Common questions.

Kairos joins your sales calls in one of two ways. Either as a meeting bot in Zoom, Microsoft Teams, or Google Meet, disclosed to participants per your tenant's consent settings, or silently via our Chrome extension where the rep is the consenting party (valid in 39 US states and Canada). The choice is per-tenant and configurable by an admin.

Kairos integrates with Salesforce, HubSpot, and Microsoft Dynamics 365. Read access is included on Starter, Growth, and Enterprise tiers. Write-back (creating contacts, updating deal stages) is available on the Enterprise tier.

Those tools are post-call analytics platforms. They listen, record, and analyze conversations after they're over. Kairos is real-time. The system surfaces guidance during the call, while the rep can still act on it. The difference matters when you're trying to win a deal in the next 15 minutes, not coach a rep next week.

Challenger Sale and MEDDPICC are built in. Custom playbooks (your own methodology) can be ingested as RAG context on the Growth tier and above. The system tracks methodology coverage in real time and surfaces guidance grounded in the framework your team is trained on.

One rep seat, 20 calls per month, full feature parity on guidance. No CRM integration, no email monitoring, no SMS delivery. The free tier exists for individual reps to experience the product. Most upgrade to Starter or Growth when their team is ready.

30 days. Your team, your deals, your call data. We work with you to configure the platform for your sales motion, integrate with your CRM, and ingest your existing playbooks. Kairos joins live calls from day one. We meet weekly to review guidance quality and rep adoption. At the end of the 30 days, you decide whether to convert to an annual subscription.

Every call session has a recorded consent method: bot disclosure, one-party (rep is consenting party), or two-party (all participants explicitly disclosed). Kairos has a jurisdiction-aware policy engine that blocks hidden mode in two-party consent states. SOC 2 Type 1 is in active progress. BAA is available on the Enterprise tier.

All tenant data is isolated through row-level security policies in Supabase. No query can return data across tenant boundaries by design. Audio retention is configurable per tenant: no retention by default, or up to 30 days. All OAuth tokens are encrypted at rest. Anthropic does not train models on customer data.

From audio chunk to guidance on screen, under one second at the 95th percentile. The real-time path uses Anthropic's Claude Haiku 4.5 with pre-assembled context and constrained 15-word outputs. Post-call analysis (email drafts, deal summaries) uses Sonnet for quality.

Yes. Every guidance item has Use it or Skip buttons. Skipping clears the item and triggers a 60-second cooldown before the next one. The system learns from rejection patterns: if a rep consistently skips a category of guidance, the system suppresses that category for them.

Generic or mistimed guidance is worse than no guidance. The system is calibrated to surface nothing rather than something low-confidence. Confidence thresholds suppress predictions below 75% accuracy on objection prediction and 70% on buying signals. If predictive accuracy falls below 60% in a deployment, the predictive layer is automatically suppressed until calibration improves.

Per-seat per-month, with annual prepay discount of 20%. Free tier is fully free. Starter is $79 per seat per month, Growth is $179 per seat per month, Enterprise is $299 per seat per month plus a platform fee starting at $5,000 per year depending on tenant size. Onboarding is included on Free and Starter; $1,500 on Growth; bundled with platform fee on Enterprise.