Foundation · 1 day
Practical Generative AI at Work
A working day for knowledge workers and business teams: using generative AI for research, writing, analysis and document work, with enough discipline to trust the output.
1 day · Foundation · Foundations · Knowledge workers and business teams
Overview
Generative AI is already in daily work, often without a shared method. People paste confidential text into consumer tools, accept fluent answers as evidence, or abandon the technology after a few disappointing attempts. This course is for you if you have to use it on real documents, real numbers and real decisions.
The day is organised around work, not around a catalogue of tips. You practise research, summarisation, writing, analysis, brainstorming and structured extraction, then spend as much time on verification as on generation. Hallucination is treated as an expected property of the system, not as a rare defect. Privacy and confidential information are handled as operating rules.
The course is deliberately not a tour of a single product’s interface. You learn a method you can take back to whatever approved tools your organisation uses: how to specify the task, constrain the output, check the result, and decide when not to use a model at all.
By the end of the day, you can use generative AI as a serious work tool — faster on the tasks it is good at, slower and more careful where the cost of a fluent error is high.
Audience
Knowledge workers and business teams: analysts, operations, research, communications, programme managers and specialists who work with documents, data and decisions.
Prerequisites
No technical background is required. Bring typical work examples where possible, and use an approved organisational AI tool rather than an unmanaged consumer account if one is available.
Duration
1 day
Can be delivered as a half-day for a single function, or extended with domain-specific exercises.
Learning outcomes
- 01
Design prompts for research, writing, summarisation and analysis that specify task, audience, constraints and output shape.
- 02
Extract structured information from documents and meetings into tables, checklists and action lists.
- 03
Verify generated output against sources, and identify claims that need evidence before they can be used.
- 04
Handle confidential and personal information according to organisational rules rather than convenience.
- 05
Use generative AI on spreadsheets and data with enough scepticism to catch invented figures and false structure.
- 06
Decide when not to use a model, including legal, numerical, personal and high-stakes work.
- 07
Build a small set of reusable work patterns for recurring tasks in your role.
Outline
- 01
A working method, not a set of tricks
- What current systems do well in knowledge work, and where fluency is a liability.
- A repeatable pattern: purpose, source, constraints, output, verification.
- The difference between drafting support and an authoritative answer.
- Exercise: take a real request from the group and make the task spec explicit.
- 02
Research, reading and summarisation
- Briefing, comparison and synthesis from source material rather than from the model’s prior.
- Summaries for different audiences: executives, specialists, and people who were not in the meeting.
- Citations, quotations and the habit of sending the reader back to the source.
- Exercise: summarise a long document, then mark every claim that is not grounded in the text.
- 03
Writing, analysis and structured output
- Drafting emails, papers, plans and options without handing over the decision.
- Brainstorming that produces alternatives rather than a single polished paragraph.
- Structured outputs: tables, scores, risks, options and action lists.
- Spreadsheets and data: useful transformation versus invented numbers.
- Exercise: turn unstructured notes into a structured brief, then check it against the original.
- 04
Verification, privacy and when not to use AI
- Hallucination, omission and confident error as expected behaviours.
- Checking facts, figures, names, dates and legal or policy statements.
- Confidential information, personal data and the difference between approved and unmanaged tools.
- Work that should stay human: high-stakes decisions, original judgement, and anything requiring authority.
- Exercise: review a fluent but unsafe output and write the rule that would have stopped it.
Practical work
The day is a sequence of workplace exercises on your own material where possible: research and briefing, document summarisation, writing, structured extraction, and a verification pass. Each exercise ends with a judgement about whether the output is usable as-is, usable after checking, or not to be used.
Takeaways
- Reusable prompt patterns for briefing, writing, extraction and analysis
- A verification checklist for generated work
- Rules of thumb for confidential information and approved tools
- A short set of role-specific work patterns drafted during the day
Delivery
Cognitecta delivers private corporate training, on-site or as remote live training. Courses can be run as published, or adapted to your organisation’s stack, domain and experience level.
Instructor
Nicholas Johnson, AI architect and software engineer. He has a degree in Artificial Intelligence and around twenty years of professional technology training, including hundreds of courses for engineering teams and large organisations. About.
Related courses
1 day · Foundation
Generative AI Fundamentals
A technically accurate, practical introduction to modern generative AI for mixed technical and non-technical audiences. Enough depth to make sound decisions, with no requirement to write code.
1 day · Intermediate
How Large Language Models Work
A one-day explanation of LLM internals for developers, architects and technically curious professionals. Enough mechanism to improve engineering decisions, without a research-paper prerequisite.