BidAgents AI: Stop Searching, Start Winning.

Frees up valuable time for small and mid-sized contractors through a multi-agent approach for searching, filling, and submitting public bid proposals - letting them focus on the wins! Built with Google Agent Development Kit.

Link Link Link Link Link Link Link Link Link Link


🧠 Inspiration

“Having worked alongside a general contractor, I’ve seen just how time-consuming and fragmented the bidding process can be. Orchestrating AI agents to handle the grunt work could free up valuable time for contractors.”
Darwin Vargas, Co-Founder

So we've built a revolutionary platform to transform the industry forever.


🛠️ The Problem: PUBLIC BID PROCUREMENT IS BROKEN

Contractors spend up to 70% of their time chasing bids. For many, this is a tedious daily grind:

  • Monitoring 50+ procurement portals
  • Logging into siloed systems
  • Parsing dense PDF documents manually
  • Filling repetitive forms
  • Tracking important deadlines

But what if intelligent agents could handle the grunt work while you focus on the wins?


🚀 Introducing BidAgents AI

Leveraging Google’s recently released Agent Development Kit (ADK), Vertex AI and Google Cloud, BidAgents AI orchestrates a supervised team of intelligent web-scraping and browser-automation agents in an end-to-end pipeline for public bid procurement. All with minimal human intervention.


💎 Our Unique Value Proposition (UVP) - What differentiates us from the crowd

Feature Description
Continuous Bid Discovery 🔍 Monitors 50+ procurement portals autonomously and discovers new opportunities in real time.
Automated Document Access 📥 Logs in, downloads PDFs/Word/ZIPs, and invokes Document AI to convert them into machine-readable formats.
Precision Analysis Agent 🧠 Uses Google Document AI + Vertex AI Embeddings to extract deadlines, requirements, and evaluation criteria with high accuracy.
End-to-End Form Pre-fill & Submission 📝 Auto-fills forms with company data and manages submissions (uploads, emails, e-filing) seamlessly.
Real-Time Notifications 🔔 Sends process summaries and deadline alerts via Gmail/SendGrid, keeping contractors informed at every stage.
Analytics & Insights Dashboard 📊 Visualizes bid pipeline performance, win-rate suggestions, and efficiency gains using BigQuery and custom dashboards.

🌍 Market Competitor Analysis

Compared to leading competitors, BidAgents AI stands out with its holistic agent-based approach and real-time automation:

Feature / Company BidAgentsAI AI Tenders Brainial Altura Xait Mytender.io IBM Watson Supply Chain SAP Ariba
End-to-end bid process orchestration
Real-time bid discovery & monitoring
Automated document parsing & form fill Partial Partial Partial
Multi-agent AI architecture
Focus on mid-sized contractors
Advanced analytics & performance insights Limited Limited Limited Limited Limited
Cloud-native & scalable architecture
Integration with multiple procurement portals Limited Limited Limited Limited Limited

Sources: Competitor websites and industry reports


📈 Market Value Analysis

  • Total Addressable Market (TAM): $13 trillion (12% of global GDP)
  • Serviceable Available Market (SAM): $200 billion
  • Serviceable Obtainable Market (SOM): $2–4 billion (1–2%)

🤖 Why We Chose Google Agent Development Kit (ADK)

ADK’s open-source Python framework gives us:

  • Modularity & Precise Tool Control for Agents
  • Seamless Agent Deployment across Google Cloud using Cloud Run containers
  • Extensibility to add new agents or integrate services

Our implementation of Google’s Agent Development Kit (ADK) goes beyond the basics—BidAgents AI demonstrates a modular, scalable multi-agent system that tightly follows ADK's core principles:

✅ Multi-Agent Orchestration in Action

BidAgents AI consists of a chain of specialized AI agents, each running in its own isolated Cloud Run container and coordinated via event-driven logic defined in the Flask backend. These agents communicate through Pub/Sub-style handoffs (now abstracted), with clean boundary interfaces:

  • DiscoveryDownloadAnalysisPre-fillNotificationAnalytics

Each agent is initialized using a consistent ToolContext, ADK Agent subclass, and modular toolsets:

from google.adk.agents.llm_agent import Agent
from google.adk.tools.tool_context import ToolContext

📂 Clean, Modular Code Structure

/backend
  ├── app.py                         # Flask orchestrator entry point
  ├── routes/                        # RESTful endpoints for triggering agents
  ├── agents/
  │   ├── agent.py                   # Orchestrator agent handler
  │   ├── Bid_Discovery/agent.py    # Discovery agent using browser automation
  │   ├── document_processing_agent/
  │   │   ├── agent.py              # Handles parsing & extraction
  │   │   └── tools.py              # Custom Document AI tools
  │   ├── pdf_filler_manager/       # Pre-fill logic using Gemini 2
  │   └── notification_agent/       # Sends notifications via Gmail/SendGrid
  ├── firebase_admin_init.py        # Firebase setup
  ├── pubsub_publish.py             # Internal trigger queue
  └── requirements.txt              # Carefully scoped Python deps

All agents implement the Agent.run() method and are wrapped with input validation, async error handling, and structured logging to track state transitions.


🧙‍♂️ Where the Magic Lies — Meet the team of AI Agents

🕵️ Bid Discovery Agent

The Discovery Agent is a web browser agent powered by Selenium. It's responsible for continuously and autonomously monitoring multiple procurement portals (such as SAM.gov, state, and city portals) to identify new bid opportunities relevant to the contractor’s profile. It uses browser automation and web scraping techniques to log in, navigate, and extract bid data from these portals. The agent is designed to handle dynamic web content, authentication, and even CAPTCHA challenges if necessary.

Here is a list of portals currently supported:

  1. SAM.gov (Federal)
  2. California State Portal
  3. San Francisco City Portal
  4. Los Angeles City Portal
  5. Private Sector RFPs

📂 Document Access Agent

Automates downloading of PDFs, Word files, or ZIPs, then uses Document AI to convert them into machine-readable formats.

  • Downloader Subagent: Handles the actual retrieval of documents from various sources, managing authentication, session handling, and download logic.
  • Preprocessing Subagent: Prepares the downloaded files for analysis, converting them to suitable formats and ensuring compatibility with Document AI.

🔎 Analysis Agent

The Analysis Agent leverages Google Document AI to extract critical information from bid documents with high precision. It processes the machine-readable documents to identify and extract deadlines, submission requirements, evaluation criteria, and other key data points. The agent uses advanced natural language processing and document parsing to ensure accuracy, even with complex or unstructured documents.

  • Field Extraction Subagent: Specializes in identifying and extracting specific fields (e.g., deadlines, requirements) from the processed documents.
  • Validation Subagent: Cross-checks extracted data for completeness and accuracy, flagging any inconsistencies or missing information for review.

📝 Pre-fill & Submit Agent

Automates the completion of bid forms by scanning the extracted requirements and auto-filling them with the contractor’s company data and context (such as certifications, past performance, and contact details). It ensures that all required fields are populated accurately, reducing manual effort and the risk of errors.

🔔 Notification Agent

Sends timely alerts and process summaries in each step of the workflow - letting contractors stay on top of their search.

📊 Analytics Agent

Leverages Google Gemini-2.0-Flash to provided crucial performance insights and history of trends, empowering data-driven decision making to enhance success.


🏗️ System Design Architecture & Tech Stack

please see the attached architecture diagram This sophisticated orchestration is powered by a robust tech stack within the Google Cloud ecosystem.

  • Agents & Orchestration: Google ADK, Cloud Run, Cloud Functions
  • AI & NLP: Vertex AI Embeddings, Document AI
  • Data Stores: Firestore (metadata), Firebase Storage (documents)
  • Notifications: Gmail API / SendGrid
  • Security: Secret Manager, IAM, encryption at rest & in transit

🎯 Expected Product Outcomes

  • 70+% reduction in manual bid processing time
  • 30% increase in bid win rate
  • 24/7 opportunity monitoring — never miss a deadline
  • ROI payback in under 3 months for mid-sized contractors

🧩 Challenges We Faced

  • Diverse Document Formats: Built robust retry logic and prompt engineering for the Discovery Agent (browser agent) to handle unstructured information from public procurement portals.

  • Secure Credential Management: Integrated Cloud Secret Manager and fine-grained IAM rules.


🏆 Accomplishments We’re Proud Of

  • 70%+ form-fill accuracy, powered by custom-engineered prompts & Document AI logic
  • 30-second processing and notification of new bids post-publication
  • Gemini Computer Vision to solve CAPTCHA and website bot blockers
  • Tested end-to-end agent workflows across various industry use-cases
  • Seamless zero-downtime updates via Cloud Build & Cloud Run

🔮 Future Improvements

  • NLP-Driven Recommendations to improve the accuracy of semantic search
  • International Portal Coverage with compliance to GDPR Policies (EU, UK, Australia)
  • Offer 3 month pilot programs with local small to mid-sized contractors with dynamic revenue models
  • Roll out a beta waitlist to gather and later implement user feedback for the current prototype

🤝 Contributing to the ADK Developer Community

We’re actively improving the Google ADK Python SDK:


✨ Our Final Remarks

In this AI-era, BidAgents AI exemplifies how intelligent multi-agent workflows can transform bid procurement—letting contractors focus on winning, not chasing.


🙋 Frequently Asked Questions

Q: How secure is my data?
A: We offer end-to-end encryption and IAM controls, all ISO 27001 & SOC 2 compliant.

Q: Can agents handle non-PDF formats?
A: Yes — supports Word, Excel, HTML/XML via preprocessing subagents.

Q: Do you offer SLA-backed support?
A: Absolutely — 24/7 support with guaranteed response times for enterprise clients.


Try BidAgents AI today. Feel free to leave feedback in the comments, or reach out directly on LinkedIn.

Built With

Share this project:

Updates