Our point of view
Most compliance work is not invention. It is organisation.
Most of a compliance programme is moving what you already know into a shape an auditor will accept. That is not work a person should spend their week on. A machine does it well. This is how we think AI compliance software should work. The agents organise your facts. A person always decides. Give your facts once. The agents draft. You decide.
The name is with pun intended and with a serious point
We train your organisation in compliance - and compliance itself is a train journey. The destination is fixed: a system an auditor will accept. But everyone boards at a different station. Some arrive with binders full of procedures; some with almost nothing written down. Where you get on is your level of maturity, and it is nobody’s fault. The job is to get you moving from wherever you are, without asking you to describe your own company from scratch.
Why “just write it” is the wrong instruction
Most compliance software that added AI did the obvious thing: it put a blank box on the screen and asked the model to write. We think that is the wrong instruction. A blank box invites invention, and invention is the one thing a compliance document must never contain. Our agents are built to do the opposite - to move known information into a known structure, and to mark clearly where a fact is still missing rather than fill the gap with a plausible guess.
One - the information already exists
The great majority of what a standard asks for, you already know. Where you operate, who does what, which supplier carries which risk, what your policy actually says. ComplyTrain’s job is to collect that once and keep it. It ingests from the places the knowledge already lives, into secure, per-organisation vaults, and organises it so the agents can find it by meaning rather than by keyword:
- Your public website
- The documents you already hold
- A spoken conversation or a recorded meeting
- A chat message made in passing
Nothing leaves your walls to answer a question about your own company. Ask about your organisation, and the answer comes from your organisation - from your approved documents and your confirmed facts, not the open internet.
Two - the thinking is done once, by experts
A standard means the same thing for every company that answers to it. So the expensive, expert part - reading the clauses, deciding what evidence counts, laying out the milestones and the templates - should be done once, properly, and reused. Skylen’s compliance experts pre-process each standard into the parts every organisation on it will need:
- The requirements, clause by clause
- The milestones that get you from here to audit-ready
- The facts the standard needs to know about you
- The forms and document templates, ready to be filled
Improve that work once and every organisation on that standard benefits. And the agents that use it are not general chatbots - they are senior subject-matter experts, trained by the same Skylen experts who prepared the standard: always on, and consistent from the first document to the last.
Three
We never ask an agent to “just write”. Every document is built, section by section.
A document here is not one long request to a model. It is a structure - sections, each mapped to the requirements it has to satisfy. For every section, the agent is given a purpose, an expert-reviewed template and your own knowledge, then asked to do one job.
Start from the requirement
Each section exists to answer specific clauses of the standard. The agent begins with those, not with a blank page - so it is always clear what the section is for.
Work inside an expert-reviewed template
The section’s shape - its purpose and the form good evidence takes - was prepared and reviewed by Skylen’s experts for this standard. The agent drafts within that template, not around it.
Ground it in your own knowledge
For the content, the agent draws only on your confirmed facts and your approved documents - your own words about your organisation, retrieved by meaning and handed to the agent as the basis for the draft.
Mark what is not known - never invent it
When a fact is still missing, the draft carries a clearly-marked placeholder and the question comes back to a person. The agent does not close a compliance gap with a plausible-sounding guess.
A person approves
The agent drafts and proposes; only a person accepts, edits or signs off. Nothing an agent writes reaches an approved document without someone deciding it should - a line that does not move.
Four - compliance is training your organisation, and running good processes
A certificate on the wall is not compliance. Compliance is that the people in your organisation know what is expected of them, and that the work runs on processes which hold up when someone checks. So ComplyTrain is not only a document engine. It trains your people - the AI Training module builds and runs the learning your standards require - and it runs the recurring work on a process engine, so reviews, approvals and evidence happen on time and leave a trail, rather than living in one person’s calendar and memory.
Five - tell it once, and everyone works from the same place
Provide a piece of information a single time and it is available everywhere it is needed: to the next interview, to every document that cites it, to every agent that answers a question about it. People and agents share one structured system - the same facts, the same requirements, the same templates. The agent is never handed a blank box; it is handed structured tools and your own knowledge. That is what lets the machine take the busywork while a person keeps the judgement.
The result is a compliance programme that moves at the speed of review rather than the speed of transcription - minutes, not months of typing your own company into someone else’s template.
What this comes down to
Three commitments hold across everything the agents do.
Grounded in your own words
Agents answer from your approved documents and confirmed facts - your organisation’s own record - not the open internet. The knowledge stays in your secure, per-organisation vault.
A person always decides
AI drafts and proposes; only a person approves. No agent can approve, publish or sign off a document, an evidence record or a risk. Every AI-influenced change is attributable and reversible.
Speak to it, or type
Every agent takes dictation as readily as typing. Talk a section through in plain language, or write it - whichever is faster for the person doing the work.
Become enlightened
Learn more - about your own data
You most likely already have a lot of information about your processes, organisation and products. With ComplyTrain you can resurface your data into new templates and re-use the information you have already produced to generate compliance documentation and much more