RingCentral Call Tracking, Transcripts & AI Analysis

The problem

The sales team handles hundreds of calls a day. The owner had no easy way to know who was available, which calls were missed, or what was actually said: lost sales, complaints, customers waiting for a callback. Finding out meant listening to recordings one by one.

What I built

• Call and staff tracking: every call, staff status change (available, busy, Do Not Disturb, offline) and call queue is saved to Supabase automatically.
• Daily report: a clear 8am–5pm view of each salesperson's availability and call activity.
• Transcripts: a few minutes after each recorded call ends, it is turned into text, with who said what.
• AI analysis of every call: each call is summarised in plain language and tagged with the outcome, whether it was a lost sale (and why), whether a follow-up is needed, any complaint, the customer's mood, and the products asked about.
• Issues list: complaints, lost sales and missed follow-ups found by the AI analysis go onto one list to work through, instead of getting lost.

Privacy and cost

Audio is passed straight through for transcription and never stored. Card numbers are removed before the text is saved. Phone numbers are masked in reports. AI tools only get read-only access, through views set up for them.

The result

Each call has a transcript and AI analysis 5–10 minutes after it ends. The owner can ask "Which calls this week ended in a lost sale, and why?" and get an answer straight away, without listening to a single recording.

Built on the same setup as my Podium Customer Data Sync & AI Analysis project.