Build an automated call-summary pipeline — RingCentral (RingSense) → AI shift reports

We're a multi-state services company using RingCentral for our business phone system, with call recording and RingSense (RingCentral's AI add-on) enabled. We want an automated pipeline that pulls our recorded calls and produces a per-employee, per-shift summary — so we can see, without listening to recordings, what each employee handled during their shift: the calls they received, what clients needed, the tasks/action items generated, how they responded, and any unresolved or flagged items. RingSense already generates a summary, transcript, highlights, and action items for each individual call. Your job is to build the layer on top: pull that data via API, group it by employee and shift, and roll it up into one clean summary per employee per shift, delivered automatically on a schedule. What you'll build A scheduled job that, on a defined cadence (end of shift / end of day): Pulls the call log filtered by extension and time window (RingCentral Call Log API) Retrieves the RingSense AI insights for each call — summary, transcript, highlights, action items, sentiment (RingSense Insights API) Groups calls by employee (extension) and shift Sends each grouped set to an LLM API (Anthropic Claude preferred) to produce a single rolled-up shift summary per employee Writes the output to our destination (Google Sheets to start; we'll confirm) Technical requirements RingCentral developer app: server-only, JWT auth, super-admin authorized (required for RingSense data) Scopes/permissions: ReadCallLog and ReadCallRecording (ReadCallRecording is granted on request via RingCentral support — you'll guide us through that request) Familiarity with RingCentral's RingSense Insights API endpoints Rate-limit handling: throttling and exponential backoff on the recording/insights endpoints LLM API integration for the summarization step A scheduler/host for the recurring job (cloud function + cron, or an equivalent; Make.com acceptable if you prefer no-code orchestration) Note: RingCentral truncates recordings at ~90 days / 100k files, so summaries must be stored persistently Deliverables Working, scheduled pipeline running end-to-end Clear setup/handoff documentation (credentials, config, how to adjust shift windows and the summary format) A short walkthrough so our team can maintain it Suggested milestones RingCentral app + auth set up; successfully pull call logs, recordings, and RingSense insights for a test date range Employee/shift grouping logic + LLM summarization producing sample shift reports Output to destination + scheduling + error handling/throttling Documentation and handoff What we'll provide RingCentral super-admin access and confirmation that RingSense is active API credentials and support raising the ReadCallRecording request Our shift-window definitions, the list of extensions to cover, and the exact fields we want in each summary The output destination Ideal candidate Demonstrated RingCentral API experience (call log + recordings; RingSense a strong plus) OAuth/JWT, Node.js or Python LLM API integration experience (Anthropic or OpenAI) Google Sheets API or Make.com experience To apply: Briefly describe a past RingCentral API integration you've built, and note whether you've worked with RingSense specifically.