I need a solid AI layer that lets customers who use our mobile app reach my company’s call center without friction. The core of the project is real-time speech recognition: as soon as a client taps “call support” in the app, their spoken request should be transcribed, interpreted, and routed to the most suitable live agent (or the right IVR branch) inside our existing call-center system. Here’s what I’m after: • A lightweight SDK or API that can be embedded in our iOS and Android builds, listens for the customer’s voice, runs accurate speech-to-text, and hands the text off to our backend. • A routing mechanism that takes that transcript, checks caller intent or keyword triggers, and opens the correct queue/extension in our on-premise PBX (we use SIP). • A simple dashboard so supervisors can see transcripts, routing decisions, and call metrics in real time. • Clear documentation and a short demo app that proves the flow end-to-end. Acceptance criteria 1. Speech recognition latency under two seconds on a standard 4G connection. 2. Accuracy ≥ 95 % on everyday support vocabulary in English. 3. Successful automatic connection to the right agent group in at least nine out of ten test calls. 4. Codebase builds cleanly in Xcode and Android Studio and includes installation notes for our dev team. If you’ve previously integrated speech engines such as Google Cloud Speech-to-Text, Amazon Transcribe, or open-source Kaldi, and you’re comfortable with SIP/VoIP routing, let’s talk.