Client
Services
Platform
Bringing accountability to government, starting with Kansas. This platform tracks every legislator across voting records, campaign finance, hearing transcripts, lobbying disclosures, and public statements — exposing the gap between what politicians say and how they actually vote. The Kansas beta is live now at kansas-accountability.expertaiservices.com.
The Kansas Accountability Platform is a full-stack civic tech application that ingests data from eight distinct government sources — OpenStates, KLISS, SLIQ Harmony, KPDC, Kansas Secretary of State, social media, and press releases. Hearing audio is automatically transcribed using Voxtral (Mistral) with speaker diarization, then Claude identifies each speaker by matching against committee rosters. All data flows through Prefect-orchestrated ETL pipelines into a PostgreSQL database with 46 OCD-standard tables, served by a FastAPI backend with 80+ endpoints and a Next.js frontend with 28 interactive pages.


Automatically discovers hearings via SLIQ Harmony, downloads video with yt-dlp, transcribes audio with Voxtral speaker diarization, and identifies speakers using Claude AI cross-referenced with committee rosters.

AI-enriched profiles combining voting record, campaign finance, lobbying connections, public statements, and floor speeches. Includes education, career history, and policy positions.
Tracks all 704 vote events and 8,177 individual legislator votes across the session from OpenStates and KLISS journal PDFs. Detects when legislators change positions on the same bill.

3,433 contributions ingested from Kansas Political Disclosure Commission. Search by legislator, committee, or amount. Links donors to lobbying clients for corruption pattern detection.

Aggregates Twitter/X, Facebook, and press releases. Compares stated policy positions versus actual voting record on bills to expose inconsistencies.

Live sync of upcoming and recent hearings from SLIQ with full document attachments, extracted PDF text, and timestamped video links for every hearing session.

Context-aware AI assistant that detects the user's current page and injects bill details, legislator profiles, hearing data, or transcript metadata directly into the prompt. Ask questions about whatever you're viewing — the AI already knows.

Mem0 + Qdrant vector database powers persistent conversation recall across web, SMS, and email channels. OpenAI embeddings enable semantic search of past interactions, with automatic entity extraction, topic categorization, and workspace-scoped privacy.
Legislators Tracked
Votes Recorded
Contributions
API Endpoints
The Kansas Accountability Platform is live and free to explore. Government transparency shouldn’t require a team of researchers. We design AI systems that automate the tedious work of tracking legislators, connecting financial relationships, and surfacing inconsistencies — so citizens and journalists can focus on what matters.
Try the BetaPython / FastAPI / Next.js 15 / React 19 / TypeScript / PostgreSQL / SQLAlchemy 2.0 / Alembic / Anthropic Claude / Mistral Voxtral / Prefect 3 / Mem0 / Qdrant / OpenAI Embeddings / MongoDB / Twilio / Postmark / Redis / CrewAI / BeautifulSoup / yt-dlp / ffmpeg / PyMuPDF / structlog / Docker / nginx / HLS.js / Recharts / TailwindCSS 4