LLM gateway serving millions of users
One API in front of multiple LLM providers, both commercial and self-hosted. Automatic cross-provider failover, per-feature rate limits, per-user cost metering.
Millions of calls per day · 99.99% availability
About
Ingenious Networks: the team and its projects
Who we are, which organisations we have worked for, and what we have built.
About us
A team of AI engineering and telecommunications
Ingenious Networks GmbH combines long-standing UC experience with engineers who run AI platforms and voice agents at scale.
| Competence | What we bring |
|---|---|
| AI engineering | Agents, RAG, MCP, LangGraph, Pydantic AI, OpenAI Agents SDK, evaluation (LLM-as-a-judge) |
| LLM platform & ops | LLM gateways, self-hosting (vLLM), observability (Langfuse, OpenTelemetry), cost control |
| Voice AI | Real-time voice agents (LiveKit, WebRTC), STT/TTS streaming, speech-to-speech |
| Unified Communications | Cisco UCM, Webex, Webex DI, Expressway, CUBE, contact centre, Microsoft Teams |
| Software & cloud | Python, TypeScript, FastAPI, Node.js, React, PostgreSQL, Kafka, Kubernetes, GCP, AWS, Azure |
| Certifications | Vimal Bergmiller: Cisco CCIE Collaboration #38515, CCNP Voice, CCNA, Microsoft MCITP |
| Languages | German, English, Spanish |
References
Clients and projects
Ingenious Networks has delivered projects for companies including:
AirbusCommerzbankSwisscomT-SystemsNTTBT GermanyAOKDB SystelA1 Telekom AustriaSociété Générale Private BankingBSI BankHasso Plattner VenturesBSYS Stadtwerke ErfurtNextiraOnexevIT
Connecting two complex systems – where others had failed, Mr Bergmiller of Ingenious Networks, as a CCIE, did it in under three days. As a follow-up order, we engaged him for three years of support. Highly recommended.
Real-time voice agent
Streaming speech → LLM → speech pipeline with turn detection, interruptions and tool calling, including a speech-to-speech mode.
Millions of audio streams per day
Sovereign AI platform for a group of companies
Self-hosted models behind a central gateway for 7 teams across 5 companies, with SSO, tracing and per-team cost attribution.
98% of tokens in-house · $52.5k/year licence cost saved
Quality-gated AI workflow
Automated evaluation with a fixed release gate for a copywriting workflow, followed by a model swap with no loss in quality.
F1 0.55 → 0.93 · latency 18 s → 3.8 s · −40% cost per request
Research agents and MCP server for a market intelligence platform
Multi-step AI agents with RAG over market data, and an OAuth 2.0-secured MCP server that lets B2B customers query the platform from their own AI assistants.
Customers migrated with zero downtime
UC environment of an aerospace group
Planning, configuration and handover to operations of a Cisco environment with session management, SRST, call recording and provisioning.
60,000 IP phones · 30,000 Jabber clients
AI projects come from our team's experience.
Approach
Start small, scale measurably
Analysis
We look at the use case, data and infrastructure and say honestly whether AI is worth it.
Pilot
Fixed scope, fixed timeline, clear success criteria. The result is a working system – and a measurement.
Operation
Monitoring, evaluation on every model change, cost control and support – or a clean handover to the client's team.
Contact
Your AI. Your data. In the EU.
Discuss a use case with us
20 minutes, without obligation: the situation is described, and we say what AI can realistically do, and what it cannot.
Open the form
- contact@ingnetworks.com
- Phone
- +49 1515 3851511
- Address
- Ingenious Networks GmbH
Eisenacher Straße 6
10777 Berlin, Germany