Selected Work | Close the Loop, Toronto

AI adoption

Seneca Polytechnic

Generative AI adoption, 2024 to 2026

Role AI Thought Leader at Seneca, co-lead with Kent Peel.

One of Canada's largest polytechnics wanted faculty, staff and students using generative AI in their actual work, with policy and practice moving together.

Co-led with Kent Peel, Seneca's other AI Thought Leader. A Viable System Model session located the missing piece at System 4, and we designed the AI Lab to fill it.

The AI Lab, from the case to the Senior Executive Committee to opening day, and a community of practice of hundreds of colleagues.

The Seneca AI Lab Opening day The adoption story at AskRight

The adoption, read through the Viable System Model

S5: Knowing who we areThe shared identity and values that hold everything together and settle the pull between today and tomorrow.
The Generative AI Policy, and input to the AI Governance Charter
S4: Looking outward and aheadWatching the world, preparing for the future, and keeping a clear picture of ourselves.AI Lab
Training, consultation and early prototyping
S3: Making the whole work togetherGetting more from the parts together than they would get alone, and agreeing who gets which resources.
A simple way to approve new AI tools, and everyday guidance
S2: Keeping the parts in stepThe schedules and shared rules that stop units from clashing with each other.
A community of practice of hundreds of colleagues
S1: Doing the workThe schools, programs and services that actually deliver for students.
Hands-on workshops and dozens of consultations

The AI Lab became System 4, the part of Seneca that looks outward and ahead.

Viable System Model after Stafford Beer. Figures as published on askright.ai/seneca, from the Seneca AI Work Retrospective, October 2026.

Hundreds

colleagues in the AI community of practice

4.6out of 5

average rating for the hands-on workshops

course sections running MyTutor within five months

80+

employees in one institution-wide session

Organization and strategy

Ecolabel Decision Support System

Funded research, 2012 to 2015

More than 450 ecolabels exist. Companies pick one on instinct, then fail its requirements or buy a label their market does not recognize.

A five-step online system: filter the field to the labels that fit the company's products and strategy, score the survivors on 56 characteristics weighted by that company's priorities, then assess compliance and return the actions that close the gap. Built on the Viable System Model.

A three-year, EUR 140,000 project delivered as a working tool, tested with 8 companies and 9 products across tourism, agriculture and construction.

Worked example: a hotel's services

The hotel already met only 57% of Green Seal's requirements. The three labels it met almost in full, 96 to 97%, were worth least to it.

One hotel's services, from the project's System A and B report, June 2014. Compliance is the share of each label's requirements the hotel already met: Green Seal 56.9%, Green Key 95.5%, BIO Hotels 95.7%, Carbon Reduction Label 97%. Horizontal position is the report's value order, not a score: Green Seal ranked highest, and the other three scored 16 to 24% on market impact. Only labels with a published compliance figure are shown.

Apigea

Agro-bio cluster, Greece, 2017 to 2018

A new cluster was to bring farmers, academia, research institutes, businesses and NGOs together around organic bee and medicinal plant products. The ambition was clear, the structure was not.

We mapped the idea onto a Flourishing Business Canvas, then used the Viable System Model to find the processes that deliver the purpose and nest them, rather than flattening everything onto one board.

Three value streams, each with a case for its own autonomy. The founder's vague coordinating role became an explicit choice: a cooperative the members run, or a cluster the founder governs.

One board, or three nested areas

Three areas, nested inside the holding. Each has a case for its own autonomy, and the holding coordinates between them.

Final project report, February 2018, and Panagiotakopoulos, Weisfeld, Upward and Pinnel, Sustainable Business Modelling. Area names as in the report; member dots are illustrative.

The advisory practice

Carbon accounting and life cycle assessment

City of Toronto

The City had committed to cut food-related emissions 25% by 2030, with no baseline to cut from.

We built the baseline across three divisions to WRI Cool Food Pledge and GHG Protocol methods, then the reporting tool the City tracks it with.

44,000 tonnes CO2e from 7 million meals a year, and a costed path to cut a quarter of it by 2030.

Project report CBC coverage

What the City buys

3%of the weightA rounding error in the shopping basket.

48%of the emissionsNearly half the footprint, in one segment.

Beef and lambOther animal-basedPlant proteinsOther plant-based

All 13 items, every measure

Beef and lamb 3% of weight, 48% of emissions
Dairy 29% of weight, 21% of emissions
Poultry 4% of weight, 6% of emissions
Pork 2% of weight, 4% of emissions
Eggs 3% of weight, 3% of emissions
Seafood 2% of weight, 3% of emissions
Grains 13% of weight, 4% of emissions
Legumes, nuts, seeds 2% of weight, 2% of emissions
Fruits and vegetables 32% of weight, 3% of emissions
Roots and tubers 5% of weight, <1% of emissions
Vegetable oils 2% of weight, 2% of emissions
Added sugars 2% of weight, 1% of emissions
Alcohol, stimulants, spices 1% of weight, 3% of emissions

Panagiotakopoulos, Onwudiwe and Wills (2024), City of Toronto Food-Related GHG Emissions and Analysis. 2022 figures, shares as published, rounded.

Ieropoulos Wines

Nemea, Greece

An organic winery needed its wine's carbon footprint, and the place to act, before making any claim about it.

Cradle-to-grave footprint to International EPD System and wine product category rules, modelled in GaBi against 2014 production data, every process rated for data quality.

2.9 kg CO2e per litre. The easiest lever was the bottle, not the cooling: a bottle half the weight would cut the footprint by about 13%, and the winery only had to order it.

Summary report

Where the footprint sits

23% 23% 54%

Bottle productionFermentationEverything else

23%of the footprintThe bottle the winery buys, and could simply order lighter.

Ieropoulos Wines carbon footprint summary report, 2015. Shares of 2.9 kg CO2e per litre, cradle to grave, rounded; "everything else" is the balance. The 13% is the report's sensitivity analysis for a 50% lighter bottle (12.7%), rounded.

The carbon and LCA practice

Also worked with

Business model and organizational design: BMO, Vivenda, MEDASSET, Apigea, Nisa Homes, Wuxly
Strategy: Grivalia Properties, PoliSpark
Carbon accounting: Strathallen, Grivalia Properties
Life cycle assessment: Ingredion, International Paper, ThredUp, Hanes, Rent the Runway, Copernicus Educational Products, StackTeck

Executive AI coaching and the rest of the practice

Work you can verify.