White Cane / AI Consulting Guiding those who walk blind into the era of AI.

It is worth it when there are tasks that repeat, rely on text or documents and take a lot of time: quotes, replies to customers, reports, searching files, meeting notes. It is not worth it when the work is mostly judgement and accountability, or when there is nothing you can measure. If you already pay for ChatGPT or Copilot but only a few people use it, on personal accounts and without access to the company's documents, what you need is the set-up, not another tool. If we don't see a benefit in your case, we will tell you.

One company platform for the whole team is usually enough. On Google Workspace, Gemini is the quickest start. On Microsoft 365, Copilot already sees your files and email. For long documents and precise writing, Claude. For general use and the widest ecosystem of apps, ChatGPT. We add a second tool only if a department has a need the first platform does not cover. Before we recommend anything, we test it on your own Greek documents.

It depends on the plan and the settings. On free and personal plans, what an employee uploads may be used to train models. On a company plan, with a data processing agreement (DPA), training on your data switched off, sign-in with a company account (SSO) and the right access rights, it is safe for most files. For data that must not leave the company at all, we run models on your own infrastructure. We check GDPR before anything real goes in.

We start with the tasks that repeat and pass through many hands: quotes, orders, invoices, customer requests, reports. For each one, AI writes the first draft or extracts the data, and an automation (n8n, Make or Power Automate) moves the data between email, files, CRM and ERP with no copying by hand. We record today's time first, so afterwards it shows whether the workflow really got faster. When a task has several steps and clear rules, we add an agent: a model given tools and permission to act, such as reading a file, querying a database or drafting a reply. It is the wrong choice when judgement and accountability matter more than speed, and then a plain search box serves you better. We will tell you which is which. That is what AI transformation looks like in practice: department by department, measured at every step.

We can. White Cane AI is a team from Athens with experience setting up AI tools in large Greek companies. We choose the tools, set them up, train your team and stay with you for two months. We start with a free 30-minute call, where you tell us what you want to solve and we tell you what we would set up and which package fits. You can ask for it through the contact form.

Not always. Some models handle Greek grammar and business documents worse than English, even when their English output looks strong. We test candidates on Greek text specifically (contracts, invoices, customer messages), and we know the Greek market, so we can tell you upfront which tools are worth trying and which aren't.

Yes. Open-weight models have closed most of the gap for what businesses actually need: drafting, extraction, classification, search across internal documents. What it costs you is hardware and the work of running it, which is why we size the GPU to your real load rather than to the largest model available.

No. We connect to what you already run (ERP, CRM, Notion, Slack, M365, Drive) through MCP servers, with explicit limits on what each agent is allowed to see.

The model is the part we deliberately keep replaceable. Your documents, your access rules, your evaluations and your workflows are the assets, and they outlive any single model. Swapping one out should be a configuration change and a test run, not a rebuild. That is the main reason we avoid architectures locked to one vendor.

It depends on the size of the company and on how many tools and connections are needed. Small business (5–15 people): €1,500 – €2,500. Mid-sized business (20–100 people): €2,500 – €7,500. Large organisation (100+ people): €8,000 – €20,000+. Prices exclude VAT and include research, set-up, training and two months of support.