AI & Call Intelligence
AI on the platform you already run.
Thirdlane runs post-call analysis, transcription, and call intelligence as first-class platform capabilities — not as a bolted-on third-party UI. Your tenants get summaries, sentiment, QA scoring, and action items from the same console they already use to operate the system.
Post-call analysis
Summaries, sentiment, action items, and structured signals on every call you choose to analyze — all surfaced in the recorded-calls UI, not behind a separate product.
Call summarization
Automatic, structured summaries of completed calls — the gist, the decisions, the next steps. Surfaced in the call detail view and on the recorded calls grid.
Sentiment analysis
Per-call sentiment scoring with breakdowns by speaker. Drill into the moments that drove the score from the consolidated recording detail window.
Action item extraction
Identify follow-ups discussed during the call so reps and supervisors can act on commitments without listening back to the whole recording.
QA scoring
Configurable QA rubrics scored automatically against the transcript. Flag scoring outliers for human review without manual sampling.
Categorization
Tag calls by topic, intent, or category from a configurable taxonomy — drives reporting, routing decisions, and post-call workflows.
Entity extraction
Pull names, accounts, dates, and order numbers from the transcript so your downstream systems get structured data, not blobs of text.
Compliance checking
Verify required disclosures, prohibited language, and policy adherence against per-tenant rule sets.
Transcription & speech
Voice in, text out — across calls, conferences, voice messages, and greetings. Use Thirdlane’s speech provider, or wire in your own.
Recording transcription
Calls and conferences are transcribed end-to-end with speaker turns, attached to the recording, and searchable from Reports.
On-demand voice-message transcripts
Tap any incoming voice message in Connect to transcribe it locally with one click. Copy-to-clipboard and inline display, no separate workflow.
Greeting transcription
Non-TTS greetings can be transcribed for informational use and to provide text input back to TTS — useful when migrating between providers.
Custom Speech Services provider
Bring your own speech stack. Custom TTS and transcription providers run as executable scripts in /usr/local/share/thirdlane/service/{tts,transcribe} with configurable per-service environment variables.
Operating the AI
Practical knobs and surfaces that make AI features usable for the people running the system day to day.
Per-service configuration
AI analysis options are configurable per service — you decide which calls get summarization vs. QA vs. compliance. No all-or-nothing toggle.
Automatic model discovery
Provider integrations enumerate available models so admins pick from a current list rather than hand-maintaining provider-specific config.
Consolidated detail view
Call summary, transcription, action items, and AI insights (sentiment, QA, entities, compliance) live in one tabbed window — open from a single Details icon on the recorded calls grid.
Visibility cues
Call history avatar badges indicate when AI analysis is available; agents and supervisors don’t need to hunt for which calls have been processed.
Bring your own provider
Thirdlane treats AI providers as pluggable. We support major commercial providers out of the box, and the platform will discover their models automatically. If you need to use a provider we don’t ship — including a private/on-prem model — drop in a Custom Speech Services provider as an executable script and Thirdlane will route TTS and transcription requests through it.
That matters in two scenarios: regulated industries that can’t send audio to a third-party cloud, and service providers that want to differentiate on AI without locking customers into a single vendor’s pricing.
See AI in your tenant.
Walk through summarization, sentiment, QA scoring, and transcription on a sample tenant — and discuss which provider model fits your customers and compliance posture.