Adaptive collection

Two ways to collect

Every AI product built on Productised collects information through a structured conversation. You define what needs to be collected — the fields, the goals, the order. The AI carries out the collection. But there is an important choice about how that collection happens.

Structured collection works through your collection goals in order, one at a time. The AI asks a question, records the answer, and moves to the next goal. Every person gets the same questions in the same sequence. The experience is consistent, predictable, and controlled. You always know exactly what data will be in the record at the end.

Adaptive collection works through the same goals in the same order — but when an answer is vague, incomplete, or contradictory, the AI asks one brief clarifying follow-up before moving on. A person says "yeah I think so" to a question about their current revenue. In structured mode, that becomes the recorded answer. In adaptive mode, the AI notes the ambiguity, asks one targeted question to resolve it, and records the clearer response.

The same collection goals. The same field structure at the end. Better data quality along the way.

Why adaptive is not chaos

The natural concern is that giving the AI latitude to ask follow-up questions will result in an unpredictable, open-ended conversation that wanders off-topic. This is not how adaptive collection works.

The AI has one specific task: when an answer is insufficient to record confidently, ask one follow-up to resolve it. One follow-up. Then it moves on, regardless of whether the follow-up fully resolved the ambiguity. The AI does not invent new collection goals, does not explore tangential topics, and does not extend the conversation beyond what the collection requires.

Adaptive collection does not change what you collect. It changes how reliably you collect it.

The structure is preserved. The goals are preserved. The field schema at the end is identical to what structured mode would produce. What changes is the signal quality of the data inside it.

When to use Structured

• Simple lead capture where you need a clean, fast experience
• Quick qualification products where the questions are unambiguous and the answers binary
• Products where brand consistency is the priority — every person should have an identical experience
• Any product where you need to minimise conversation length and maximise completion rate

When to use Adaptive

• Diagnostic products where the quality of the collected data directly affects the quality of the outcome — a vague answer to a critical question can undermine the entire report
• Scorecard tools where scores are computed from collected values — a poorly recorded answer skews the score
• Products that collect qualitative data like "describe your biggest challenge" or "walk me through your current process" — these are the answers most likely to be incomplete on first pass
• Anything where the outcome is a high-value personalised deliverable that will be shown to a real client

If you are building a product whose output you would be proud to show a client in a discovery call, use adaptive. If the collection is a gateway to the outcome rather than the value itself, structured is fine.

Where to set it

The Collection Style setting is on the Content tab of the AI Product node in the canvas sidebar. Select Adaptive or Structured from the dropdown. Adaptive is the default.