Agentic AI forOrganisational Research

From verified AI assistance to governed, autonomous research pipelines.

2 days / levelHands-on3 progressive levels

LLMs · Memory · Agents · Skills · MCP

The journey

Use confidently. Customise intelligently. Build responsibly.

Every level uses the complete agentic toolkit at increasing depth, anchored in authoritative sources, citation integrity and evidence-based decision making. You begin by using and applying ready-made skills and connectors, progress to optimising retrieval workflows grounded in legislation, official standards and your organisation's own knowledge base, and arrive able to create and share your own skills, MCP servers and agent harnesses with your own guardrails.

Basic

Use & Apply

Prompt, search and use ready-made skills and connectors while verifying every answer against original texts. No prior AI experience required.

CapstoneEvidence-backed research note

Intermediate

Optimise & Customise

Tune RAG workflows, adapt skills and combine MCP connectors for systematic, source-grounded research and citation-integrity practice.

CapstoneMini literature review + verified bibliography

Advanced

Create & Share

Build skills, MCP servers, knowledge vaults and governed multi-agent pipelines with adversarial verification.

CapstoneReview-ready research brief

What participants will master

Interrogate policies, standards and authoritative research sources with AI assistance
Build verification-first workflows with source grounding, citations and audit trails
Progress from ready-made tools to customised RAG, skills and MCP connections
Design governed research agents that search, draft, verify and cite
Preserve verified institutional knowledge in a long-term research vault

Programme format

2 days per level

9:00 am – 5:00 pm

Morning: 70% theory · 30% practical
Afternoon: 30% theory · 70% practical

Designed for

Policy · Research · Strategy

Compliance · Innovation · Digital

One toolkit, three depths

LLMs: prompt and verify answers → tune retrieval workflows and model behaviour → build new harnesses with your own guardrails
Memory: manage short-term session context → ground answers in curated notes → build a long-term knowledge vault agents read and write
Agents: use a single agent with tools → direct agents through multi-step workflows → build autonomous research agents
Skills: run ready-made research skills → edit and adapt existing skills → author and share new skills across your team
MCP: use ready-made connectors → configure and combine connectors in workflows → build MCP servers for vaults and databases

Adaptable to your sector

Content is grounded in your sector's legislation, official standards and internal knowledge base. See the Shariah Research edition for an example.

Programme logistics

Delivery: in-person and hands-on; participants bring a laptop. Up to 25 participants per cohort for effective lab supervision
Assessment: capstone deliverable at each level, assessed on source accuracy, citation completeness and verification discipline; certificate of completion awarded
Prerequisites: levels are sequential: Level 1 open to all, Level 2 requires Level 1 or equivalent, Level 3 expects basic comfort with the command line and formats such as JSON
Pre-course checklist: sent before each cohort; covers accounts, software installs and laptop requirements per level
Information security: all labs use public or sanitised documents only; no confidential organisational data enters unapproved AI services