Designing Services with AI
I'm teaching a new course. It's for those who must turn novel ideas and creative solutions into services that efficiently deliver outcomes at a particular price point — using AI to design at a deeper, more structural level and get the engineering and economics right. It’s also about exploring ways of delivering the same quality of outcomes at lower costs — necessary under competitive and regulatory pressures. Designing with AI both allows for and requires a radically different approach that’s unconstrained by legacy formats such as blueprints and journey maps. Designs can exist and evolve in a high-dimensional space of parts, parameters and paths. And with AI tools capable of generating prototypes and low-level instructions, designers can move up the responsibility ladder and focus on developing ideas, declaring intent and defining behaviour.
A major skills upgrade
After this course you'll be able to do at least five things better: 1) Design services with a deeper understanding of what makes them beat expectations even when others in the same category struggle to do well. 2) Critically examine the design of any service and identify structural problems not obvious to others. 3) Develop designs that adapt and evolve with changing market conditions without the need for major changes. 4) Argue for and against design decisions that have board-level implications — with clarity and confidence appreciated by legal, finance, and operations. 5) Instruct AI tools with engineering precision to generate prototypes, functional specifications, and financial models.
What prompted this course
"For every dollar spent on software, six are spent on services."
That quote is from Services: The New Software, in which Julien Bek of Sequoia Capital predicts the growth of AI autopilots that will do the work and deliver outcomes instead of merely assisting teams or professionals. Adjacent to that is the emergence of service-as-software, in which customers pay for the outcomes of a performed service instead of access to platforms or tools. Both suggest there will be plenty of new work for those who design services. They also highlight a knowledge gap — traditional practices that see services mostly as journeys and interactions focus almost entirely on designing user experiences and not outcomes.
This course addresses that gap.
What makes it different
First, this course departs from traditional thinking. Great user experiences are necessary but not sufficient. Nobody buys an airline ticket just to enjoy check-in, screening, boarding, seating, or baggage claim. Everybody pays for the outcome of arriving at a destination on a particular date and time. As low-fare airlines have shown, experiences are actually the pains passengers are willing to go through for the outcome. From that perspective, everything necessary for the airline to deliver the outcome — from aircraft and crew, time slots and gates, to fuelling and maintenance — defines the critical function of the service and drives costs. Outcomes won’t materialise if any part of the critical function fails, no matter how good the user experience.
This course teaches critical function design.
Second, services are far more contractual in nature than software seats and subscriptions. Selling services and promising outcomes requires a greater acceptance of risk and responsibility. The commitments and considerations are fundamentally different. Several performances and affordances must happen in series and in parallel for the outcomes to materialise as expected. A poorly designed service, even when it doesn’t fail, falls short of expectations when outcomes and experiences are costlier than expected — not just for customers but also for the provider. Concurrency of demand and supply is part of the problem. In a sense, half the design of a service is about predicting, controlling, and adjusting for what customers do and don’t; have and don’t have. That’s why design efforts cannot ignore or be blind to the contractual dynamics at the heart of every service.
This course covers the contractual dynamics.
Third, services that beat expectations are well-designed systems free of gaps and conflicts in their designs and with the right levels of frictions and tensions between their components. But such optimality will always be out of reach if poor decisions have been made about which part should be physical, not virtual; fixed, not moving; transient, not persistent. Decisions on division of labour and delegation of control are also important — between customer and provider, principal and agent, human and machine. Balancing and alignment matter. Asymmetries are alright but incongruences and conflicts of interests can make a service costlier for everyone. Thoughtful decisions on such aspects of design can improve the overall system performance, allowing the service to beat quality and cost benchmarks.
This course teaches structural analysis.
Lastly, solutions must evolve with customer needs and the use of AI in designing and delivering services could mean cheaper and faster evolution. AI can discover new paths along which a service can evolve, keeping in mind structural integrity and performance. An evolutionary approach to design would also minimise the gap between what the service offers and what the market wants. It would also help avoid strategic blunders often made when reacting to competition or regulation, because there would seldom be the need to urgently implement major changes. But such an approach requires a high-dimensional layer of abstraction and reasoning within which AI can explore thousands of small, selective changes to the design of a service to improve core function.
This course introduces evolutionary design.
Course outline
The course will be taught online over a period of four weeks and eight 90-minute sessions. In each session, a short lesson introducing new concepts followed by a lengthy discussion around a case study. You may be asked to comment on a design decision, what you’d do differently and why. In between the sessions, you get to practice what you have learned — offline, at your own convenience, and with the assistance of an AI tool.
Week 1
— The basic unit, structure and ontology of a service.
— Building blocks, stereotypes, and patterns.
— Stories, frames, and sequences.
— Versions and variants.
Week 2
— Windows of opportunity and potentialities.
— Multi-factor assurance and thresholds of commitment.
— Arrangements (things) and agreements (people).
— Deviations, errors and exceptions.
Week 3
— Prospect theory and the pricing and packaging of outcomes.
— Experiences as inversions of pains and cost-coefficients.
— Division of labour and separation of concerns.
— Convergences and conflicts of interests.
Week 4
— Strategic choices and structural limitations.
— Functional specifications, flexibility, and tolerance.
— Problems of adaptation, scalability and growth.
— Reliability engineering as applied to services.
Two weeks after the course, there will be a special session during which you get to present the design you have developed and have a lively discussion about it with everyone. This session will last 120 minutes. Attendance is optional.
Study approach
You will study the design of a service, take it apart and then put it back together into a new format — an axiomatic, multi-layer abstraction using which AI can further develop the design for implementation according to your strategic intent.
The service chosen for the purpose is the collection of waste across municipalities. Although it sounds simple and mundane, the service has several subsystems including transport, processing and containment that present real design challenges. That affords us plenty of surface areas and several angles from which to examine its design. Also, the service is easier to visualise for everyone, being common. And while it is firmly in the physical realm, several of its components require analytics and data processing. Lastly, it will be fun to explore the potential for using AI in a service like that. In many cases the service operates at sub-optimal levels even as pressure grows on governments to be both fiscally responsible and reduce carbon.
The fictional client for this assignment is C40 Cities — a global network of mayors of 97 cities representing 920 million people, focused on climate action. The mayors are exploring the idea of a city genome — a set of service system ontologies covering all the infrastructures and services that form the DNA of a city, with waste collection being one of them. The genome will facilitate knowledge sharing across the network. What can Boston learn from Bogotá, or Cincinnati from Copenhagen, at an abstract level? That’s why the service system ontology you develop should be like a genotype, capturing the invariant logic: the critical functions, the contractual dynamics, and the structural elements that make waste collection a service anywhere. From that, local solutions can be generated by introducing city-specific parameters such as area, population, geography, climate, land use patterns, existing infrastructure, demographics, and regulatory framework.
While we maintain focus on a single service throughout the course, we will cover hundreds of other examples along the way to solidify and reinforce learning.
The assistance
You will need an Anthropic Claude subscription for the duration of the course. Claude will play the role of a design associate — analysing arguments, conducting research, and helping make key decisions. Before the start of the course you’ll receive a system prompt developed by the course instructor, that encodes the rules, principles and procedures of the design method being taught. The prompt will configure Claude to generate output that is more meaningful and relevant to the course and prevent it from drifting into generic thinking. The prompt will also later help you practice what you’ve learned and apply it to your own work, long after the course.
The level of thinking
The level of thinking in this course is set for the need to design with AI through abstraction and reasoning. According to Charles Owen, founder of the Institute of Design, it's through the process of abstraction that creative minds find new ways to conceptualize — "with the deliberate cultivation of wonder in combination with a quest for the essence of function". This course forces you to think at such an abstract level that the very concept of a service feels strange at first, before a new clarity emerges. As Kenya Hara writes in Designing Design (2011):
To understand something is not to be able to define it or describe it. Instead, taking something that we think we already know and making it unknown thrills us afresh with its reality and deepens our understanding of it.
So whether you are a designer, engineer, or product manager — or any other stakeholder in the success of a service — this course is for you if a deepened understanding of what services really are, why they come to exist, and why they fail, motivates and excites you. The emphasis on strategy, risk, engineering and economics is there to help you have more useful discussions with legal, finance, and operations. But above all, the thinking in this course is about applying creativity and imagination — to develop designs that are so good that others would want to study and copy them.
The instructor
Majid Iqbal is an industrial engineer and management consultant with thirty years of experience in the United States and Europe. He has advised Fortune 100 companies and major governments in several areas. His speciality, though, is structural improvements to the designs of services with the aim of increasing reliability, responsiveness and resilience. He is known for novel approaches to solving problems that have resisted conventional thinking. His book Thinking in Services explains why services exist, why they succeed and why they fail, by revealing the characteristic patterns and basic elements found in all services. In addition to speaking at conferences, Majid has taught master’s-level courses at Carnegie Mellon University and workshops across the world.
Registration
Purchase your ticket on Eventbrite.
