Welcome to Askalot¶

Askalot is the Research Operating System with mathematically verified questionnaires. One platform escorts the researcher across the whole lifecycle — ideation → design → campaign → execution → data quality analysis — coordinated by a central Research Paper that every AI agent reads from and writes to. Formal mathematical proof confirms every path, every condition, and every answer rule in the questionnaire is logically sound, before you field it.

What is Askalot?¶

Askalot combines AI-powered questionnaire generation with SMT-based formal verification to ensure survey logic is mathematically sound. The platform relieves AI from the cognitive burden of simultaneously managing question content, conditional dependencies, logical consistency, ordering, and avoiding contradictions—allowing it to focus on generating meaningful questions while mathematical validation ensures coherence.

Mathematical Foundation¶

We created a declarative model for questionnaire specification that fundamentally differs from traditional flow-based approaches. Rather than viewing questionnaires as sequential execution paths, we formalize them as collections of items with preconditions (Boolean predicates determining when questions appear) and postconditions (constraints on acceptable responses). This declarative paradigm, inspired by Hoare Logic, treats the questionnaire as a set of logical relationships where the actual flow emerges dynamically based on respondent answers. Our static analysis investigates all possible paths simultaneously through constraint satisfaction, reasoning about relationships between immutable states rather than simulating execution through mutable values. This enables validation of questionnaires containing hundreds of interconnected questions with complex conditional dependencies—a level of complexity that exceeds human comprehension and often makes traditional validation approaches computationally intractable.

See Theory Documentation for complete mathematical foundations.

The Research Paper — central source of truth¶

Every research project on Askalot has one Research Paper. It is opened during ideation, grown through design, consulted during campaign and execution, and closed during analysis with the verdict on what the data showed. It is what ties the five lifecycle stages into one continuous workflow rather than five disjoint tools — the Designer, Manager and Analyst agents all read from and write to the same paper, enriching it by accretion: an agent reads a chapter before changing it, and edits are targeted rather than wholesale, so earlier-stage and human-approved content is preserved. (The Respondent agent only answers surveys and has no access to the paper.) See the AI agents for the agent contract and each agent's specialist sub-agents.

Platform Architecture¶

Four integrated services implement the lifecycle, each consuming and updating the Research Paper:

🛠️ Armiger — Ideation & Questionnaire Design¶

Browser-based development environment that turns a research question and source documents into a Research Paper, then into a verified QML questionnaire with AI assistance and real-time SMT validation.

AI ingests documents (PDF, Word, spreadsheets) and natural-language objectives, drafts the Research Paper
AI-assisted questionnaire generation grounded in the paper
Z3 constraint solver integration for logical validation
Interactive flow visualization with interactive diagrams
Reachability analysis (ALWAYS, NEVER, CONDITIONAL)
Postcondition validation (TAUTOLOGICAL, CONSTRAINING, INFEASIBLE)

Access: armiger.dev.askalot.io

🎯 Targetor - Campaign Management¶

Build, launch, and track survey campaigns with demographic targeting.

Import respondents from user databases
Target specific demographic factors
Track campaign progress and response rates
Manage multiple concurrent campaigns

Access: targetor.dev.askalot.io

📋 SirWay - Survey Execution¶

Lazy-evaluation based survey platform with dynamic form creation.

Functional flow navigation with path evaluation
Dynamic question rendering
Contradiction-free survey execution
Interviewer-Assisted Mode for phone and in-person interviews

Access: sirway.dev.askalot.io

📊 Balansor — Data Quality Analysis¶

Don't just export results — measure and defend them. Balansor weights, scores, and refines responses against the Research Paper.

Medallion pipeline: Bronze (raw) → Silver (open-ends coded, then raked / post-stratification weighted) → Gold (publication-ready)
Representativeness metrics: RMSE, MAE, Chi-square, max deviation
Response-quality flags: straightlining, speeder detection, Cronbach's α (internal consistency)
Survey-level metrics: completion rate, design effect (DEFF), effective sample size
Data export to CSV, Excel, SPSS, Parquet with full lineage tracking
The Analyst agent interprets the dataset against the paper, not in isolation

Access: balansor.dev.askalot.io

Quantitative Analysis Philosophy¶

Askalot is fundamentally designed for quantitative analysis where all survey responses are represented as numerical values, enabling rigorous mathematical analysis and formal verification.

Integer-Based Outcomes¶

Every question type in QML produces integer outcomes:

Closed questions: Direct numerical encoding (defined by scalar range or enumeration)
Multiple selection: Bit mask encoding allowing mathematical operations
Ranges: Szudzik pairing function encoding two integers as one

Open-Ended Question Handling¶

For open-ended questions, Askalot transforms free-text responses into numerical representations through:

AI-Assisted Coding: Open-end coding proposes the dimensions an open question's answers vary on, each with its own categories; you review them and pick which to keep, and each one you keep becomes one labelled categorical column. Coded before weighting, as the first step of a Bundle's Bronze→Silver derive, so a dimension is available as a weighting factor in the same derive. Available in the Balansor UI; readable — and the question selection writable — via the get_bundle_coding and code_open_ends MCP tools.
Sentiment Analysis: AI analysis of text sentiment, converting qualitative responses into numerical sentiment scores. Planned for a future release.
Hybrid Approaches: Combination of clustering and sentiment analysis for multi-dimensional numerical representation. Depends on sentiment analysis implementation.

This quantitative-first approach enables:

Mathematical validation of survey logic via SMT solvers
Statistical analysis with standard quantitative methods
Formal verification of consistency and reachability
Data export to standard formats (CSV, Excel, SPSS, Parquet)

Complete Workflow¶

The Research Paper moves with the project across every stage:

Ideate & Design in Armiger → AI drafts the Research Paper from documents + natural-language objectives, then generates a verified QML questionnaire
Campaign in Targetor → Define campaigns, sampling strategies, and demographic targets aligned with the paper
Execute in SirWay → Collect responses with contradiction-free dynamic flow navigation (or simulate with persona agents)
Data Quality Analysis in Balansor → Weight, score, and refine the dataset; the Analyst agent reads the paper to interpret findings against the original objectives

Application Domains¶

While Askalot was designed for survey research, the combination of AI-assisted generation and formal mathematical validation makes it applicable to any domain requiring logically consistent, adaptive questioning:

Academic Research - Survey research with mathematical validation guarantees
Clinical Trials - ePRO/eCOA (electronic Patient-Reported Outcomes/Clinical Outcome Assessments) with adaptive questioning
Emergency Medical Triage - Diagnostic tools for first responders using adaptive questioning to identify critical conditions rapidly
Legal Case Discovery - Dynamic question generation for legal interrogation, efficiently navigating complex decision trees
Critical System Verification - Formally verified operational procedures for safety-critical systems (spacecraft pre-launch, nuclear reactor protocols, particle accelerator startup)

The formal verification ensures that critical steps cannot be bypassed, safety interlocks are properly enforced, and every possible state leads to appropriate verification steps—transforming traditional checklists into mathematically guaranteed protocols.

Next Steps¶

Introduction

Platform overview and core concepts

Learn More
Quick Start

Follow the complete workflow from design to analysis

Start Now
Creating Surveys

Learn QML syntax and best practices

Guide
Mathematical Theory

Understand the formal foundations

Theory
AI-Assisted Features

Questionnaire Design, Campaign Management, Simulation, Analysis

AI Features