Almost every useful AI decision system an Australian business builds falls into one of two patterns. A Guide system is customer-facing: it understands what someone needs and matches them to the right service, product, or person. A Decide system is operational: it assesses information and recommends an action, escalating anything unusual to a person. Knowing which one you need first is the single most useful decision you can make before any AI project starts.
Most businesses default to whichever pattern is more visible in their day-to-day. A training provider sees the enquiry backlog and thinks Guide. A maintenance team sees the fault log and thinks Decide. That instinct is usually right, but it is worth checking deliberately rather than assuming.
What is a Guide system?
A Guide system sits between a customer and a complex catalogue of services, products, or people, and does the matching a staff member would otherwise do by hand.
The pattern shows up wherever a business offers more options than a customer can reasonably compare themselves: a professional education provider with hundreds of workshops, an advisory firm with multiple service lines and specialists, a retailer with a large or technical catalogue. The customer knows roughly what they need but not which specific option fits. Without help, they either pick the wrong thing, give up, or land in a generic "contact us" form that a staff member then has to manually triage.
A working Guide system understands the enquiry, checks it against the full catalogue, and returns a short, ranked list of the right fit, with the reasoning attached. The result is a qualified opportunity instead of an unsorted enquiry.
What is a Decide system?
A Decide system takes information, a photo, a document, a technician's notes, a sensor reading, and recommends what should happen next. It escalates anything it is not confident about to a person, rather than guessing.
This pattern fits operational and asset-heavy businesses: maintenance teams assessing equipment faults, operations staff triaging incoming documents or images, any team that currently relies on one or two experienced people to make a judgment call quickly and consistently. The business risk here is specific: that judgment usually lives in someone's head, not in a system, and it walks out the door when that person leaves.
A working Decide system captures that judgment as a working tool. It does not remove the person from the loop. It handles the routine cases at the same standard every time and puts the genuinely unusual ones in front of a person, with the context already assembled.
What does each pattern look like once it's built?
A Guide system in production for a professional education provider works like this: someone submits an enquiry describing what they want to learn or achieve. The system reads it against the full catalogue, hundreds of workshops and experts in some cases, and returns the two or three that genuinely fit, with plain-English reasoning for each. The enquiry that would have sat in a shared inbox for two days now becomes a qualified lead in under a minute, and the person who used to do that matching by hand is free for higher-value work instead.
A Decide system in production for an operations team works differently: a technician photographs a fault, or an intake officer receives a document, and the system assesses it against the business's own technical documentation and past decisions. Straightforward cases get a recommended action immediately. Anything genuinely ambiguous, low confidence, high stakes, or simply new, gets escalated to a person with the relevant context already attached, rather than starting from a blank page.
How do you know which one to build first?
Ask where your business currently loses the most time or the most opportunities to manual judgment. If the answer is "matching people to the right thing we sell," that is Guide. If it is "assessing information before we can act on it," that is Decide.
A useful test: picture the specific person who does this work today. If they mostly deal with people asking what they need, you have a Guide candidate. If they mostly deal with information they have to interpret before deciding what happens next, you have a Decide candidate. Most businesses have a clear answer to this within a few minutes of thinking about it properly.
What do Guide and Decide systems have in common?
Both patterns depend on the same three things being true: there is a real decision being made repeatedly, the business already has the knowledge needed to make it well (even if that knowledge is scattered across documents, spreadsheets, and people's heads), and getting it right or wrong has a measurable cost.
Both patterns are also built the same way at Qode: an AI Decision System Sprint produces a working prototype against real inputs, a business case showing what the system is worth, and a fixed quote for production, before any larger commitment is made. Neither pattern is theoretical by the time you decide whether to proceed.
Which pattern fits your business?
Most Australian mid-market businesses have at least one clear candidate for each pattern, but rarely need to build both at once. Start with whichever one is costing you more right now, measured in missed opportunities for Guide or in risk and inconsistency for Decide, and prove it before moving to the second.
If you can already picture the person and the decision, you are ready for the next step. Book a Decision System Sprint fit call and we will help you work out which pattern to prototype first, and whether it is worth building at all.
