OAIRA: Research Methodology as Executable Software

Market research has an expertise problem. Designing a good survey requires trained researchers most teams don't have. Validating survey design before it goes to real respondents requires tools most platforms don't provide. Synthesizing findings across dozens of source documents requires analyst time most organizations can't spare. OAIRA addresses all three by encoding research expertise directly into software.

What It Is

OAIRA is an AI-powered market research platform built on Next.js with a Supabase backend and Claude as the primary intelligence layer. It combines eight professional research methodologies, a pre-deployment simulation engine, an autonomous AI interviewer, an 8-phase deep research pipeline, real-time analytics, and streaming report generation into a unified system accessible via web UI, REST API, and MCP server.

The platform's foundational claim is that methodology is code, not advice. Research rigor is embedded in executable workflows, not documentation that users are expected to absorb before starting. An AI-powered fit scorer evaluates any research goal against all eight frameworks simultaneously and recommends the best match. The selected methodology then guides the researcher through a multi-turn AI conversation that produces a complete, structured survey.

Why It Matters

The simulation engine is the capability most distinct from anything in the existing market. Before deploying a survey to real respondents, teams run it against pools of AI personas — 10 to 500 synthetic respondents with configurable demographics, psychographics, and behavioral profiles. The simulation produces a full synthetic dataset with the same schema as real respondent data. The same analytics pipeline, charts, and methodology-specific analysis apply. A comparison engine flags divergences between synthetic and real responses, identifying confusing or leading questions before they contaminate fieldwork.

This creates a new research pattern: simulate first, deploy second. The cost of a simulation run is predictable and low. The cost of discovering a design flaw after real fieldwork is not.

The autonomous AI interviewer extends this further into qualitative research. Deployed via a public interview link, a conversational AI agent conducts structured, open-ended interviews following a research brief. It tracks question coverage, probes vague responses, and transitions gracefully between topics — wrapping up when coverage thresholds are met and extracting structured responses from the conversational transcript.

For research synthesis, the 8-phase deep research pipeline ingests uploaded source documents, performs semantic search via pgvector, extracts claims with evidence markers, reconciles contradictions across sources, assigns confidence scores, and produces a structured artifact with full citation chains.

How It Works

Each of the eight methodologies — Jobs-to-be-Done, User Journey Mapping, Gap Analysis, Hypothesis Testing, Comparative Analysis, Sentiment and Opinion, Audience Segmentation, Exploratory Discovery — is implemented as a step engine that maintains workflow state across a multi-turn conversation.

Per-methodology generators produce structured survey output with sections, questions, and branching logic appropriate to the chosen framework. Methodology-specific analytics adapt accordingly: JTBD surveys produce Ulwick opportunity scores, User Journey surveys produce friction rates by stage, Hypothesis Testing surveys produce validation results with confidence intervals.

All AI interactions use Vercel AI SDK data streams. Tool calls — methodology analysis, question generation, survey creation — are observable events in the stream. Multi-tenancy is enforced through Supabase Row-Level Security with a four-role hierarchy and plan-aware quota enforcement at the API layer.

Where It Fits in Worksona

OAIRA is the research intelligence layer of the Worksona portfolio. It sits downstream of strategic planning and upstream of business decision-making — the system that turns questions into structured, validated, actionable insights.

Its position is distinct from the other portfolio projects. Where Meshwork captures and serves organizational knowledge passively as meetings and documents accumulate, OAIRA is activated deliberately: a team identifies a research question and instructs the platform to pursue it. The two systems are complementary — OAIRA's research outputs can feed Meshwork's knowledge graph; Meshwork's accumulated organizational knowledge can provide context for OAIRA's AI agents.

The deeper portfolio claim is that research methodology belongs in software, not in the heads of specialist consultants. When methodology is code, research quality becomes systematically improvable. Teams don't need to hire expertise they don't have — they inherit it from the platform.

---

Live: oaira.worksona.io