Pro AI feature
Speaker Themes.
Per-person speaking patterns drawn from your recordings — what they bring up, how they hedge, the words they reach for.
Some patterns are too quiet to notice in the moment.
Sarah ends every difficult sentence with "if that makes sense." Maria brings up her workload twice in every meeting before agreeing to anything new. Jamal hedges with "I think" before stating what turns out to be a hard requirement. The CFO uses the word "meaningful" only when she means "expensive."
You feel these patterns on some level, but you don't track them. Speaker Themes does — drawn from the recordings you're already making.
What a Speaker Theme report contains
Recurring topics
The five to ten things this person brings up most across your recordings. Not just words — themes. "Workload concerns" might appear under different phrases each time; the model groups them.
Hedging signature
The phrases they use to soften a position — "if that makes sense," "I might be wrong but," "just thinking out loud." When the hedging spikes, something is being protected.
Distinctive vocabulary
Words this person uses disproportionately compared to the rest of the conversation. Useful for catching jargon you've been letting drift past unexamined.
Talk-time patterns
Words per turn, average turn length, interruptions made vs. received, share of meeting time. Useful for managers, sales coaches, therapists noticing dynamic shifts.
Energy markers
Recurring high-engagement and low-engagement signals — speech-rate changes, vocal stress, the subjects that cause the energy to spike or drop. Inferred from the audio, not just the transcript.
Trend lines
"They've been talking about deadlines 40% more in the last month than the prior six." Useful as an early-warning signal for stakeholder concerns or account momentum changes.
Where this is especially useful
Sales coaches use Speaker Themes to understand a rep's style — where they're losing ground in the call, which moments cause them to hedge. Managers use it to understand what their direct reports actually care about across 1:1s. Therapists use it to track narrative shifts in long-term clients. Journalists use it to characterize a source's language over a sequence of interviews.
It is also useful self-applied. Run Speaker Themes on yourself across the last three months of recordings. The results are quietly clarifying.
A note on what this isn't
Speaker Themes is descriptive, not diagnostic. It does not score people on "credibility" or "leadership." It does not classify communication styles into named buckets ("Driver," "Analytical"). It surfaces patterns and hands you the recordings; the interpretation is yours.
Like the rest of the cross-recording Pro AI features, it's processed on Bonfiyah's backend under a no-training commercial contract — your transcript text is analyzed for the patterns, never used to train any model, and your raw audio is deleted from our servers within 7 days.
FAQ
Which plan includes Speaker Themes?
Speaker Themes is a Pro AI feature, part of the cross-recording AI suite. It draws its patterns from the recordings you're already making and analyzes how each person speaks across all of them.
How does Bonfiyah attribute a speaking pattern to the right person?
Bonfiyah identifies speakers by voice biometrics, never by transcript text. Each person's voice fingerprint carries across your recordings, so their recurring topics, hedging signature, and talk-time patterns accumulate against the right person rather than getting mixed together.
Does Speaker Themes score people or sort them into personality types?
No. Speaker Themes is descriptive, not diagnostic. It does not score people on "credibility" or "leadership," and it does not classify communication styles into named buckets like "Driver" or "Analytical." It surfaces patterns and hands you the recordings; the interpretation is yours.
Where does the analysis run, and what happens to my audio?
Like the rest of the cross-recording Pro AI features, Speaker Themes is processed on Bonfiyah's backend under a no-training commercial contract — your transcript text is analyzed for the patterns, never used to train any model, and your raw audio is deleted from our servers within 7 days.
Can I run Speaker Themes on a hiring or performance decision?
No. Speaker Themes is a reflective tool, not an assessment instrument, and it is not for hiring, promotion, performance evaluation, termination, or any other consequential decision about a person. The EU AI Act and the Colorado AI Act both apply to using conversation-inference AI as an input to consequential decisions — don't use Bonfiyah's reads that way without your own legal review and your own ground-truth instruments.
Used in these workflows
Where this surface earns its weight.
See the full eighteen-workflow catalog →
What Speaker Themes is — and isn't — for.
It is for: noticing how the people you talk with regularly shift over time — the topics they return to, the way they hedge — so you can follow the thread across your own conversations.
It is not for: hiring, promotion, performance evaluation, termination, or any other consequential decision about a person. Speaker Themes is a reflective tool, not an assessment instrument, and its per-utterance signals are descriptive acoustic and linguistic measures — not inferences of an emotional or mental-health state. The EU AI Act — which restricts inferring traits and emotions about people in a workplace and classifies personal-trait profiling as high-risk — and the Colorado AI Act's parallel employment framework both apply to anyone using conversation-inference AI as an input to consequential decisions about other people. Don't use Bonfiyah's reads that way without your own legal review and your own ground-truth instruments.
See a real Speaker Theme report
We'll send a sample report drawn from a real (anonymized) sequence of meetings. You'll see the recurring-topic clustering, the hedging signature, and what to ignore.
No spam. We use ConvertKit. See our privacy policy.