Customer support · Artificial Intelligence
AI-powered customer support platform
One grounded AI at the centre of every support channel — web chat, email, messaging and live phone calls — resolving routine enquiries around the clock and escalating to a person only when it should.
Project
Multi-tenant · built for deployment worldwide
Phased delivery, usable at each step
Discuss a similar brief
The challenge
Where things stood.
A lean support operation was running across disconnected tools — live chat in one place, email in an inbox, phone calls on personal mobiles — with no shared view of any customer's history. A large share of enquiries were the same handful of questions, answerable from material the business already had, yet consuming the team every day.
Anything arriving outside office hours simply waited until morning, and the phone left no record, transcript or link to a customer's other conversations — so phone support was impossible to scale, measure or improve. The status quo did not just cost time; it capped how much the business could grow without hiring more people.
How we built it
Delivered in deliberate phases.
- 01
A channel-agnostic core
Every incoming message — widget, email, messaging app or phone — is normalised into one internal shape, so the same answer engine and routing run identically regardless of channel. The channel a customer uses never changes the intelligence, the knowledge or the record behind it.
- 02
A grounded answer engine
Each question passes through four stages before a word is sent — curated FAQs, semantic vector search, document retrieval, then language-model composition from the retrieved facts. Because every answer is assembled from the business's own content, the system minimises invented replies and hands off when it cannot answer confidently.
- 03
Real-time AI voice
Callers have a natural spoken conversation with an AI answering from the same knowledge base as chat and email. A human joins only when the caller asks — and the live call rings inside an agent's browser with the transcript and an AI summary already on screen.
- 04
An agent workspace built around context
A unified inbox brings every channel into one queue with an in-browser softphone, on-demand thread summaries and suggested replies. When a conversation reaches a person, they arrive already knowing what it is about.
- 05
Self-service, automation and oversight
A branded help centre, structured intake forms and a no-code flow builder deflect work upstream, while per-site configuration, explicit escalation rules, quality scoring and complete transcribed records keep people in control and the service measurable.
What we delivered
The work.
- A channel-agnostic core serving web chat, email, messaging and live voice from one AI brain and one inbox
- A four-stage, knowledge-grounded answer engine with a confidence gate and explicit human hand-off
- Real-time AI phone answering with seamless, context-rich transfer into an in-browser softphone
- An agent console with unified inbox, AI thread summaries, suggested replies and full cross-channel history
- Self-service help centre, structured forms and a no-code conversation flow builder
- True multi-tenancy with per-site knowledge, channels, business hours and behaviour, plus quality scoring and oversight
Outcome
What changed.
- Instant answers across chat, email, messaging and phone, around the clock, including evenings and weekends
- People spend their time on the conversations that genuinely need human judgement, not repetitive FAQs
- Every hand-off arrives with full context and an AI summary, so customers are not asked to repeat themselves
- The same accurate, on-brand answers regardless of how a customer chooses to get in touch
- Every conversation, including transcribed phone calls, captured in one place and measurable
- A multi-tenant foundation that extends to new brands, sites and teams as the business grows
Put one well-grounded AI at the centre of every channel, keep people for the moments that need them, and a lean team gains the reach of a much larger one.
Sales enablement
Sales AI — real-time call intelligence
Real-time intelligence that listens to live sales calls, transcribes both sides, reads sentiment and buyer interest, and delivers in-the-moment coaching — with live quality control and performance-based lead routing for managers.
Enterprise systems · EdTech
EduFlow — multi-tenant School ERP
A config-first, multi-tenant School ERP spanning the full school lifecycle — admissions to accounting — where each school tailors its own rules, numbering, approvals, documents and branding by configuration rather than custom code.
Related insights
The engineering behind the work.
Artificial intelligence
One registry, many modules: reproducible ML with a CI that runs everything
Twenty-three models stay consistent because one registry is the single source of truth and a CI aggregator actually runs, and re-checks, all of them.
Artificial intelligence
On-Device and Edge Speech Recognition: Real-Time Transcription at Near-Zero Cost
The cheapest, most private way to transcribe a conversation is to never send it anywhere. Modern compact models make that practical.
Artificial intelligence
From Transcript to Intelligence: Real-Time Sentiment, Interest Scoring, and Semantic Compliance
Turning a transcript into live coaching means measuring meaning carefully -- without punishing the very skills you want to reward.