Apr 6, 2026
How We Built an AI System That Automates Senior Solution Architect Workflows
Discover how we built a 4-agent AI co-pilot that converts complex RFPs into draft technical proposals in 15 minutes — with built-in conflict detection, assumption surfacing, and confidence scoring.
Author
Book a call
Table of Contents
30% of a senior architect's time goes into reading RFPs and writing proposals rather than designing systems or solving engineering problems.
A 50-page enterprise RFP lands in an inbox. Two hours later, the architect has concluded what they suspected on page three: React Native frontend, Node backend, PostgreSQL, and Stripe for payments. The decision was predictable, but the process was manual and time-consuming.
What the AI Presales Co-Pilot Does
This system assists with the presales cycle. You upload a client's documents—PDFs, Word docs, Excel sheets, or emails—and the system produces a draft technical proposal that an architect can review and refine in 15 minutes instead of writing from scratch for 3 hours.
The value extends beyond document generation to reasoning:
- Detects Contradictions: It flags tensions, such as a request for Native Mobile only paired with Critical SEO.
- Surfaces Assumptions: It states the assumptions it is making (e.g., assuming Stripe for payments) so an architect can confirm or override them.
- Confidence Scoring: It scores its own decisions, indicating where reviewers should focus attention.
How It Works: Four Specialized Agents
The system runs four specialized AI agents in sequence. The entire pipeline runs over a WebSocket connection, streaming progress to the dashboard in real time.
Agent 1: Requirement Interpreter
This agent parses raw, unstructured documents using format-specific tools (PyMuPDF, pandas, etc.). It extracts features, technical constraints, and non-functional requirements.
- Pattern Enrichment: If a two-sided marketplace is mentioned, the system suggests features like ratings and booking based on a built-in knowledge base.
- Conflict Detection: A rule engine identifies over-engineering or technical mismatches during initial analysis.
Agent 2: Solution Engine
This agent produces tech stack recommendations and effort estimates. It uses a hybrid approach: deterministic rules suggest the baseline stack (e.g., NestJS/PostgreSQL), while the LLM validates that choice against the specific client context. Effort estimation provides optimistic-to-pessimistic ranges (e.g., "120–210 hours") to reflect real-world uncertainty.
Agent 3: Self-Critique
Before a proposal is written, a critique agent reviews the solution for consistency. If an assumption contradicts a user input, the system sends a clarifying question back to the architect via WebSockets. Only when the critique approves does the pipeline proceed to generation.
Agent 4: Proposal Composer
The Pipeline at a Glance
| Stage | Role | Output |
|---|---|---|
| 1. Parse | Ingestion of any file type | Combined raw text |
| 2. Extract | Conflict & pattern detection | Structured requirements JSON |
| 3. Design | Rule-based stack matching | Solution + effort mapping |
| 4. Review | Validation & User Clarification | Approved technical solution |
| 5. Write | Template-based generation | Polished Markdown/PDF proposal |
Engineering Insights
Rule-based systems and LLMs complement each other: Conflict detection and effort mapping do not need an LLM—they need deterministic rules. LLMs handle ambiguity and prose generation.
The Technical Stack
- Backend: Python with FastAPI for async orchestration.
- AI Layer: OpenRouter (Claude, GPT-4, and Gemini) for multi-model reasoning.
- Communication: WebSockets for real-time streaming to a React-based frontend.
- Knowledge Base: JSON-based datasets for requirement patterns and stack decision rules.
Related Articles.
More from the engineering frontline.
Dive deep into our research and insights on design, development, and the impact of various trends to businesses.
May 15, 2026
Build vs Buy: Choosing the Right AI Strategy for Insurance Companies
Build or buy AI for insurance? Learn how to avoid vendor lock-in, lower AI operating costs, and build scalable, compliant insurance platforms.
May 15, 2026
Beyond AI Pilots: Building Production-Ready RCM Platforms for Denial Prevention, Coding Accuracy, and Smarter Billing
Build production-ready RCM platforms for denial prevention, coding accuracy, smarter billing, compliance, and scalable healthcare AI revenue operations.
May 15, 2026
Why AI Insurance Projects Fail in Production
Why do most AI insurance projects fail in production? Discover the hidden architectural, compliance, and scaling gaps behind failed AI deployments.
May 14, 2026
A 50-Point Production Readiness Checklist for AI-Generated Products
This 50-point AI production readiness checklist helps engineering leaders determine whether an AI-generated prototype is ready for enterprise production, or whether it needs to be hardened, refactored, or rebuilt before launch. It covers five pillars: architecture, model and data readiness, observability, security and compliance, and product and business readiness.
May 14, 2026
Building a Production-Ready Image Cropper in React Native
A practical guide to building a custom gesture-driven image cropper in React Native, with support for both profile and cover photo crops.
May 11, 2026
From MVP to Scale: Designing Architecture for AI-First Products
A panel of architects and engineering leaders at thegeekconf mini 2026 discuss how to build and scale AI-first products — from MVP decisions to production-level challenges. The conversation covers data quality, model selection, security, token economics, and the mindset teams need to navigate a fast-moving AI landscape.