TenderPilot: Agentic AI for Tender Analysis

Tender documents are massive, complex, and Hebrew

Governmental procurement professionals face a critical challenge: tender documents run hundreds of pages, are legally dense, and are written in Hebrew - a morphologically rich language that breaks most standard NLP tools. Finding specific clauses, checking compliance, and drafting responses requires hours of manual work per document. At scale, organizations managing dozens of tenders simultaneously simply cannot keep up.

Agentic workflow architecture

TenderPilot takes a fundamentally different approach from traditional RAG systems. Instead of simply chunking documents and embedding them, we built an autonomous multi-agent workflow where specialized AI agents collaborate to process, understand, and index tender documents with human-level comprehension.

At the heart of TenderPilot is a stateful orchestration engine managing four specialized agents:

Each agent makes context-aware decisions, can retry when encountering ambiguous content, and works in parallel while respecting dependencies. The result: human-level document comprehension with complete data sovereignty - no data leaves the local environment.

Solving the hard parts

PDF extraction from complex Hebrew layouts

Tender documents contain complex tables, multi-column layouts, and Hebrew text that standard PDF parsers mangle. We integrated LandingAI - a specialized computer vision product - for high-fidelity extraction that preserves document structure and accurately handles mixed-direction text before the agentic workflow begins.

Retrieval accuracy at legal-grade precision

We implemented a Relevance Filter to grade retrieved documents before surfacing them, and a Generated Questions mechanism where the system embeds questions answered by the text rather than just the raw text. This "Reverse-HyDE" approach dramatically improves retrieval precision - the system finds the right clause even when the query uses different terminology than the document.

Interface walkthrough

Click any screenshot to enlarge.

Dashboard View

Document Analysis

Chat Interface

Metrics

Memory Index