Every call teaches the AI what it still needs to know.
Upload historical call recordings today; nightly analysis of IBIOS-handled calls is coming soon. The AI identifies questions callers asked that the knowledge base could not cover and flags them for review — so the KB improves automatically over time.
Speech recognition with trade vocabulary correction. Gap detection against existing KB. Recurring pattern highlighting. Staff approve before any new article goes live.
The AI that never learns from its own calls stays average forever
Coming soon: an AI Receptionist handles a call and the caller asks a question the AI cannot answer — it escalates to a human or gives a vague response. Without call learning, that gap in the KB persists. The same question comes up again next week, and the week after, and the AI gives the same inadequate response every time.
Call learning breaks that cycle. Every call where the AI struggled is analysed. The gap is flagged. A suggested KB article is generated. A staff member reviews and approves it. The next caller who asks the same question gets a proper answer. The AI gets better with every call it handles.
Call learning at a glance
What knowledge gaps does call analysis flag?
Questions with no KB match
Caller asks about a product range, service area, or policy that does not appear in any current KB article. IBIOS flags the gap with the exact phrasing used.
Questions with low-confidence matches
The KB has a related article but the match confidence is below your configured threshold. The AI may have given a vague answer. The gap is flagged so the article can be improved.
Recurring questions not yet in the KB
The same question pattern appears across several recent calls. Recurring gaps are highlighted as priority enrichment items in the gap report.
Coming soon: questions the AI escalated to a human
Coming soon with the AI Receptionist: whenever it triggers a handoff to a human because it could not answer, the triggering question is automatically added to the gap queue.
Coming soon: out-of-hours caller questions
Coming soon with the AI Receptionist: calls handled outside business hours often reveal questions that regular callers would normally ask staff directly. Out-of-hours gap patterns are tracked separately.
Post-call sentiment drops
If call sentiment analysis detects a negative shift correlated with a specific question, that question is flagged for immediate KB enrichment review, not the standard nightly batch.
How does a call recording improve the knowledge base?
Recording transcribed
Uploaded audio or call recordings from your phone provider are transcribed using the integrated speech recognition engine with trade vocabulary post-correction.
Questions extracted
The transcript is analysed for caller questions. Speaker diarisation identifies business vs caller turns. Questions are extracted with context.
KB cross-referenced
Each question is checked against the existing KB. Questions without a good match, and recurring question patterns, are flagged as knowledge gaps.
Gap report generated
Flagged gaps appear in the knowledge gap queue with AI-suggested KB articles. Staff review and approve. New articles are live immediately on approval.
Two modes: onboarding batch and continuous learning
Historical recording upload
At onboarding, upload MP3 or WAV files of past calls. IBIOS batch-processes them and generates a gap report covering your most common real-world conversations. This is a quick way to build a KB that reflects your actual caller questions rather than what you assumed they would ask.
- ✓ Batch process past recordings in one go
- ✓ Gap patterns ranked by frequency
- ✓ Process before the AI goes live
Coming soon: continuous live call analysis
Coming soon: once IBIOS is handling your calls, every transcript will be automatically analysed nightly, with new gaps added to the queue without any staff action, so the KB compounds over time.
- ✓ No staff action needed to trigger analysis
- ✓ Nightly batch, urgent gaps flagged immediately
- ✓ Gap volume trends over time visible in dashboard
Questions about customer call learning
How does IBIOS analyse call recordings for knowledge gaps?
IBIOS transcribes audio using the integrated speech recognition engine and then passes the transcript through a gap-detection prompt. The AI identifies questions the caller asked and cross-references them against the existing KB. Questions that do not match any current KB article are flagged as knowledge gap candidates, along with the caller’s phrasing, so you can see exactly what was asked and write an article to cover it. Flagged gaps appear in a dedicated section of the knowledge management screen.
What is the difference between historical recording upload and live call analysis?
Historical recording upload is for onboarding: you upload MP3 or WAV files of past calls and IBIOS batch-processes them to identify recurring questions. This builds a starting KB from your most common real-world conversations. Continuous analysis of calls handled by IBIOS is coming soon with the IBIOS phone system and AI Receptionist — when it launches, every call transcript will be analysed nightly and gaps flagged without any staff action. Both modes produce gap reports that staff review and convert into KB articles.
How accurate is the call transcription for trade terminology?
The speech recognition engine delivers strong accuracy on trade terminology in standard UK English. Accuracy is higher when callers speak clearly and lower with strong regional accents or very specialised technical terms. Post-transcription, IBIOS applies a trade vocabulary correction pass that improves accuracy on common bathroom and tile trade terms. Each transcript is shown in the gap report alongside the detected gaps so staff can read the original wording and correct any transcription errors before approving a KB article.
Does IBIOS use customer call data responsibly?
Yes. Call recordings are processed in accordance with your existing call recording consent notice — IBIOS’s processing for AI training is covered by the same “calls may be recorded for training” disclosure your business already uses. Customer names and phone numbers from call metadata are not stored in KB articles. Only the question pattern and the suggested answer are retained. Raw recordings uploaded for batch processing are stored encrypted.
What happens after IBIOS flags a knowledge gap?
Flagged gaps appear in the knowledge gap queue with the caller’s question phrased exactly as asked, plus a suggested KB article title and body generated by IBIOS. Staff can approve the suggested article as written, edit it before approving, or dismiss the gap as not relevant. Once approved, the new KB article is immediately available to IBIOS (and, when it launches, to the AI Receptionist for the next call). Dismissed gaps are tracked so that if the same question comes up in future calls, IBIOS can re-surface it as a pattern rather than a one-off.
Which IBIOS product line includes customer call learning?
Customer call learning is part of the IBIOS AI Brain. Learning from calls the AI receptionist hands to a human is coming soon. It is built for showroom retailers and bathroom retailers. See also: the AI knowledge base.
IBIOS (Integrated Business Intelligence & Operations System) is made by IBIOS Limited (NI743167), Northern Ireland. Access is by waiting list; there is no free trial.
An AI that learns from your calls.
Upload historical recordings at onboarding. Continuous nightly analysis of IBIOS-handled calls coming soon. Staff approve every new KB article before it goes live.