AI strategy and business architecture for the Government of Canada
Astraterra helps federal and Crown executives turn AI investment into results they can report: work that has been redesigned, benefits that are measured, and value that reaches the budget.
The problem
The time AI saves is real. It rarely shows up in departmental results, and the reasons are organizational, not technical.
AI is bought for its capability, not for an outcome. With no target measure and no plan for the capacity it frees, saved time is absorbed by email, meetings and rising demand. Below a certain gain, it stays a convenience.
Benefits-Driven AIAn assistant speeds up each task, but the file still waits in every queue between teams. Elapsed time barely moves, because the task was never the constraint. Return comes from redesigning the work around what AI does well.
AI-native process redesignStrategy sets ambitions the organization can't yet deliver. Governance is built to approve pilots, not to scale them or hold anyone to their benefits. Each works on its own, and the value is lost in the gaps between them.
A connected foundationThe pattern is visible in the Government of Canada's AI Register: reviewed entry by entry, few of its 400+ systems identify measurable fiscal or mission benefits.
Services
Each practice area answers one of the reasons AI value goes missing. Most engagements draw on more than one.
Start every AI investment from the outcome it should move, how that will be measured, and what happens to the capacity it frees. A process typically needs about a 1.3× productivity gain before its benefits can be harvested. Below that, AI is a convenience.
Built on Benefits-Driven AI
Redesign the work around what AI does well, instead of speeding up the steps it already has. Start from the program's goal and its binding constraint, and keep accountability where it belongs.
Built on CLAIR
Connect what the organization intends, what it is able to do, and how it decides, so pilots scale and benefits have owners. Includes the delivery lifecycle that takes solutions from proof of concept to production with the traceability the GC expects.
Built on AI Organizational Maturity and the AI Solution Development Lifecycle
The method
The three practice areas work as one method. The harvest is planned first, the work is redesigned to deliver it, and a connected foundation lets it scale.
The departmental result the investment is meant to move.
How the result will be measured, and the baseline it starts from.
The work rebuilt around what AI does well, starting from the binding constraint.
Freed capacity redeployed and the value banked, only once it shows in production.
Practice 03 · The connected foundation
Which outcomes matter, in what order, and who sponsors them.
What the organization must become able to do so the results last.
How AI is approved, delivered, scaled and held to its benefits.
Inside the redesign stage
CLAIR draws on the Theory of Constraints, process reengineering, Lean, Six Sigma, decision modelling and benefits realization management, and adapts them to the ways AI breaks the assumptions those methods were built on. It starts from the program's goal and its binding constraint, never from "where could we use AI?"
Redesign in practice
Follow a mocked-up grant application through today's five-team process. Add AI to every step and each task gets faster, but the file still waits in every queue. Redesign the process around AI, with the same accountability, and the queues between teams disappear.
Today140 daysAI added to every step122 daysRedesigned around AI25 days
Track record
Government of Canada
Industry
The principal
Astraterra is led by Terry Feltham, an engineer and business architect with 30 years of experience in redesigning how organizations work. He also brings over 13 years of experience in machine learning and artificial intelligence, having published a first article on AI commercialization in 2017. Terry served 20 years in the Royal Canadian Navy, and has held senior roles with PwC, IBM and the federal public service.
The firm rests on one view: adopting AI is mostly an organizational design problem. The technology is capable. What's usually missing is a clear strategy, redesigned work, and the discipline to measure and capture what AI delivers.
Working together
Start with a briefing or a Scan, and decide what comes next on the evidence.
Half day
For a DG, ADM or CFO leadership team: where AI value is realistically available in your organization, and what it would take to capture it.
3 to 4 weeks
One value stream: its goal, its binding constraint, the key decisions and redesign options, ending in a clear go or no-go.
6 to 9 months
The full CLAIR lifecycle on one value stream, through a pilot to measured, captured value.
Scoped to need
A departmental AI strategy, benefits plan and governance model, with a prioritized investment roadmap.
Latest thinking
September 2026
Faster work is not the same as realized value. Five management disciplines that turn AI efficiencies into results a department can count.
July 2026
Why deputies must personally lead their department's AI transformation: the gap between AI's promise and its realized value can only be closed at the deputy's desk.
June 2026
AI is now everywhere, but demonstrated value is not. Why every AI investment should begin with the departmental measure it is meant to move.
Contact
A 30-minute conversation, without a pitch deck. Bring a program, a process or a pressure.