AI interviews and expert systems
We build AI interviewers that ask the questions your experts would ask, record the answers as structured requirements, and run on rules your experts edit themselves. AI does the predictable part. People make the decisions.
Where it fits
Already built
Your professions, the questions your experts ask, the sections your requirements need and the rules for what counts as an answer, configured in Structed's back office and tested on simulated clients built from your real cases.
Interview links go out from your own flows, and finished requirements land where your team works: your CRM, helpdesk or project tool, through the API, webhooks or MCP.
New domains and dozens of specializations go in without rewriting the core: domain knowledge lives as versioned data, not hard-coded logic.
Structed's subscription is billed by Structed under its own plans. Setup and integration are fixed price from us. When your conversation doesn't fit Structed's shape, we build a custom interviewer: see the packages below.
How it works
In the client's words, not your jargon. Long forms get abandoned; a conversation gets finished.
"I don't know" and "I'll send it later" are recorded, not asked again and again.
Every requirement points back to the moment the person said it. Nothing invented, nothing quietly dropped.
What the agent asks lives as versioned data experts edit, not as code a developer has to change.
Reasoning, speed and cost tuned per stage, any major model through one router, swapped without a rewrite.
A separate layer treats what clients type as data, never as instructions to the agent.
Test personas, written from scratch or copied from a real interview with the AI's mistakes; golden sets; and a human check on the AI judge.
Results can be reviewed before the client sees them, and anything that can't be undone waits for a yes.
Packages
We interview your experts and write down what they ask, in what order, and how they decide. Useful on its own: it is the onboarding doc nobody had time to write.
An agent that runs the conversation your expert would run, by link with no account, and ends with structured requirements your team and your AI tools can work from.
The people who own the knowledge change what the agent asks and knows. They edit in a sandbox, test on simulated clients, publish, and roll back in one click.
For an agent that works in the demo and breaks with real users. Traces on every run, a test set built from your real failures, an automatic check before each release, and a person in the loop where a mistake costs money.
In my daily work I change the agent's behaviour myself: I set up a sandbox, test there and publish to production.
Implementing our CMS, which empowered technical writers to update in-app integration docs independently, saving engineers countless hours.
Model and hosting costs are billed at cost, without a markup, and estimated per conversation in the plan.
Proof
16 professions and their specializations live as versioned data, not code, so a new domain doesn't need a rewrite. Experts edit the taxonomy in a sandbox, test it, and publish to production without a developer.
Check it yourself
Questions
No. It takes the repeated questions off them, so their time goes to decisions. The experts decide what it asks, and they change it themselves.
Every answer traces back to the conversation, a person can review results before the client sees them, and the case that went wrong becomes a test that runs before the next release.
Any major model, through one router. We pick per stage on speed, cost and quality measured on your own cases, and you can switch later without rebuilding.
We agree on it before the build starts: your cloud or a hosted setup, which model providers are allowed, and how long conversations are kept.
Yes, when the work combines AI with expert judgement: triage, review, planning. If a plain rule or a ready-made tool would do, we'll tell you that instead.
Start with Structed, not from scratch