Our Technology

Built for enterprise scale

Modern AI infrastructure designed for reliability, security, and performance at any scale.

Architecture Principles

The foundation of our platform

AI-Native Architecture

Every model call goes through odnoga, the AI gateway SalesDots runs on: routed per task across 100+ models, with automatic fallback, and metered to the cent.

Security First

Enterprise-grade security with data encryption, audit logging, and infrastructure hosted on certified platforms.

API-First Design

Every feature accessible via REST APIs, enabling deep integrations with your existing tech stack.

Horizontal Scalability

Auto-scaling infrastructure that grows with your data volume and user base without performance degradation.

Technology Stack

Best-in-class tools and frameworks

AI & Machine Learning

odnoga AI gateway
100+ models, 8 vendors, one endpoint
OpenAI, Anthropic & Google models
Routed per task, with fallback
Vector embeddings (pgvector)
Semantic search & tender history
Trained classifier heads
Notice classification at $0 per notice

Infrastructure

Vercel
Global edge network for the web apps
Supabase
Postgres, Auth, Realtime, Storage
Edge Functions (Deno)
Harvesters, engines and scheduled jobs
PostgreSQL + pgvector
Primary store and vector index

Security

SOC 2 Type II infrastructure
Via Supabase
AES-256
Data encryption at rest
TLS 1.3
Encryption in transit
Row-level security
Tenant isolation inside Postgres
Powered byodnoga

Every AI call SalesDots makes goes through odnoga

odnoga is the AI gateway and control plane SalesDots runs on. Tender screening, notice classification, embeddings, bid analysis — none of it talks to a model vendor directly. One OpenAI-compatible endpoint routes each request to the right model, falls back when a vendor fails, and writes every call to a cost ledger.

SalesDots
odnoga gateway
Model vendors
OpenAI
Anthropic
Google
xAI
Mistral
Perplexity
Groq
DeepSeek
Fallback chain
gpt-4.1-mini→gpt-4o-mini→gemini-flash→claude-haiku-4-5
Request
2Route
3Vendor
4Metered

Illustrative request flow. The routing, the fallback chain and the metering are how it really works; the sample numbers are not live.

100+
models in one catalog
8
model vendors, one endpoint
50%
genuine vendor batch discount
EU
hosted control plane

Vendor-agnostic by design

OpenAI, Anthropic, Google, xAI, Mistral, Perplexity and more behind one OpenAI-compatible API. Switching the model behind a SalesDots feature is a routing rule, not a deploy.

Fallback before you notice

Every route carries a fallback chain. When a vendor rate-limits, errors or retires a model, the next model in the chain answers — and the switch is recorded.

Batch mode, real discounts

Nightly screening runs through the vendors' own batch APIs at their genuine 50% discount. odnoga never bills a discount it did not obtain.

Every call on the ledger

Model, prompt version, tokens, latency and cost per request — attributed to the workspace and the end user. That is how SalesDots shows AI cost in Dots instead of estimates.

odnoga is a separate product with its own model catalog, API documentation and live status page — all published at odnoga.com.

Ready to see it in action?

Join the waitlist and be among the first to experience our technology.

Join Waitlist