✦ AI consulting · Updated 2026-10-26
How to implement AI in your business without wasting money.
Implementing artificial intelligence is not buying licenses for everyone: it is choosing a few tasks where it genuinely saves time, testing them with a small group, measuring and only then scaling. That is how I do it with companies and small businesses.
At a glance
- First step
- Assessment of tasks, data and current tools
- First result
- A pilot with 2 or 3 measured use cases
- What gets measured
- Time saved, quality and errors avoided
- Main risk
- Customer data in uncontrolled tools
- Who it is for
- Small and large businesses in any industry
The 6 steps to implement AI
- Assessment: which tasks take the most hours, what data is involved, which tools you already pay for (Microsoft 365, Google Workspace, ERP, CRM) and who already uses AI on their own.
- Use cases: a short list of candidate tasks, ranked by impact, effort and risk. The best ones are repetitive, text-heavy and have a person reviewing the output.
- Tools: choose the platform that fits your ecosystem and buy a business plan, not personal accounts.
- Rules: a short use policy stating which tools are approved and which data is never uploaded.
- Pilot: two to four weeks with a small group, measured against how the work was done before.
- Scale: train each team with its own examples and roll out what worked.
Use cases that usually work first
- Drafting and answering emails, proposals and communications in the company's voice.
- Summarizing meetings, contracts, tenders or long documents.
- Customer service: draft replies and FAQs, including over WhatsApp.
- Back office: extracting data from invoices and receipts into a spreadsheet or system.
- Sales and marketing: customer research, social media content and pipeline follow-up.
- Spreadsheets and analysis: formulas, data cleanup, tables and explanations in Excel or Google Sheets.
Common mistakes
- Buying licenses for everyone before knowing what they will be used for.
- Letting everyone use their free account with customer data.
- Starting with the most ambitious project (an agent that does everything) instead of a concrete task.
- Not measuring: without a baseline there is no way to know whether AI saved time.
- Forgetting human review: AI makes mistakes confidently, and that has to be expected.
What implementing AI costs
It depends on how many people will use it and whether it needs to be integrated with your systems. For many companies the first step costs less than expected: use the AI already included in Microsoft 365 or Google Workspace, train the team and automate a couple of processes. Vendors change business plan prices often, so we quote them at current rates.
→ How I help
Related terms
- Assessment of tasks, data and tools, including AI already used without control.
- Use case prioritization by impact, effort and risk.
- Platform selection and secure configuration.
- A pilot with metrics and a roadmap to scale.
? FAQ
How long does it take to implement AI in a company?
A first pilot with measurable results can be running within a few weeks. Rolling it out company-wide depends on size and on how many integrations are needed.
Do I need a technical team to use AI?
Not to get started. Generative AI tools work from the browser or inside Office and Google Workspace. Integrations with your systems or agents do require technical work, which I can coordinate with your team or provider.
Will AI replace my employees?
In practice, what changes is the tasks: AI speeds up drafts, summaries and repetitive work, and people spend more time reviewing, deciding and serving customers. The companies that benefit most are the ones that train their people to use it.
Sources