Add-ons for AI assistants
Teach your AI assistant your own way of doing things
Do you have to explain your templates and rules to the AI assistant again every time? We teach it your own procedures and give it controlled access to your documents. For example, it can write property listings from your own template. That way the answers are the same, whoever asks. The assistant only gets access to what you choose.
- The problem
- Do you have to explain your templates and rules to the AI assistant again every time?
- What we do
- We teach it your own procedures and give it controlled access to your documents. That way it gives the same answers, whichever of your people asks.
- What it costs
- First solution from 15,000 DKK at a fixed price. Operations from 2,500 DKK/mo. All prices excl. VAT. See all prices, and work it out for your own task
What it is and how it works
We teach your AI assistant your own templates, rules and procedures, so you do not have to explain them again every time. A skill is a set of instructions that tells the assistant how to do one task your way. A plugin adds a function to the assistant, and a connector gives it controlled access to one of your systems. Together they let the assistants your team uses work with your own procedures and your own data.
Most AI assistants can already write, summarise and answer questions in a general way, but they know nothing about how your company specifically does things unless someone tells them each time. A skill is a written set of instructions, and often example files alongside it, that teaches the assistant your company’s own way of doing one task, so it does not have to be re-explained in every conversation.
A plugin is a smaller addition that gives the assistant a function it did not have before, such as a particular kind of calculation or a specific output format your company relies on. A connector is different again: it gives the assistant controlled, limited access to one of your own systems, such as a document archive or a customer database, so its answers can be grounded in your company’s own information rather than only in general knowledge.
None of the three replaces the assistant your team already uses. They sit alongside it, adding what it was missing for your company specifically, while the assistant itself stays the same tool people already know how to use.
Where it fits
Architecture firms
The assistant drafts project descriptions and tender texts in the firm’s own structure and wording.
Accounting and bookkeeping
The assistant can look things up in the folders you give it access to, so answers come from your own files.
Real estate agents
The assistant writes property listings from the firm’s template and checks them against its rules.
How you get started
We always work in the same five steps. Here is what they mean for add-ons for AI assistants:
We look at your daily work We find the tasks your team repeats with AI assistants and the systems they need.
We build the first solution We write one skill for one task with the people who do it, and test it on real work.
We put it to work for you We install it in your team’s assistants and connect only the systems it needs, with access you control.
We build on what works We build more, one thing at a time, based on how you used the first.
We look after it We update them when the assistants or your procedures change, and remove what nobody uses.
What it gives your company
Without a skill, the same instructions get typed out again in every conversation, by whoever happens to be using the assistant that day, with small differences each time. With a skill in place, the assistant already knows your company’s own way of doing the task, so the result is consistent regardless of who asked for it.
A connector means the assistant can answer from your own systems directly, rather than a person first having to find the relevant document or record and paste it into the conversation by hand. That step, small as it looks, is often the one that made using an assistant for real work more trouble than it was worth.
As your team’s assistants and your own procedures change, the skills, plugins and connectors around them can change too, rather than becoming a fixed setup that quietly falls out of date while everything around it moves on.
Why Lynggaard Jardim
Lynggaard Jardim is run by its two founders together: Alexander across operations and finance, Victor across development and IT. A skill we write reflects both, how the task is actually done in your company, and how safely it can be connected to your systems.
We are based in Copenhagen, work across Denmark and build each skill, plugin or connector around the assistants and the systems your team already uses, naming no vendor as a partner and adding only the access a task genuinely needs. What we build follows your own procedures; it does not ask your team to change them to fit a tool.
Your data
As far as possible, your data stays with you, so we have no access to it. When it is necessary, we sign a data processing agreement.
An AI agent gets access only to the programs we have agreed with you in advance. Everything it does is written down, so you can see it afterwards. For the most important tasks, it stops and waits until a person has approved it.
The solutions themselves run in the EU, and we never publish your data. Which AI works in the solution depends on the setup. Wherever possible it runs on your own account, for example your own ChatGPT, and then your own terms apply. If you would rather use ours, it runs on our agreement, and then your data is not used to train AI. We follow the rules and your own requirements for data.