AI portfolio for practical tools and thoughtful systems

Esther L.B. Portfolio

I build AI-supported tools that make complex work easier to plan, explain, measure and improve.

Portfolio focus

Practical AI tools that improve planning, decision-making, learning and evidence-led delivery.

3 AI case studies

4 step method

Portfolio focus

What I want this portfolio to show

Clear problem framing

I start by defining the user, the pressure point and the decision the tool should support.

Human-centred workflows

I design AI around real handovers, checks, approvals and moments where judgement matters.

Useful outputs

I focus on briefs, plans, summaries and dashboards that help people act faster and with more confidence.

Responsible automation

I keep source evidence, context and review points visible so AI supports the work rather than replacing accountability.

Selected AI projects

Three AI tools I would build

Each case study shows the output and the thinking behind the build: the goal, the objective, the design choices and why I chose this version.

Inclusive Learning Copilot

I designed this as an AI assistant that turns a programme brief into a blended learning pathway, facilitator guide and safeguarding-aware delivery checklist.

Learning design Inclusive pedagogy CPD planning

Programme Builder

Module 1 Psychological safety

Scenario exercise, reflection prompt, facilitator note.

Module 2 Inclusive practice

Adapted tasks, accessibility checks, feedback loop.

Module 3 Action planning

Personal commitment, partner follow-up, evidence capture.

Goal

I wanted to help programme teams move from broad learning aims to structured, inclusive delivery assets faster.

Objective

My objective was to reduce manual curriculum drafting while preserving quality, safeguarding expectations and learner context.

Thought process

I started with audience, risk and outcomes before asking the AI to suggest activities, because learning tools are only useful when they respect context and inclusion.

Why this version

I chose a guided copilot over a fully automated course generator because human review matters in safeguarding, inclusion and programme tone.

Daily Briefing & Priority Intelligence Assistant

I designed this as a planning assistant that reads meeting context, deadlines and logistics, then produces a concise daily brief for busy teams.

Planning support Priority management Stakeholder prep

Daily Priorities

09:30

Partner review: confirm action owner, budget note and decision needed.

12:15

Travel buffer: 28 minutes. Attach venue access notes and contact number.

16:00

Board document pack: flag missing appendix and prepare one-line summary.

Goal

I wanted to give busy teams a reliable, low-noise view of the day so decisions, documents and logistics are ready before they are needed.

Objective

My objective was to combine calendar context, meeting purpose, logistics and stakeholder notes into a single preparation layer.

Thought process

I made the assistant prioritise risks, dependencies and next actions rather than long summaries, because preparation tools should reduce cognitive load.

Why this version

I chose a daily briefing format over a chatbot-only interface because people need to scan quickly and act without hunting for information.

Impact Insight & Evaluation Engine

I designed this as an AI workflow that collects feedback, clusters themes and drafts evaluation summaries for programme reporting and continuous improvement.

Evaluation Reporting Continuous improvement

Impact Dashboard

Confidence increased More practice time needed Strong facilitator trust

Goal

I wanted to turn open-text feedback and programme evidence into clear insight that helps teams improve delivery and show impact.

Objective

My objective was to speed up reporting while keeping interpretation grounded in participant voice, outcomes and objectives.

Thought process

I kept source evidence visible beside AI-generated themes so the person reviewing the report can see what each insight is based on.

Why this version

I chose a transparent insight engine over a black-box score because useful AI should make reasoning easier to inspect, challenge and improve.

How I approach AI tools

Human judgement first, automation second

Start with the user

I define who needs the tool, what pressure they are under and what decision the AI should support.

Map the workflow

I identify the repeatable steps: inputs, checks, handovers, approvals and reporting needs.

Build the smallest useful version

I create a narrow tool that solves one real problem before adding dashboards, integrations or extra prompts.

Keep review visible

I use AI to draft, organise and surface insight while keeping final judgement with the responsible human.