Know within seconds when a session needs you.
Or when it just landed you a customer.
Rule-Triggered Alerts watch every conversation your protege runs and notify you, by email or Slack, the moment a session matches a condition you defined. Escalation requests, knowledge gaps, lead-qualification thresholds, performance summaries. The protege doesn't drop the ball; you don't have to listen to every recording to find out.
Why Rule-Triggered Alerts exist.
The point of the protege is that you're not in every conversation. But there are conversations you want to be in. The client who started asking about a custom engagement. The session where the protege hit a knowledge gap. The call where the lead-quality score spiked.
Without alerts, you discover those moments in the Monday-morning digest: three days too late. With Rule-Triggered Alerts, you find out in seconds. Slack ping. Email. Optional escalation hand-off back to your calendar.
How it works.
Define a rule
Use natural language ("notify me when a session asks about pricing for an engagement over $50K") or structured conditions (sentiment score below 0.3, lead-quality score above 0.8, specific keyword present, etc.). The platform supports both.
Pick a channel
Email (instant, with the session detail link), Slack (with a one-click "review session" button), or both. Per-rule channel selection.
Iterate
Every alert that fires is logged. You can refine the rule, mute it for a period, or escalate it to a per-session action (e.g., "auto-create a calendar slot for me to follow up with this client").
What you can rule on.
- Session signals— sentiment, engagement, lead-quality scores
- Conversation content— keyword match, topic match, intent classification
- Client behavior— return-client flag, foundation-subscription cadence, coupon used
- Knowledge-base coverage— protege hit a gap it could not answer from KB
- Billing events— high-revenue session, escalation request, churn risk signal
- Refinement signals— protege delivered an answer the expert hasn't yet approved
Example rules in production.
Custom engagement over $50K
Notify me when a session asks about a custom engagement over $50K. Slack ping, no auto-action. Drew or Alan review within the hour.
Mid-session sentiment drop
Alert me when sentiment drops below 0.2 mid-session. Triggers a follow-up email automation to the client and a Slack ping to the expert. Catches frustrated-client moments before they churn.
KB gap on a published topic
Email me when the protege hits a KB gap on a topic I've published about. Surfaces protege refinement opportunities; feeds the EPR-1 and EPR-2 refinement loops.
These example patterns are drawn from real internal Apex Replicant ruleset patterns. Verify with rule-service config before publishing.
Pricing.
AI-evaluation usage and rule limits scale with the tier.
- Basic
Basic
$50/moUp to 5 active rules, standard AI-evaluation volume.
Best for: Solo experts running one protege.
- Up to 5 active rules
- Email and Slack channels
- Natural-language and structured rules
- Per-rule snooze and mute
- Standard AI-evaluation volume
- Most Popular
Professional
$150/moUp to 25 active rules, high AI-evaluation volume.
Best for: Multi-protege experts or small practices.
- Up to 25 active rules
- Everything in Basic
- Rule families (per-protege, per-segment, per-topic)
- High AI-evaluation volume
- Per-session action hooks (calendar slot, follow-up email)
- Enterprise
Enterprise
$500/moUnlimited rules and AI evaluations.
Best for: Sentinel + multi-team deployments.
- Unlimited active rules
- Unlimited AI evaluations
- Custom-tuned AI evaluators
- Shared rule libraries across teams
- Audit-trail exports + SOC 2 Type II readiness controls
Rule-limit numbers (5 / 25 / unlimited) and AI-evaluation volumes are illustrative; tier prices are canonical. Verify with billing-service config before relying on the limit numbers for procurement.
Common questions.
- How fast do alerts fire?
- Email and Slack alerts fire within seconds of the session event that triggered them: typically inside the live session itself, not post-session.
- Can I mute a rule temporarily?
- Yes. Each rule has a "snooze" control per-instance and a global mute toggle.
- Can I write rules in natural language?
- Yes. The Rule Service uses the platform's AI evaluator to parse intent. You can also drop down to structured-condition form for precision.
- What's the difference between Basic and Professional?
- Active-rule count and AI-evaluation volume. Basic suits a solo expert with a handful of high-value patterns. Professional opens the door to rule families (one per protege, one per client segment, one per topic).
- Does the Enterprise tier come with SOC 2 attestation?
- Enterprise customers get audit-trail exports and the underlying SOC 2 Type II readiness controls. Full SOC 2 Type II attestation is in audit; see /security for the compliance roadmap.
- Can I share a rule with my team?
- Yes. Rules are scoped per-expert by default; Enterprise tier supports shared rule libraries across teams.