Pitline - AI Voice Agent and Call Review Console for a Garage Chain
AI Phone Agent for a UK Garage Chain
Designed and built a clickable staff console for an AI phone agent that books, moves and cancels car servicing and MOTs, reads out job status and answers common questions for a chain of 14 Midlands garages - with every call reviewable turn by turn, every tool call on the record and a warm transfer to a person whenever the agent should not decide.
Most of our work is under NDA. Per agreement, we can share the work but not the company details - all names and identifying information shown here have been changed.
Impact
Dashboard, live calls with the live call monitor, call log, call review with the turn inspector, and agent bookings with the booking panel. Each of the 30 demo calls opens its own review.
From find_customer and lookup_vehicle to create_booking, send_sms and transfer_call. Each tool call shows inline in the call timeline with its arguments, result and duration, and writes are highlighted.
Disclosure played before the first question, no card details spoken, booking matched the caller's request, and no promise outside the knowledge base - each marked pass or warn next to the reviewer form.
30 calls across 14 sites, 6 calls live now, 24 customers and their vehicles, and 24 agent bookings: 19 synced, 3 pending and 2 rejected by the garage management system.
The Problem
Each Wrekinvale site has one or two service advisors who also run the counter, write up jobs and ring customers with quotes. Between 08:00 and 09:30, and again at collection time, the phones rang out: the chain's own records showed around a third of inbound calls unanswered at peak, and most answered calls were routine questions like "Can I book an MOT?" or "Is my car ready?".
Our Solution
Pitline answers the chain's phone lines day and night, says up front that it is an automated assistant, and books, reschedules or cancels appointments, reads out the status of a car in the workshop and answers questions from an approved knowledge base. Anything else goes to an advisor at the right site with a summary already on screen, and staff review every call afterwards with the recording, transcript and each tool the agent called.
The Challenge: An Agent That Knows Its Limits
A phone agent for a garage chain has to sound helpful, act inside a booking system it does not own and follow UK rules on recording and personal data, while handing a caller to a person the moment it should not carry on.
Recording and Disclosure on Every Call
A recorded call and its transcript are personal data under UK GDPR and the Data Protection Act 2018, and business call recording falls under the Investigatory Powers Act 2016. Callers must hear the recording notice before anything is captured, and a caller who declines gets an unrecorded call with only a minimal record kept.
A Right to a Person at Any Time
A caller can ask for a human at any point and must be offered a transfer in hours or a callback after hours. Quotes, complaints, disputes and refunds always go to a person, and a caller describing an emergency is told to hang up and call 999.
Bookings in Someone Else's System
The garage management system is the system of record, and the diary changes at the counter while the agent is on the phone. Every write goes to that system first, can be rejected, and has to be visible to staff so nobody turns up for a slot that no longer exists.
Hard Lines the Agent Must Not Cross
No card details on a call, which keeps the voice path out of PCI DSS scope. No repair price that is not in the knowledge base. No job status or booking change before the caller passes an identity check on registration plus postcode or surname. Confirmation texts must stay service messages under PECR.
Our Approach: From Call Rules to a Clickable Console
Spec and Rules Research
Mapped 8 roles, from service advisor and contact operations lead to conversation designer, data protection lead and the caller, and 30 screens: 27 staff screens and 3 public caller pages, plus 5 slide-over panels. Defined the data model for calls, turns, tool calls, transfers, callbacks, bookings, knowledge articles, flows and versioned agent releases, and wrote down the rules the design follows: UK GDPR and the Data Protection Act 2018, the Investigatory Powers Act 2016, PECR, Ofcom caller ID rules, MOT timing and PCI DSS.
Clickable Demo on a Production UI Kit
Built 5 key screens on a production dashboard kit: the dashboard with provider health, a degraded speech-to-text banner, 5 KPI cards and calls by hour with opening hours shaded; live calls with ticking durations and Listen in, Take over and End call; the call log with masked numbers and recording state; call review with speaker lanes, the turn timeline, tool call cards and the reviewer form; and agent bookings with sync status, change history and the SMS sent.
What a Production Build Adds
Telephony (such as Twilio, Vonage or Telnyx), streaming speech-to-text, a language model with tool use, text-to-speech, the garage management system (such as Garage Hive or MAM Autowork), DVLA and DVSA vehicle lookups, SMS and a pgvector knowledge base. Staff sign-in through Microsoft Entra ID, an append-only audit log, and the remaining screens: review queue, callbacks, flow editor, persona, tools, knowledge base, test sandbox, releases with a separate approver, analytics, settings, the advisor handoff screen and the caller booking pages.
Why It Works: Every Call Can Be Checked
Every Turn Has a Reason
Call review lays out each turn with its timestamp: caller speech with its speech-to-text confidence and low-confidence words underlined, agent speech with the flow node it came from and its response latency, and knowledge article chips where an answer came from the knowledge base. The turn inspector shows the variables, retrieved articles and raw tool payloads behind a step.
Disclosure Comes First, and Shows
The disclosure checklist records when the AI disclosure and recording notice played and the caller's recording choice. The call log marks each call recorded, not recorded at the caller's request or deleted under retention, and call review explains why there is no audio instead of showing an empty player.
Handoffs Carry the Context
A transfer shows its reason, target site, ring time, who answered and the whisper summary the advisor heard, such as "Mrs Patel, Ford Focus AB12 CDE, wants to know why the car isn't ready; job says awaiting parts". The caller does not have to repeat their registration.
Agent Bookings Checked Against the Diary
Every booking the agent made, moved or cancelled shows whether the garage management system synced, is pending or rejected it, with the rejection reason such as a slot already taken at the counter. Staff can confirm, edit, retry sync or cancel with an SMS, and the dashboard lists rejected bookings and full diaries under Needs attention.
Related case studies
View all: AI agents & automation
Tallyman - Intelligent Document Processing for Trade and Customs Documents
Designed and built a clickable document processing platform for a Rotterdam freight forwarder and customs representative: it classifies commercial invoices, packing lists, bills of lading, air waybills and certificates of origin, extracts each field with a confidence score, checks it against business rules and sends only the uncertain fields to a person in a side-by-side review workspace.
Sanbuk - Accounts Payable Agent with an Operator Console and Approval Gates
Designed and built a clickable back-office agent for a Dubai mainland distributor: it reads the AP and vendor mailboxes, matches invoices to POs and goods receipts, checks UAE tax invoice fields, drafts replies and posts to the ERP, while people approve by amount, see every tool call and handle every bank detail change themselves.
Enterprise Software Engineering - Human + AI
Contact Us
© 2026 - Copyright. All Rights reserved Oceanlab
Location
Global Headquarters
21 Merchant Road, 3rd Floor, Singapore 058267
Main corporate office and leadership operations
North America Office
548 Market Street, PMB 12345, San Francisco, CA 94104
Client operations and business development
India Office
402 Shivalik Shilp II, SG Highway, Ahmedabad 380015
Engineering and delivery center
Enterprise Software Engineering - Human + AI
Contact Us
Locations
Global Headquarters
21 Merchant Road, 3rd Floor, Singapore 058267
Main corporate office and leadership operations
North America Office
548 Market Street, PMB 12345, San Francisco, CA 94104
Client operations and business development
India Office
402 Shivalik Shilp II, SG Highway, Ahmedabad 380015
Engineering and delivery center
© 2026 - Copyright. All Rights reserved Oceanlab