FAQ
Every answer, before we talk.
Budget, timelines, data, tools: the questions we’re asked most, sorted by topic. Another question? A call is enough.
Getting started
Do we need to change tools?
No. Automations connect to your existing setup: messaging, CRM, management, calendar, spreadsheets, industry software and in-house tools. They work inside them, where your teams already work. The system adapts to you, not the other way round.
How long before it’s running?
It depends on the process and the tools to connect, so we won’t give you a deadline off the top of our heads. Scoping sets the scope and the schedule. We start with one specific flow, tested on your real cases, before expanding to other tasks.
What do we end up with, concretely?
A system running in your tools, the access and the rules, and a team trained to use it. The system belongs to you. And you keep the same contact afterwards.
Do we need to code, or an in-house tech team?
No. For a tailored system, design and integration are on our side, then training happens on the delivered system, with your real cases: reading what the system did, correcting a rule, adding a case. For training on its own, we start from the tools you already have and your own tasks: no technical skills needed.
Do you work outside Lille, in Paris for example?
Yes. Kalhyps is an AI agency based in Lille that travels to Paris and across France. Remotely, we work in France and internationally.
Data and control
Where does our data go, and who can access it?
The system uses permissions limited to its task. Depending on the solution, data is sent to an external AI service. Processing locations, access and retention are agreed with you during scoping. Your data is not used to train a model.
Connection and security, in detailWhat can happen without my approval?
You define the permissions and rules with us. A routine action can be authorised in advance. You choose which actions require human approval. Business decisions stay with your team.
What happens if the system gets something wrong?
Anything that commits you goes through your approval: someone on your team checks the action before it goes out, which limits the risk of a mistake having consequences. Actions authorised in advance are recorded in a log: you can see what happened, and the rule at fault is corrected. A kill switch also lets you pause the system at any time, for example while a rule is being adjusted.
Choosing the right solution
How is this different from a ChatGPT subscription?
A subscription gives you access to a tool. Kalhyps helps you put it to use in your business. In training, your teams learn to use it on their own tasks, and to know its limits. In a project, we design and integrate a system tailored to you: connections to your tools, action rules, tests on your files and training for your team. Either way, we start from what you already use, including your AI assistants.
What if we already use automations?
We keep what works. An automation follows fixed rules: same input, same result. AI comes in when a step takes understanding: reading a document, finding information, preparing a reply. The choice depends on your needs and your tools.
And if an off-the-shelf tool already does the job?
We say so. If existing software solves the problem, it’s better to use it than to build. Scoping is also for that: we only build what doesn’t exist, or what has to connect to your tools.
And if AI brings nothing on a given topic?
We say so. Scoping is first about finding where AI is useful, and where it isn’t. Sometimes a simple rule or better use of your tools is enough, and that’s perfectly fine.
Budget and support
How is the budget determined?
The budget depends on the task, the tools to connect and the support required. The quote separates implementation, training and, for a delivered system, monthly support. Recurring costs are made clear: AI service usage, hosting and tool subscriptions. You know the scope and these cost categories before committing. As a guide: an audit from €500 excl. VAT per day, free if you go ahead with the project that follows, and training from €500 excl. VAT per day. The starting price of each offer is on the Pricing page.
Which AI models and hosting should we choose?
A model is the engine that lets the system read, understand or write. We choose it based on your needs, budget and data sensitivity. An external service, a European provider or a model hosted on dedicated infrastructure: we assess the options during scoping. The provider’s origin alone is not enough: we also check where data is processed and retained.
How does the system improve?
Results are tested on your real cases. Your feedback is used to correct the rules and procedures. Each improvement is checked before the system’s autonomy is expanded.
What happens after implementation?
You get the access and the rules, and your team is trained to use the system. We agree who monitors it, how to report a problem and when to step in. Monthly support covers every system we deliver: updates, maintenance and, depending on the plan, small improvements. It starts at €149 excl. VAT, depending on how many automations and agents are covered.