AI consulting & implementation from readiness check to daily team use

Consulting answers the question of what you should do: which processes AI pays off in, which tool fits, what rules are needed and how to train the team. Building an agent, a prediction model or an automation is the execution stage, and it is handled by our separate services. The split matters because much of the value comes from off-the-shelf tools that already exist, and sometimes the honest answer is that nothing needs to be developed. If the mapping shows that a custom solution is worth building, you receive an initial specification that you can take to another supplier as well.

Readiness comes down to four things: whether processes are repeatable and written down, whether data is accessible rather than scattered across WhatsApp messages, whether one person owns the subject, and whether management is willing to try something small first. You do not need an IT department or a huge dataset. In the readiness review we score each area and say what is missing, and sometimes the first recommendation is to tidy documents or name an owner for a process, and only then pick a tool. You can get a first picture at no cost with the /tools/ai-readiness-checker tool on our site.

No single tool wins every task, and the choice depends on your working environment and the kind of work. An organisation living in Microsoft 365 gets value from Copilot, one on Google Workspace from Gemini, and teams that write and analyse a lot of Hebrew text should test ChatGPT and Claude closely. So we do not recommend from online rankings. We run the same five to ten of your real tasks through each tool and compare quality, speed, user management and privacy terms. The result is a reasoned recommendation, and quite often a combination of two tools rather than one.

Yes, even for a small business, because without one each employee decides alone what to paste into an external tool: a contract, a client list, payroll data. A good policy is short and practical, usually a page or two: which tools are approved, which types of information must never be entered, when human review is mandatory before something goes to a client, and who to ask when in doubt. It also helps when a business customer or a public body asks how you use AI. We draft it with you and fit it to the tools you chose; binding legal advice should come from a lawyer.

Before the pilot we pick a single metric and measure it as it stands today: how long a quote takes, how many enquiries are answered per day, how many errors are caught in review. After a few weeks we compare against the same metric. We do not rely on impressions such as "it feels easier", because those do not justify licences. The calculation also includes costs that are easy to forget: the time spent on human review, licences and training, so the saving is not one that exists only on paper. If the gap is too small, we will tell you to stop or try a different process.

It depends on the tool and the plan. On business plans such as ChatGPT Team or Enterprise, Claude for Work or Gemini inside Workspace, vendors generally commit not to train their models on your data, but that is not the case for free personal accounts. So the policy requires company accounts only, removal or coding of sensitive details, and a list of information types that must never be entered at all. On Amendment 13 to the Israeli Privacy Protection Law we document what is entered and where it goes, and in sensitive cases we refer you to a privacy adviser.