Free demo
Working demo
The checking exercise a refresher runs, on four sets of sample notes.
Open the Notes to Record demoRead more
Phishing sessions and tabletop exercises sit beside our cyber security reviews.
See the cyber security reviewsBuild it together. Keep improving it.
What we do
A workshop should leave something useful behind.
Good prompts get lost in chat threads. A chatbot can become out of date when nobody owns its instructions or reference material. Teams need a way to turn experiments into repeatable work.
We agree a task and platform, prepare suitable sample material and work alongside your team. A scoped session can produce a starter custom skill or chatbot with business instructions, approved reference content and test questions. We practise updating it and leave an owner, boundaries and a handover. Live integrations, wider rollout and ongoing support are scoped separately.
From a useful task to a repeatable tool
Choose a workshop around a defined task. We can build a starter with you, teach your team to create and maintain it, or combine both.
Custom AI skills for your team
Turn a recurring task into reusable instructions, examples and checks for your approved AI tools.
- Choose one task and define the expected result
- Write the instructions, output format and boundaries
- Test with ordinary cases and exceptions
- Name an owner and a review date
A chatbot shaped around your business
Build a scoped starter together around approved business material and a clear purpose, such as finding guidance or preparing a draft.
- Agree the platform, audience and permitted information
- Configure instructions and approved reference material
- Test supported answers, missing information and access boundaries
- Teach the owner to update and retest it
- Agree handover and what is required before wider use
Learn to build and maintain it
Coach your team through creating and improving its own assistant.
- Hands-on configuration with your team
- Test questions and a record of changes
- Review permissions and knowledge-source updates
- Guidance for maintaining the starter
AI tools refresher
Cover what has changed in your approved AI tools since the last session, and what each change means for your rules, your checks and your team's work. Practise the new features on current tasks, then update the team's written guidance before the session ends.
- A change log of new features, model releases, agents and connectors in the tools you have approved
- Each change read against your acceptable-use policy and data rules
- Prompt injection and connector risks explained with the OWASP Top 10 for LLM Applications
- A checking exercise on sample data: a prepared record tested against its source
- Updated one-page guidance for the team, dated and versioned
Prompt library clinic
Review the prompts, saved instructions and custom assistants your team shares, retire the ones that no longer give reliable results and add the ones that do. Give every entry an owner and a last-checked date, kept in the document or workspace tool your team already uses.
- An inventory of shared prompts, saved instructions and custom assistants
- A retest of each entry against a worked example, with the result recorded
- An owner, a last-checked date and a data-handling note on every entry
- Retired entries archived with the reason they were retired
- A template for proposing new entries
AI champions programme
Train a small group of people across your teams to answer first questions, collect problems and useful examples, and bring them to the next refresher. Give them a clear route to escalate what they cannot answer, including anything that touches personal or client data.
- Champion sessions ahead of each refresher
- A champions' guide: first answers, when to escalate and where the rules are kept
- A shared log of questions, problems and examples of work
- An escalation route to your policy owner and your data protection lead
AI office hours
Hold a regular drop-in where anyone brings a task and works it through with a trainer, on your approved tools and within your data rules. Log what keeps coming up and feed it into the prompt library and the next refresher.
- One-to-one and small-group help on current tasks
- A log of recurring problems and useful examples
- Candidate entries for the prompt library
- Topics carried forward to the next refresher
What you get
What the team keeps between sessions
- A scoped starter skill or chatbot, or the capability to build one, as agreed for the session.
- Instructions, approved sources and test cases your team can maintain.
- A named owner and a handover that distinguishes workshop work from production rollout.
A sample
Inside a business-specific skill
An illustrative workshop artefact using sample information. The actual task and platform are agreed with your team.
Sample, on sample data
- Purpose
- Prepare a first draft of a customer reply from the approved service policy.
- Instructions
- Use only the supplied policy. Identify missing information and do not invent a commitment.
- Reference material
- A sample policy and examples approved for the exercise.
- Output
- A draft, the sources it relies on and the questions for the reviewer.
- Checks
- Test a routine request, a missing answer and an instruction that conflicts with the policy.
- Owner
- The named team owner updates the material and repeats the checks before sharing a change.
A starter assistant is not a guarantee of accuracy or production readiness. Access, data protection and any live integrations need the agreed review.
How it runs
How a workshop runs
- 01
Scope the task
Agree the purpose, audience, platform, time available and what a starter can reasonably include.
- 02
Prepare the material
Choose approved sources and sample cases; agree the access and data rules.
- 03
Build and practise
Create the instructions or starter together and teach the team how it works.
- 04
Test and hand over
Try routine and difficult cases, record the limits and teach the owner how to maintain it.
Try
Open the demo. The research is in preparation.
In preparation
Practical AI training: building capability that lastsIn preparation
Practical AI training: building capability that lasts
A guide to planning AI training around the tasks people do, their confidence and the responsibilities that come with new tools, including how learning continues after a first session.
Working demo, on sample data
- Four sample notes, from a project meeting, a client call, a site visit and a supplier review, prepared as records with decisions, actions, owners and open questions.
- Statements the source does not support, such as a condition written as a decision or an action given to the wrong owner, flagged beside the numbered source lines.
- Filing held until each flag is edited, removed or kept with a reason, then the record filed with its corrections shown. The drafts are scripted and no language model runs.
Worth asking first
- Which task should the starter help with?
- Which platform and accounts are approved?
- What knowledge may it use, and who can access it?
- Who will maintain it after the workshop?
Talk to us if
- Your team wants a business-specific chatbot or reusable skill.
- You want to learn by building together.
- Shared prompts and assistants need an owner and regular checks.
- People need follow-up help applying what they learnt.
Questions
Questions buyers ask
Can you build a chatbot during our time together?
Yes, where the agreed task, platform and available material fit the session. We scope a starter in advance and can build with you, teach your team to build it, or both. We do not promise an unspecified production system within a session.
Is this training a new AI model?
Usually it means configuring an existing assistant with business-specific instructions, approved knowledge and tests. Fine-tuning a model is a separate technical choice, not something every business chatbot needs.
What happens after the workshop?
We agree an owner and handover, including how to update instructions and repeat the checks. Follow-up clinics or refreshers can be arranged. Hosting, licences, live integrations and ongoing support are agreed separately.
Can we use confidential business material?
Only through a route approved for that material. We prepare sample information for exercises where live records are unsuitable and agree access before connecting business sources.
Make something your team can keep using.
Bring a recurring task. We will agree a practical workshop and what you can take away from it.
Discuss a custom AI workshop