the smallest capable model class, the per-task cost logic, and when to escalate to a frontier model.
Reference configs for the classify → compress → gate → route → evaluate loop, sized from a single workstation to a colocated cluster.
Every placement tagged against the EU AI Act, GDPR, SOC 2, and HIPAA, and mapped to the NIST AI RMF core functions, so the compliance conversation happens at design time, not audit time.
The break-even spreadsheets behind our audits (cloud vs owned vs hybrid) pre-wired so you can drop in your own quotes and usage.
Model prices, capabilities, and the placement map move fast. The canvas and library are re-issued every quarter; your subscription stays current.
The Playbook sequences the same operating layer we install by hand: one quarter, three phases, each one shippable on its own.
The General edition ships at launch. Vertical editions, the same canvas re-drawn for the workloads, regulators, and data boundaries of your industry, roll out to founding members in order of demand. Your signup is a vote.
The Playbook launches Q3 2026. Nothing is charged today. This is the founding list. Members lock $59/month for life (versus $99 at launch), get the General edition the day it ships, and set the order the vertical editions are built in.
No, it's the productized output of our consulting. The audit engagement replaces every assumption with your measured numbers; the Playbook gives you the same placement logic, canvas, and governance overlay to apply yourself. Many teams start with the Playbook and upgrade to an audit when the decision gets board-sized.
The General edition: the editable Operating Canvas (FigJam + PDF), the placement library (60+ workloads), the routing blueprints, the governance overlay, and the TCO spreadsheets. Vertical editions follow, sequenced by founding-member votes. If launch slips past Q3 2026, founding members hear first, and nobody is charged until it ships.
Because the math is public. The free TCO calculator and readiness assessment on this site run the same models the Playbook packages, so you can pressure-test our logic before paying anything. We publish break-even math, not testimonials.
That's the advisory practice: TCO audit, spend audit, implementation, and the managed private stack. The Playbook and the practice run on the same operating layer; you can move between them without starting over.