Preparation: getting ready for AI
Preparation is the first stage of the AI-Native Roadmap. It costs little, needs no new software and decides whether everything after it works. Four things get done: culture, processes, data and a readiness check.
Get ready for AI before you buy anything. Tell staff what AI is for and what stays with them, write down the processes you want to automate and tidy the data those processes use. Then run a readiness assessment. Preparation is cheap, needs no new software and decides whether every later stage works.
How do we get ready for AI before we buy anything?
Get ready by fixing four things that cost little and need no new software: what staff believe AI is for, how your processes are written down, where your data lives and an honest readiness check. Preparation is the first stage of the AI-Native Roadmap, and it decides whether every later stage works.
Buying a tool first is the common mistake. The tool arrives, nobody knows which job it is for and it quietly goes unused. Research from RAND Corporation (opens in a new tab) found that by some estimates more than 80% of AI projects fail. The first root cause it names is that people misunderstand, or miscommunicate, the problem the AI is meant to solve.
Preparation fixes that before any money is spent. It also sets up the goal the whole roadmap works towards. Hartz AI's definition, used on every page of the roadmap, is this:
An AI-native business is one where AI assistants, department brains and agents do the repetitive work in every team, with people keeping the judgement.
Preparation does not need to cover the whole organisation at once. It needs to be done for the first team that will pilot an AI assistant in Phase 1.
What should we tell staff about AI?
Tell staff plainly what AI is for and what stays with them: AI does the repetitive work, and people keep the judgement. Then tell them which tools are approved, because many are already using AI without asking.
A Censuswide survey for Microsoft (opens in a new tab) found that 71% of UK employees have used unapproved consumer AI tools at work. Banning AI does not stop this. A short policy and an approved tool do.
Most staff have also had no help. Government research by DSIT (opens in a new tab) found that 84% of UK workers have had no AI-related training in the past 12 months. Charities are exposed too: the Charity Digital Skills Report 2026 (opens in a new tab) found that 51% of UK charities do not have an AI policy.
Three messages for the first staff briefing
- AI takes on the repetitive work, so people have more time for the work that needs them.
- A person checks every AI output before it reaches a customer, a member or a decision.
- These are the approved tools, and this is the information that never goes into any AI tool.
Involve staff early rather than announcing a finished plan. The CIPD (opens in a new tab) says organisations that involve employees early and create space for dialogue are more likely to succeed with AI. Its research also found that 31% of employers worked on a generative AI policy in the past 12 months (opens in a new tab), up from 16%.
Hartz AI's team AI training runs this briefing as a practical session, with the approved tools open on the screen.
Why write processes down before automating them?
Write the process down before you automate it, because AI can only follow steps that exist. A process that lives in one person's head cannot be handed to an assistant, checked by a colleague or measured in a pilot.
The document does not need to be long. One page per process is enough: what starts it, each step in order, who does each step, what information each step uses and what a good result looks like. This is a standard operating procedure, written in plain words.
Take membership renewals at a sports club. Written down, the process shows that three volunteers each keep their own list, reminders go out at different times and nobody owns the follow-up. Fixing that is worth doing before any AI is involved. Once fixed, it is an obvious first automation.
A process written down for Preparation becomes the brief for Phase 1. It tells you which steps an AI assistant could draft, which a background automation could run and which must stay with a person.
What does AI-ready data look like?
AI-ready data is organised, findable and current, not perfect. Each kind of document lives in one agreed place in the cloud, files follow one naming pattern and the latest version is the one everyone uses.
- Structured: customer, member or donor records in a system rather than scattered spreadsheets.
- In the cloud: shared drives and business systems the AI tools you approve can connect to, not files on a laptop.
- Kept current: a regular cycle for uploading new documents and retiring old ones, so the AI never quotes last year's price list.
- Meetings captured: notes or summaries saved where the team can find them, because decisions made in meetings are data too.
Decide who can see what before any AI tool connects to your data. An AI assistant sees only what its connections allow, so a shared drive where everyone can open the HR folder becomes a problem the moment AI can search it. Hartz AI scopes those connections in discovery.
A synagogue office is a good example. Its records of members, festival rotas and room bookings are often spread across email, a spreadsheet and a paper diary. Bringing them into one system is unglamorous work, and it is exactly what makes a department brain possible later.
Data protection rules apply from the first day. The AI governance guide sets out the UK GDPR questions to settle before personal data goes anywhere near an AI tool.
How do we know we are ready for Phase 1?
You are ready for Phase 1 when a readiness assessment says so. It checks culture, processes, data, skills and governance for the team that will pilot first, and names the first workflow worth trying.
The checklist below is the exit gate from Preparation. If you can tick every line for one team, that team can start a Phase 1 pilot.
- Staff have been briefed and know which AI tools are approved.
- A short AI policy says what information never goes into an AI tool.
- At least one repetitive process is written down, step by step.
- The data that process uses is in one place, current and access-controlled.
- Someone owns the pilot and has agreed how success will be measured.
The free AI readiness checklist takes 20 to 30 minutes. For a fuller review with recommendations, AI consultancy runs the assessment with you. Then move on to Phase 1: AI assistants.
Preparation is the cheapest stage of the AI-Native Roadmap and the one most often skipped.
Common questions
Frequently Asked Questions
Get your first team ready for AI
A discovery call works through culture, processes and data with you, and names the first workflow worth piloting.