The UK government has firmly signalled…

"...For SME leaders, the message is unequivocal: the time to engage with AI is now. It is no longer a peripheral technology but a central pillar of modern business strategy."

Why Agentimise

"To empower founders, CEOs, and leadership teams of startups and SMEs to fully harness the transformative potential of AI—ensuring British companies lead the world in performance, productivity, and innovation."

Our Audience

The Urgency vs. Fluency Gap

Leaders universally recognise AI's critical importance, yet a significant gap exists between this awareness and their own practical AI knowledge, hindering confident decision-making. A Boston Consulting Group study found that while 89% of leaders rank AI as a top-three priority for 2024, many feel personally unprepared to lead the charge.

The Guidance Vacuum

Your teams are already experimenting with AI and are looking to you for a clear vision, strategy, and ethical guardrails. Without it, innovation becomes fragmented and risky. A 2024 Microsoft/LinkedIn report found that 78% of AI users are bringing their own tools to work, highlighting a critical need for official guidance.

The Strategy Paralysis

The lack of deep AI understanding at the leadership level leads to hesitation and an inability to separate hype from genuine strategic opportunity, putting competitiveness at risk. PwC research indicates that companies struggling to scale AI often lack a clear, leadership-driven strategy that connects technology to tangible business value.

Purpose & Outcome - By the end of this course

Our Purpose

To bridge the critical gap between AI awareness and strategic action, empowering leaders to move beyond the hype and make confident, informed decisions that drive real business value.

Your Outcomes

  • Develop an actionable AI strategy aligned with your core business objectives.
  • Make smarter, high-impact investment decisions on AI initiatives.
  • Lead your teams with clarity and confidence through the AI transition.

What you'll Need for this Workshop

This is not about teaching you how to use AI but to motivate you to use AI forever going forward.

You will need:

  • Curiosity
  • A sense of adventure
  • Teamwork
  • A laptop with an AI interface (e.g., ChatGPT, Gemini, Co-Pilot)
  • A willingness to unlearn past habits and an understanding of how to delegate.

The key is to learn how to talk to AI like you are talking to your colleague.

How AI-Ready is Our Leadership?

Strategy: Experimentation or Integration? Are our AI initiatives a series of disconnected experiments, or are they guided by a central strategy tied directly to our core business objectives?

Value: Cost-Cutting or Revenue-Growing? Do we view AI primarily as a tool for internal efficiency, or are we actively exploring how it can create new products, services, and revenue streams?

People: Tool-Users or Co-Pilots? Are we simply giving our teams AI tools, or are we fundamentally rethinking workflows and upskilling our people to partner with AI for higher-value work?

Risk: Avoidance or Management? Is our primary approach to AI risk to restrict its use, or are we proactively developing governance to manage risks and unlock its potential safely?

Leadership: Delegated or Owned? Is AI considered an "IT project," or is the executive team actively leading the charge, fluent in its implications and accountable for its success?

The New Economics of Intelligence: From Hiring to Buying

The Old Rule: Hire Intelligence

Intelligence is scarce & expensive. Acquired by hiring a human. Bundled with salary and overhead. Scaled slowly, one person at a time.

The New Reality: Buy Intelligence

Intelligence is abundant & cheap. Acquired on-demand like a utility. Unbundled into precise tasks. Scaled instantly, on demand.

Strategic Impact: The ability to acquire and apply intelligence is no longer a constraint of hiring. It is now a strategic choice of procurement, fundamentally changing business models and competitive landscapes.

The Adoption Paradox and the Competitive Gap

The UK's AI Adoption Gap: A Tale of Two Speeds. Adoption is Accelerating, But Unevenly. While overall SME adoption is growing rapidly, with Moneypenny data suggesting it could be as high as 38.8% in 2025, a significant divide is emerging based on company size.

The Size-Based Competitive Chasm

15% of small companies (10-49 employees) have adopted AI.
68% of large companies have adopted AI.

The Bottom Line: Larger competitors are building insurmountable advantages with AI at a pace that threatens to leave SMEs behind. The time to act is now.

Where AI Delivers Strategic Advantage

Innovation & Product Development

What it does: Acts as a tireless brainstorming partner and rapid prototyper.
Business Impact: Drastically shortens development cycles and reduces R&D costs, getting ideas to market faster.
Example: Generating multiple software or product design mockups in minutes instead of weeks.

Marketing & Customer Experience

What it does: Creates tailored content and unique experiences for individual customers.
Business Impact: Unlocks hyper-personalisation at scale, a strategy once only available to the largest corporations.
Example: Crafting unique marketing emails for thousands of customers based on their individual purchase history.

Operational Efficiency

What it does: Summarises complex documents, drafts communications, and structures information instantly.
Business Impact: Frees up your team's time from low-value tasks to focus on strategic, high-impact work.
Example: Condensing a 50-page report into key bullet points in seconds, preparing a leader for a critical meeting.

The Unguided AI Workforce: Your Team Isn't Waiting

The Reality: "Bring Your Own AI" is Already Here. Your employees are not waiting for official policy. A staggering 78% of AI users are bringing their own tools to work (Microsoft/LinkedIn, 2024), creating a 'shadow AI' ecosystem within your business.

The Consequence: A Widening Internal Gap

Without a central strategy, this ad-hoc adoption creates chaos:
'A' Players use AI to become hyper-productive, creating new internal benchmarks.
'B' Players dabble without clear purpose, leading to inconsistent results.
'C' Players fall further behind, creating a productivity divide that damages team cohesion.

The Bottom Line: The absence of an AI strategy is, in itself, a strategy—one that allows chaos, risk, and inequality to define your company's AI transformation.

Terminology

CategoryAcronymFull Name
Foundational ConceptsAIArtificial Intelligence
Foundational ConceptsLLMLarge Language Model
Foundational ConceptsNLPNatural Language Processing
Strategic & Applied AIGenAIGenerative AI
Strategic & Applied AIGPTGenerative Pre-trained Transformer
Strategic & Applied AIRAGRetrieval-Augmented Generation
Technical IntegrationAPIApplication Programming Interface
Learning MethodRLHFReinforcement Learning from Human Feedback

Which LLM's for Which Job?: Always Changing

  • For Maximum Creative Power & Complex Reasoning: Use Anthropic's Claude 3 Opus.
  • For the Best All-Around Performance & Integration (The "Swiss Army Knife"): Use OpenAI's GPT-4o.
  • For Handling Extremely Long Documents & Video Analysis: Use Google's Gemini 1.5 Pro.
  • For High-Speed, Cost-Effective Everyday Tasks: Use Anthropic's Claude 3 Haiku or Meta's Llama 3 8B.
  • For Open-Source Customisation & Control: Use Meta's Llama 3 70B.

LLM user interfaces

VendorUser Interface (The App You Use)LLM (The Engine Under the Hood)
OpenAIChatGPTGPT-4, GPT-4o
GoogleGeminiGemini 1.0 Pro, Gemini 1.5 Pro/Flash
MicrosoftCopilotGPT-4, proprietary Microsoft models
AnthropicClaudeClaude 3 (Opus, Sonnet, Haiku)
MetaMeta AILlama 3
xAIGrokGrok-1, Grok-1.5

What Questions Do You Have??

Lab Exercise 1

  • Access Via Google Drive - Link
  • Become familiar with your preferred interface (ChatGPT, Gemini, Co-Pilot)
  • Security permissions

How an LLM 'Thinks': A Simple Analogy

An LLM works by being an incredibly sophisticated predictor, guessing the next most likely word in a sentence, like "The cat sat on the ???"

Phase 1: Building the Brain
The model is trained by reading a massive portion of the internet, learning the relationships and patterns between billions of words and sentences.

Phase 2: Sophisticated Prediction
When you give it a prompt, it doesn't "think" or "understand"; it calculates the most statistically probable next word to follow your input, then the next, and the next, to form a coherent response. For example, it predicts the best words to follow your instruction to draft a sales follow-up email.

The Bottom Line: At its core, an LLM is a complex "what-word-comes-next" engine.

How an LLM Learns to be 'Correct'

This process is how we teach an AI to move from just generating text to providing helpful, accurate, and safe answers.

Step 1: The Initial Guess - The AI is shown a picture of a cat and incorrectly labels it "Dog". ❌

Step 2: The Human Feedback - A human trainer sees the error and provides the correct label, "Cat".

Step 3: The Model Correction - The model's algorithm learns from this feedback, adjusting its internal parameters to strengthen the connection between that image and the word "Cat".

Step 4: The Corrected Guess - The next time the AI sees a similar picture, it correctly labels it "Cat". ✅

The Bottom Line: This process, called Reinforcement Learning from Human Feedback (RLHF), is crucial for aligning AI models with human values and making them useful for business.

Comparative Intelligence: Human vs. LLM (IQ Range Estimates)

IQ Score Comparison (Estimated Ranges)

80 |-----------------|-----------------|-----------------|

100 |#########| (Avg. Human)

120 |-----------------|-----------------|-----------------|

140 |-----------------|-----------------|-----------------|

160 | |###############| (Claude 3 Opus)

180 |-----------------|-----------------|-----------------|

200 | |############| (GPT-4)

220 |-----------------|-----------------|-----------------|

240 |-----------------|-----------------|-----------------|

260 | |#####################| (Marilyn vos Savant)

Key Takeaways

  • Superhuman in Narrow Domains: Top LLMs now test in the top percentile of human performance on standardized tests.
  • Range, Not a Single Number: AI performance varies by task.
  • No Consciousness: High IQ scores do not imply consciousness, understanding, or common sense.

The Most Important AI Skill: The Quality of Your Prompt

Vague Prompt (Garbage In...)

The User Asks: "Tell me about a marketing plan."
The AI Replies: "A marketing plan is a comprehensive document..."
Result: Unhelpful, generic information. 👎

Strategic Prompt (Value Out)

The User Asks: "Act as a CMO for a UK startup launching a new B2B SaaS product for the finance industry. Create a high-level, three-month marketing plan..."
The AI Replies: "Here is a 3-Month Go-to-Market Plan..."
Result: Actionable, tailored strategy. 👍

The Bottom Line: Effective prompting is not about asking a question; it's about providing a detailed brief with context, persona, and constraints to get a valuable result.

The Anatomy of a Strategic Prompt

ROLE: Tell the AI who to be. (Example: "Act as an expert financial analyst.")

CONTEXT: Provide the background and "why." (Example: "We are a B2B company preparing for our Q3 board meeting.")

TASK: Give a clear, specific, and actionable instruction. (Example: "Analyze the attached sales data and identify the top three trends.")

FORMAT: Define the structure of the output you want. (Example: "Provide the answer as a bulleted list, followed by a one-paragraph summary.")

And the crucial fifth step... REFINE: Treat the first response as a first draft. Give the AI feedback to improve it.

Organize your conversations by Project

To get the most out of your AI, stop treating it like a single, endless conversation. The most effective method is to organize your conversations by specific projects or goals.

Conversation Title: Q4 SaaS Launch - Marketing Plan
Prompt: [ROLE: Act as a CMO]...

Conversation Title: Q4 SaaS Launch - Blog Post Ideas
Prompt: [ROLE: Act as an expert copywriter]...

The Rule of Thumb: When your CONTEXT or major TASK changes, start a new conversation.

Short Term Memory: The Context Window

Think of an AI's memory as a temporary notepad. As the conversation gets longer, older information scrolls off and is forgotten. To combat this, use a "Memory Refresh" prompt to force the AI to summarise and remember what's important.

"Memory Refresh" Technique

Example Prompt: "Before we continue, I want you to refresh your memory. Your role is to act as a Chief Marketing Officer. The project we are working on is the Q4 launch of a new B2B SaaS product... Please summarise these key points to confirm you understand the mission context."

Lab Exercise 2

  • Practice good & bad prompting
  • Select from a sample of ideas or do your own
  • Bouncing chats between two Chats
  • Booking a vacation
  • Organise your Chats
  • Short Term Memory example

AI in Action: The Marketing Manager

Market & Competitor Analysis

Before: Days compiling data manually.
After: Asks AI for a competitor summary, ready in minutes.

Content Creation

Before: Hours struggling with writer's block.
After: AI generates outlines and drafts instantly.

Campaign Reporting

Before: Hours pulling data and creating charts.
After: AI analyzes data and writes summary paragraphs.

AI in Action: The Sales Representative

Prospect Research

Before: Hours of manual research.
After: AI creates a one-page briefing in seconds.

In-Call Support & Post-Call Admin

Before: 30 mins deciphering notes.
After: AI transcribes, summarises, and drafts follow-ups.

Proposal & RFP Creation

Before: Days searching for info.
After: AI drafts answers from a knowledge base.

AI in Action: The Financial Analyst

Financial Reporting

Before: Days exporting data to Excel.
After: AI generates reports and explains variances.

Investment Research

Before: Manually reading dense reports.
After: AI analyzes transcripts and summarises risks.

Scenario Planning

Before: Building complex, error-prone spreadsheets.
After: AI forecasts and models scenarios on command.

AI in Action: The HR Manager

Talent Acquisition

Before: Manually sifting through CVs.
After: AI screens candidates and drafts outreach.

Policy Development

Before: A week to draft a new policy.
After: AI creates a comprehensive draft in minutes.

Employee Onboarding

Before: Manually preparing schedules.
After: AI creates personalised plans and acts as a 24/7 chatbot.

AI in Action: The Legal Counsel

Contract Review

Before: Half a day reading manually.
After: AI reviews and flags risks in minutes.

Legal Research

Before: Hours searching databases.
After: AI finds and summarises relevant case law.

Regulatory Compliance

Before: Manually cross-referencing policies.
After: AI analyses new legislation and identifies impacts.

Customer Example 1a: A Planning Office - Reports

Challenge One

A planning office in Bournemouth can not find quality 'Planning Officers'. This was a recruitment problem.

Solution

Instead of hiring more, we took 'load' away from the existing Planners so they could deliver more interesting and challenging work. Each local authority requires planning applications in their own specific format. We ran 50 historic applications through AI so it could learn what 'good' planning application looked like. We pointed AI towards all the authority sites where rules and regulations existed. AI can now produce an 80% of a 50 page planning application, including google earth images, environmental checks etc saving each planner 25% of their weekly load. The planner is 'accountable and responsible' for the final 20% and document approval.

Customer Example 1b: A Planning Office - Researching

Challenge Two

Weekly there are changes to the construction and planning laws and planners must be aware and then understand the impact.

Solution

Build a research Agent which searches government, local authorities, and other sites for changes. Summaries the changes and priorities the relevance of the change depending on the Planning Practice are of expertise.

Customer Example 2a: An engineering company - Gathering customer insights

Challenge One

Important customer information on Sales Calls was not being recorded which meant valuable market insights were being leveraged by the marketing team. E.g Location of the enquiry, size of the property and thus likelihood to afford the product.

Solution

Record the conversation between sales person and prospect. At the end of the call AI performed multiple bits of research. Using the Post Code, AI used Zoopla to check property values. Used Linkedin to understand the job titles of the prospective buyers along with credit checks to ascertain the probability of the prospect affording the product. If there are existing customers in the same area as the prospect. This helped the sales team to prioritise prospects and gave marketing valuable insights as to where future prospects might live.

Customer Example 2b: An engineering company - Gathering Technical insights

Challenge Two

Important customer information was being lost by the pre-sales engineers visiting the site.

Solution

Either onsite and/or whilst the engineer drove back to the office AI would transcribe a voice recording extracting a variety of valuable insights e.g. site access restrictions, ground conditions, the style and dimensions of proposed construction, characteristics of the prospective client, etc. The information was then put into two buckets; 1- Market Data and 2- Engineering

Live Demo: Key AI Tools for Visual Content Generation

Google Veo Link

CategoryToolVendorWhy it Matters (For a Leader)
Graphics & Image GenerationMidjourneyMidjourneyIt produces the highest quality, most artistic images...
Graphics & Image GenerationDALL-E 3OpenAIIt's the best for integrating text and complex concepts...
Graphics & Image GenerationImagen 2GoogleIt offers unparalleled realism and brand safety features...
Video GenerationSoraOpenAIIt generates hyper-realistic, minute-long videos...
Video GenerationVeoGoogleIt offers fine-grained cinematic control and consistency...
Video GenerationKlingKuaishouIt produces high-fidelity, two-minute videos...

Live Demo - Building this Training course

The Challenge: Build a presentation which is 'self maintaining', never out of date.

This slide deck is built by AI using human language as the instructions.

  • Stage 1 - create a document which describes what I want in each slide to be
  • Stage 2 - Using Google AI Studio and a well structured prompt, have it read the document (PDF). My prompt turns the document into a 'structured prompt' whilst performing live research to ensure the content of the slides are up to date.
  • Stage 3 - Feed the 'structured prompt' into Google Gemini (in Canvas Mode). This will generate a live set of 'interactive' slides.

Live Demo - Building Quick Prototype Websites with Bolt

Demo - If we have time

https://optimisehome.co.uk/

Deep Research & Financial Insight Lab

  • Deep Research
  • Financial insight
  • Google LMNoteBook
  • Produce graphics Via Google Veo

What is an AI Agent?

A standard AI Chatbot is a tool. An AI Agent is a system. You give it a goal, and it works autonomously to achieve it by planning, reasoning, and using other digital tools.

How Does an AI Agent work?

An AI Agent operates in a continuous loop: GOAL → PLAN → ACTION. It repeats this cycle until the final goal is met. You have essentially delegated not just a task, but an entire outcome.

The AI Agent Marketplace

OpenAI: The GPT Store

A public marketplace with thousands of custom "GPTs" built by the community and partners.

Microsoft: Copilot GPTs

A library of custom AI assistants designed for business processes, deeply integrated into the Microsoft 365 ecosystem.

Google: Gemini Gems

A feature allowing you to easily create your own personalised versions of Gemini, tailored to your specific needs and workflows.

Agent Examples

Example marketplace Agents;

  • Creative writing Coach - read your work and give feedback
  • Book Magic - Become a published author - Help you write and publish your own book
  • Code - Review software code, write test cases
  • Whisper Transcriber - Real time audio transcription expert

Agents built by Agentimise

We're creating a marketplace of AI-powered virtual executives (vCEO, vCFO). They are available 24/7 to guide, analyze, and even execute plans. It's like having access to a high-performing leadership team on demand, and without full-time cost.

Build Your Role Assistant

What is a Digital Employee:

A Digital Employee is essentially an AI Agent designed to function like a virtual team member within a business. It goes beyond a simple tool by autonomously planning and using other software to achieve a specific goal you delegate to it, such as 'manage my inbox' or 'generate weekly sales reports'.

Example Link: Typetone AI

What do Human Employees do?

As AI takes over the 'doing', the most valuable humans will shift from doers to directors. The new core responsibilities: Validate, Govern, Manage Ambiguity, and Own the Outcome.

Employee JD's and Scorecards / AI First initiative

Do you actually know the tasks performed by your employees e.g. Role X performs activities A, B and C. Introducing AI Agents will be much harder without a solid grounding of each role within your business.

AI-Powered 'Delegate to Elevate'

By delegating the tasks in the "Drudgery Zone," we can systematically delegate them to AI agents, freeing up human talent to triple the time they spend in their 'Genius Zone'.

Love ItDislike It
Great AtGenius Zone (60%)Drag Zone (20%)
Not Great AtDistraction (20%)Delegated to AI

Human Employees of the future

In - Managing AI Agents and Digital Employees
Out - Managing Human employees

Lean AI Leaderboard

The Challenge - First company to $1B revenue with 1 Employee (The Founder)

Eleven Labs - $100,000,000 ARR - 50 Employees - $2M per employee

Lovable - $30,000,000 ARR - 18 Employees - $1.66M per employee

https://leanaileaderboard.com/

Recruitment - AI First

Recruitment Freeze / Digital Employees

Example AI Powered Customer Support is better than Humans: Learns quicker, knows more, investigates faster, can answer in real time with a human voice. All round better service for customers (24/7) than humans (7/5).

Open Discussion

What would you like to talk about?

The End / Thank You

Conclusion: The Future of Work is a Design Choice, Not a Prediction

The integration of AI is catalyzing a decade of high-velocity change into the next 12-18 months, and the decisions we make now will define the next decade of business.

The Central Leadership Decision: Automation vs. Augmentation

  • Redesign Work: Actively identify which tasks can be delegated to AI to elevate your employees to higher-value work.
  • Invest in Skills, Not Just Tech: Focus on training your team for the new core responsibilities: validating, governing, and directing AI.
  • Lead the Change: You must personally drive this transformation; it cannot be delegated to the IT department.

The Bottom Line: The collective decisions made by leaders like you will determine whether this powerful technology becomes a tool for human empowerment or simply a force for displacement.