Everyone is talking about AI.
But where does it actually create value?

Engagements

Strategic and evidence-based support across the AI and data lifecycle. Explore any engagement for the full picture.

AI & Data Opportunity Assessment

Before any investment, I help you find out whether AI is genuinely the right answer to your problem, or whether something simpler would serve you better.

What this covers

• Priority problems worth solving
• Genuine AI opportunities, not assumed ones
• Data availability and quality
• Organisational readiness
• Potential value and realistic risks
• Clear, recommended next steps

AI & Data Strategy

I help you turn a validated opportunity into a prioritised, resourced roadmap, so AI investment follows evidence rather than trends.

What this covers

• Strategic vision aligned to real priorities
• Opportunity prioritisation
• AI roadmap and data strategy
• Investment priorities
• Governance and implementation pathways

AI System & Value Diagnostic

If an existing AI system is not delivering, I help you find out exactly why, and what is actually worth fixing.

What this covers

• Problem definition check
• Data quality review
• System performance and user adoption
• Cost and ROI
• Success metrics and governance

Applied AI & Intelligence Design

Once a problem is validated, I help architect and design the right solution, matching the technology to the actual constraints on the ground.

What this covers

• Problem architecture
• AI solution design
• Data architecture and technology selection
• Prototype planning
• Implementation roadmap

AI Evaluation, Value & Governance

Drawing on my evaluation background, I help you measure whether an AI system is genuinely working, and govern it responsibly.

What this covers

• Performance, impact and ROI
• Adoption and responsible AI
• Governance and risk
• Ongoing monitoring

Data & Decision Intelligence

I help turn existing data into something that actually informs decisions, not just another dashboard nobody opens.

What this covers

• Data assessment and strategy
• Analytics and intelligence systems
• Decision support and knowledge systems

Training & Capacity Building

I have trained evaluators across 25 plus UN agencies in practical, no-code AI methods. I bring the same approach to your team.

What this covers

• AI literacy and AI strategy for leaders
• Responsible AI
• Data literacy and AI evaluation
• AI for public systems and development

The Lab

Applied Intelligence, Research, and Strategy only go so far without Execution, the last letter in ARSHEE. This is where that happens: ideas from advisory work and evaluation practice, turned into tools rather than left as recommendations. Some are live, some are still taking shape. This is what is in the workshop.

Sutra

Smart Unified Toolkit for Research and Assessment

Evaluation has always been a discipline of judgement: knowing what a number can and cannot claim, what a design can and cannot prove. Sutra is built on the belief that AI should sharpen that judgement, not replace it. It drafts the structure, checks for gaps, and keeps every claim traceable back to its source, while the evaluator decides what is actually true.

That division of labour, machine for structure and speed, human for judgement and accountability, is the same one explored in the Hybrid Intelligence for Evaluators work with UNFPA. Sutra is that idea built into a working tool.

Why I built this

I sit on the boards of APEA and IOCE, two of the largest global evaluation networks, and my own evaluation and AI work has taken me into organisations ranging from grassroots NGOs to Japan's Ministry of Foreign Affairs. Across all of it, I kept meeting the same problem: the evaluation process itself is scattered. The Theory of Change lives in one document, the budget in a spreadsheet, the indicators in someone's memory, and the final report gets stitched together from all of it under deadline, with the evidence trail thinning out every time something moves between tools.

Sutra exists so I would stop solving that problem project by project. It is built on what I think of as human-centric intelligence: AI speeds up structure and catches what you would otherwise miss, but it never runs unsupervised, and you can switch it off entirely if you would rather work without it. The platform holds itself to the same values I would hold any evaluation to: empathy for the people the data represents, rigor in what gets claimed, and accountability for who decided what.

From evaluability assessment through Theory of Change, indicator design, instrument building, data collection, and analysis, to a final report where every number is linked to the data, method, and instrument behind it, Sutra keeps the paperwork honest so you can keep the judgement.

A look inside

Setting up a workspace. The first step decides which of Sutra's 18 modules, spanning six phases (Governance, Research, Design, Collect, Analyse, Deliver), lead the workspace.

Choosing the Deliver-stage modules: budget, workplan, templates, the report, and the Master Export that assembles everything into one signed document.

Inside a workspace. Modules light up as they are switched on, each one built for a specific stage of the evaluation cycle.

What it does

• Evaluability assessment and ethics and safeguarding review before you commit to a design
• Theory of Change, Logframe, and Evaluation Matrix, built visually and linked together
• Indicator library with disaggregation built in
• Instrument design with KoBo, CSV, and Excel export
• Quantitative and qualitative analysis, with every finding pinned as AI-assisted until a human verifies it
• Budget, workplan, and compliance checks, including CSR Rule 8(3)
• A Report and Master Export where every claim is clickable back to its source, and AI-authorship is disclosed rather than hidden

All of it lives in one workspace organised across six phases, governance and ethics, research, design, data collection, analysis, and delivery, so a team can start wherever their project already stands instead of being forced through the same sequence every time.

Selected experience

A curated selection of work across AI, data and public systems, not a complete record.

Selected publications and speaking

• LAK'25 Workshop on Hybrid Intelligence, Dublin, Ireland (2025) — Operationalizing Hybrid Intelligence in Learning Analytics
• UNFPA Blog (2025) — Hybrid Intelligence: the intentional space between human and artificial intelligence
• 5th APEA Conference, Tokyo (2025) — Empowering Youth in Monitoring and Evaluation: Bridging AI, Ethics and Law
• UNFPA Summit for the Future of Evaluation, Colombo (2025) — Invited speaker on ethical AI and data innovation
• IIAS-DARPG Conference, Government of India (2025) — Future-Ready Public Administration in India
• BRICS NSO side events on AI Readiness, Lucknow (2026) — Nominated by Transform Rural India

Does your organisation actually need AI?

Not every problem requires AI. But the right problem, supported by the right data and strategy, can create meaningful value. Start with an initial assessment. Click each step to see what happens.

Insights

Notes on AI strategy, data strategy, value, evaluation and responsible AI. New articles coming soon.

Let's start with the problem.

If your organisation is exploring AI, struggling to create value from an existing AI investment, or looking to turn data into better decisions, I would be happy to start a conversation. Choose whichever is easier for you.