The AI SDR that lives in your Slack.
Mavrick's AI voice agent picks up inbound leads in under 60 seconds, qualifies them on a script you approved, books the meeting, and posts the handoff card to your sales channel. Built on Telnyx + OpenAI + Cartesia. Verified shipping in production.
Sub-60-second response. Sub-15-second outbound dial. Slack-native handoff.
The industry-average B2B inbound response time is 42 hours. Only 7% of B2B companies meet the 5-minute benchmark. Companies responding within 5 minutes are up to 100x more likely to qualify the lead than companies waiting 30 minutes (HBR, 2011 — replicated 2024). Humans can't hit those numbers consistently. AI voice agents structurally can. Mavrick is the AI coworker that operationalizes that math inside your Slack.
Inbound dial in under 60 seconds
Form fill → Slack alert in <1s → Mavrick voice agent dials → Cartesia 'Brandon' voice runs your discovery script → meeting booked → Slack handoff card posted. End-to-end median: 8–12 seconds to first ring.
Outbound at human-equivalent quality
Same voice stack, but for your outbound list. Mavrick dials, qualifies, drops a voicemail if no answer, and respects quiet hours per state via the regulatory database (50 states + DC + 5 territories seeded).
Slack-native handoff with full context
When a lead qualifies, Mavrick writes a structured handoff card to your sales channel: call transcript, qualification answers, next-step recommendation, CRM record link. Your human rep picks up where Mavrick stopped — zero context loss.
Real infrastructure, not vaporware.
Most "AI SDR" products are conversational wrappers around a single model. Mavrick is a complete shipping voice stack with the infrastructure named.
- Outbound carrier
Telnyx — ~$1.50/mo per DID + $0.005/min US outbound. About 70% cheaper than Twilio at scale. Shared DID pool with area-code matching for local-presence dialing.
- Voice rooms + agent runtime
LiveKit — handles the call audio + agent dispatch + SIP participant handoff. Cloud-hosted, no infra burden on you.
- Transcription
OpenAI Whisper — streaming STT during the live call. Sub-200ms latency on most inbound segments.
- Reasoning
GPT-5 with the workspace-specific discovery script + your cleared-hot qualification rules baked into the system prompt.
- Text-to-speech
Cartesia — default voice 'Brandon' (voice_id 5cad89c9-d88a-4832-89fb-55f2f16d13d3). 4 alternative voices A/B tested; Brandon won on lead-completion rate. Custom voice clones available on Pilot+.
- Compliance gates
National DNC + state DNC + workspace DNC + consent verification fail-closed before any dial. State quiet-hours table seeded for 50 states + DC + 5 territories.
- Concurrency safety
Per-workspace concurrency cap (5 simultaneous calls default), enforced by Postgres-backed slot reservation. No one workspace can starve the carrier capacity.
What @Mavrick looks like in a sales channel.
Mavrick vs the other AI SDR options.
| Tool | Speed | Channel | Slack-native | Approval arch | Pricing |
|---|---|---|---|---|---|
| Human SDR (alone) | 14+ hours avg | Voice + email | — | n/a | $70K–$110K/yr fully loaded |
| Apten | <60s | SMS only | No | No | $49/mo |
| Verse.ai | <2 min | Voice + SMS | No | Limited | ~$5K/mo |
| Drift | <60s | Chat only | No | No | ~$2.5K/mo |
| Mavrick | <60s | Voice + SMS + Email + Slack | Yes — architecturally | Yes — Constitution-enforced | Free + $50/mo |
Pricing observed mid-2026 on each vendor's public site. We'll update this table on every quarterly comparison sweep.
What we won't pretend Mavrick does.
- Enterprise outbound at 10,000+ dials/day per workspace. Concurrency cap is 5 simultaneous by default; we raise it on Enterprise but you should also evaluate dedicated dialers (Orum, Nooks) for high-velocity outbound.
- Legal compliance review. Mavrick blocks against National DNC + state DNC + workspace DNC fail-closed, but you remain the controller of record under TCPA / state telemarketing law. Review your own list legality.
- Multi-language conversations beyond English. Whisper supports multilingual STT, but our default scripts and quality bars are English-only as of mid-2026.
- Replacing senior AEs. Mavrick is built for inbound qualification + outbound dial-list grind. Discovery on complex enterprise deals, technical proof-of-concept calls, executive handshakes — those still need humans.
Questions sales leaders ask before installing.
What does Mavrick's AI SDR actually do?
+
Three things, end-to-end. (1) Outbound: dials prospects on your list via Telnyx using a Cartesia-cloned voice (default: 'Brandon'), runs a discovery script you approved, books the meeting on your calendar. (2) Inbound speed-to-lead: when a form fills on your site, Mavrick calls the lead in under 60 seconds — see /for/speed-to-lead for the full 14-year-old-problem context. (3) Handoff: posts a structured Slack card to your sales channel with the call summary, qualification notes, and next-step recommendation. Your human rep picks up where Mavrick stopped — no context loss.
How fast is 'sub-60-second' actually?
+
Median 8–12 seconds from form submit to Mavrick dialing, with the carrier ringing the lead's phone inside 30 seconds end-to-end. The cap is a deliberate 60s in case of cold-start latency on the voice-agent runtime. By comparison, the industry-average B2B inbound response time is ~42 hours (RevenueHero, 2024). Only 7% of B2B companies meet the 5-minute benchmark. Mavrick is structurally always-under-60s — humans can't compete on this axis.
What stack is Mavrick's AI voice agent built on?
+
Outbound carrier: Telnyx (~$1.50/mo per DID, $0.005/min outbound US — about 70% cheaper than Twilio at scale). Voice rooms + agent runtime: LiveKit. Transcription: OpenAI Whisper. Reasoning: GPT-5. Text-to-speech: Cartesia (default voice 'Brandon', voice_id 5cad89c9-d88a-4832-89fb-55f2f16d13d3). DID pool is shared across workspaces with area-code matching for local-presence dialing.
Does Mavrick replace my SDR team?
+
Not the goal. Mavrick is the always-on first responder — it picks up the call no human can pick up at 11pm or during your team's all-hands. When Mavrick qualifies a lead, it hands off cleanly to a human rep with full context. Most teams using Mavrick keep their senior AEs and let Mavrick eat the inbound-qualification grind that burns out junior SDRs.
What if my leads don't want to talk to AI?
+
Mavrick's voice agent identifies as AI when asked. If the lead pushes back, Mavrick offers to schedule a human call, sends an SMS with a calendar link, and posts the disposition to your Slack so your rep can follow up personally. The 'always pretend to be human' anti-pattern is explicitly banned by Mavrick's Constitution (see /constitution).
Does cleared-hot approval slow things down for sales?
+
Cleared-hot is a one-time script approval, not a per-call approval. You approve the discovery script and qualification criteria once. From then on, Mavrick executes consistently against the approved playbook. The approval gate kicks in only when Mavrick wants to mutate something downstream — book a meeting, send an email, update a CRM record. That's a one-click confirmation in Slack.
Which CRMs does Mavrick post handoffs to?
+
HubSpot, Salesforce, Pipedrive, Close, Attio, and any CRM exposed through the Pipedream Connect managed connector layer (3,200+ tools). Mavrick writes the call summary + disposition + next-step into the CRM record automatically. The same Slack handoff card includes a deep link to the CRM record for one-click human takeover.
How does Mavrick for sales compare to Apten, Verse.ai, Drift?
+
Apten is SMS-only and $49/mo. Verse.ai is voice + SMS but desktop-routed and $5K/mo. Drift is chat-only and ~$2.5K/mo. Mavrick is voice + SMS + email + Slack-native handoff, starts at free (10 missions), $50/mo Pilot. The unique combination: voice + Slack-native routing + cleared-hot approval architecture. See /for/speed-to-lead for the full comparison matrix.
Is the voice quality actually any good?
+
Cartesia is the leading voice-clone provider as of 2026 — most ElevenLabs-trained ears can't distinguish their output. Default voice 'Brandon' is a neutral US-male tone that's been A/B tested against 4 alternatives and won on lead-completion rate. Custom voice clones (your own founder's voice, for example) are available on Pilot and above.
What's the pricing?
+
Recruit (free): 10 missions to evaluate — covers ~5-8 outbound dials or ~15-20 inbound qualifications. Pilot ($50/mo): 20,000 credits ≈ 60-100 missions/mo. Enterprise: custom pricing for high-volume dial pools + dedicated DID provisioning + 99.9% SLA. See /pricing for the full breakdown. No per-seat fees on any tier.
The rest of the sales surface.
Stop losing 78% of leads to whoever responded first.
Start free — 10 missions, no credit card. Install Mavrick to Slack in 60 seconds. Connect your CRM. First inbound dial goes out in <60s after a form fill.