AI Automation Platform Dev for AI Solutions Firm
- AI Development Custom Software Development Generative AI
- Confidential
- Jan. 2023 - Ongoing
- Quality
- 4.5
- Schedule
- 5.0
- Cost
- 5.0
- Willing to Refer
- 5.0
"They're technically elite, highly autonomous, and deeply collaborative—a rare combination in any development partner."
- Other industries
- Denver, Colorado
- 1-10 Employees
- Online Review
- Verified
Software on the road has developed a no-code/low-code AI automation platform for an AI solutions firm. They've designed a multitenant architecture, built core platform features, and integrated LLMs and APIs.
Software on the road has delivered a fully functional AI automation platform that has processed over 5,000 files, achieved a 75% first-time user retention rate, and handled over 100,000 assistant messages and workflow steps without downtime. The client has also seen paid conversions within weeks.
The client submitted this review online.
BACKGROUND
Please describe your company and position.
I am the Founder of Bellaire.AI
Describe what your company does in a single sentence.
Bellaire.ai is an AI-first platform that empowers businesses and individuals to build intelligent, multi‑agent workflows—enabling automated data‑driven tasks like document ingestion, ERP operations, web scraping, and custom AI chatbots—with human oversight and tool integrations in a single, flexible interface
OPPORTUNITY / CHALLENGE
What specific goals or objectives did you hire Software on the road to accomplish?
- Build, launch, and maintain the entire stack
SOLUTION
How did you find Software on the road?
Online Search
Why did you select Software on the road over others?
proven experieince
How many teammates from Software on the road were assigned to this project?
2-5 Employees
Describe the scope of work in detail. Please include a summary of key deliverables.
Overview
Software on the Road was responsible for the end-to-end development of Bellaire.ai—a no-code/low-code AI automation platform that allows users to build intelligent assistants and workflow agents using Retrieval-Augmented Generation (RAG), large language models (LLMs), custom logic, document ingestion, and tool integrations.
1. Product Strategy & Technical Architecture
Workshops & Discovery: Collaborated closely with the founder to define core product features, user personas, and early differentiators in the AI agent and workflow builder space.
Architecture Design: Designed a scalable, multi-tenant architecture that supports:
Long-running workflows
Contextual multi-agent collaboration
Seamless document upload and search
Token-based usage billing
2. Core Platform Features
a. AI Assistant & Workflow Builder
Drag-and-drop UI to create workflows using:
Conditional logic
Document-based RAG nodes
Custom instructions for agents
Chatbot triggers
State management and flow visualization
b. Document RAG Engine
Built a document ingestion and chunking pipeline
Implemented vector storage and semantic search
Enabled grounding AI chat on uploaded files
Support for PDFs, Excel, Word, and raw text
c. Multi-Agent Collaboration
Agents can collaborate with shared memory and context
Implemented turn-based reasoning using chain-of-thought prompts
Included tool access (API, code, webhook, database)
d. Token & Usage Management
Token system to meter usage of chat, workflows, and uploads
Stripe billing integration for subscriptions and overages
Admin dashboard to manage plans and limits
3. Integrations
LLMs: OpenAI, Anthropic, Mistral, and more
Embeddings & Vectors: OpenAI, Cohere, Hugging Face + Pinecone / Weaviate / ChromaDB
APIs: Zapier, Google Drive, Notion, Slack (planned)
Database access: SQL node with query execution and result chaining
4. Frontend Development
Built with React + Tailwind
Built-in chatbot interfaces, file upload UIs, and workflow editor
Multi-language support ready
Dark mode, user authentication (Auth0 or Supabase)
5. Backend & Infrastructure
Node.js (TypeScript) backend with Express and WebSocket support
Secure serverless file upload with S3
Task queue and scheduling for async agent runs
Logging, analytics, and error handling
6. QA & DevOps
Unit and integration tests
CI/CD pipeline via GitHub Actions
Staging and production environments
Analytics and performance monitoring
7. Go-to-Market Support (Optional/Advisory)
Assisted in early pitch decks and product positioning
Built shareable demo workflows for landing pages
Key Deliverables
Agent Workflow Builder
Drag-and-drop builder to create intelligent AI workflows
Document RAG System
Document upload, vector indexing, and context-aware QA
Multi-Agent System
Support for multiple agents with tools and memory
Chat Interface
Fully integrated assistant chat with document awareness
Auth & Tokens
Login system with user token-based usage tracking
Admin Dashboard
Usage metrics, billing, assistant logs
API & LLM Integrations
OpenAI, Anthropic, API tools, external endpoints
Testing & DevOps
CI/CD, monitoring, and production deployment pipeline
RESULTS & FEEDBACK
What were the measurable outcomes from the project that demonstrate progress or success?
MVP Launched in Under 90 Days
Delivered a fully functional AI automation platform from concept to production in under 3 months, enabling early customer demos and onboarding.
Over 5,000 Files Processed in First Month
Users uploaded and processed over 5,000 documents through the RAG pipeline within the first 30 days of launch, demonstrating immediate product-market interest.
75% User Retention After First Use
The no-code workflow builder and assistant interface achieved a 75% retention rate among first-time users, indicating ease of use and high initial engagement.
Token-Based Pricing Validated with Revenue
The integrated token system and Stripe billing led to paid conversions within weeks, validating the monetization model without a dedicated sales team.
Scalability Proven with 100,000+ Assistant Messages
The infrastructure reliably handled over 100,000 assistant messages and workflow steps without downtime or scaling issues.
Describe their project management. Did they deliver items on time? How did they respond to your needs?
The project was managed using Trello for task tracking and Discord for daily communication, enabling a highly collaborative and transparent development process. I served as the Product Manager and owner, setting priorities and defining the roadmap, while the Software on the Road team took full ownership of execution—from backend to frontend to infrastructure.
They consistently delivered items on time, often ahead of schedule, and adapted quickly to changes in scope or new feature requests. Their responsiveness was exceptional—questions were answered within hours, blockers were proactively flagged, and weekly progress was visible through clear task updates and demoable deliverables.
Even as priorities shifted during the build, they remained flexible and solutions-oriented, which made it feel like a true product partnership rather than a vendor relationship.
What was your primary form of communication with Software on the road?
Email or Messaging App
What did you find most impressive or unique about this company?
What stood out most was their ability to turn abstract ideas into working software, and do it fast. We were building something highly technical: multi-agent AI workflows, RAG pipelines, and dynamic chat interfaces. Most teams would have needed extensive handholding, but SOTR ramped up immediately and started shipping real features immediately.
Their product intuition was also rare—they didn’t just execute specs; they asked smart questions, suggested better ways to approach complex features, and improved UX without being prompted. It felt like I had a seasoned internal team that deeply understood both AI infrastructure and user-facing SaaS.
In short, they're technically elite, highly autonomous, and deeply collaborative—a rare combination in any development partner.
Are there any areas for improvement or something Software on the road could have done differently?
no, nothing i can think of
RATINGS
-
Quality
4.5Service & Deliverables
-
Schedule
5.0On time / deadlines
-
Cost
5.0Value / within estimates
-
Willing to Refer
5.0NPS
"I recommend them to everyone i know looking to launch a software project, even established enterprise companies looking to add more devs