Entry, Standard, Enterprise: three steps from proof of value to managed automation

Most AI projects get too big too early. They start with strategy, vision, architecture and workshops. All important. But nothing is in production. lights-out.ai is built the other way round.

Entry: see if it works for you

Entry is the first step. The goal is not to transform the whole company. The goal is to pick one important workflow and build a working solution in 4 to 8 weeks.

It could be a product data view that replaces manual consolidation. A wholesale view that replaces weekly reports. A purchasing view that makes forecasting more direct. Or an operations workflow where the team currently pulls data from multiple systems to make a single decision.

Entry is a command centre. It reads from a limited number of source systems, shows context, can write back where it is safe to do so and gives the team a first experience of working with owned AI software.

The important thing is that Entry is not a demo. It should be in production.

Standard: replace SaaS where it already hurts

Once Entry has proven value, the next question follows: which standard systems or manual layers can be replaced?

Standard is for situations where CRM, PIM, BI or internal admin tools no longer fit how the company works. Not because standard systems are bad, but because certain processes have become too specific, too cross-functional or too expensive to force into generic tools.

Here lights-out.ai becomes more than a workspace. It becomes a custom-built system with safer data handling, more integrations, more reliable operations and clearer process ownership.

Enterprise: replace work, not just systems

Enterprise is when the company moves from replacing tools to replacing recurring tasks.

This can involve order and return workflows, purchasing processes, product data enrichment, customer service preparation, report generation or other operational logic. Here agents and workflows work together, with clear mandates, logging, human oversight and SLAs.

Enterprise requires more governance. That is why it fits best once the first workflows are proven and the company knows which parts of the work should actually be automated.

Avoid the two most common mistakes

The point of three tiers is to avoid two common mistakes. The first mistake is starting too small: a chatbot, an experiment or a demo that never reaches the business. The second mistake is starting too big: a transformation programme that tries to design the future before anyone has seen what works.

Entry, Standard and Enterprise is a path between the two.

First we prove the value. Then we replace what no longer fits. Finally we automate work where the process, the data and the accountability are mature enough.

This is not AI as a project. It is operations development with AI as the tool.