The agent fleet
Every agent working your deals, named.
Vendors sell "100+ AI agents" and hope you don't ask what they do. Here is our fleet — every card is a shipped system with a job, and the count is computed from this list, not a marketing round-up.
39agents in production today
In the meeting — live14
Running while you talk. One event stream feeds every surface: browser, phone, Watch, earpiece, sidepanel.
Cue agent
Composes the live cue — competitor mentions, MEDDIC gaps, buying signals, commitments — grounded in this account's history and dropped at the gate if the evidence isn't there.
Whisper answer
Say the wake word (or type silently) and ask; answers from your account history and knowledge base, in your ear or on screen.
Coach
Live delivery feedback — talk ratio, monologues, filler — kept quiet unless it matters.
Champion tracker
Watches who on the buyer side is leaning in and flags champion moments.
Disengagement watch
Detects the buyer checking out — energy drop, short answers — while there's still time to change course.
Morale read
Tracks the emotional temperature of the room across the call.
Price sensitivity
Flags pricing-pressure language the moment it appears.
Prep briefing
Pulls the account brief, prior commitments, and attendee context into the workspace as the call starts.
Prior-meetings recall
Loads what this buyer said in earlier calls so continuity cues can cite it.
Voice profile matcher
Suggests which contact is speaking from consented voice profiles — you confirm, it never auto-applies.
Name dossier
Builds who's-who context on named participants as they join the conversation.
Todo capture
Hears commitments and drafts the to-do while the sentence is still warm.
Meeting Q&A
Answers mid-call questions against the transcript so far.
Transcript refiners
Two agents (ASR + diarization) continuously repair the live transcript — words and who-said-what.
After the call9
Every capture path — live, bot, phone import, Otter/Gong/Plaud — lands in the same pipeline with the same outputs.
Wrap-up summary
The MEDDIC summary, agreed action items (commitments only, never inventions), and a draft follow-up email.
Quality guardian
Judges every wrap-up and runs a targeted repair pass when it's below the floor — the badge tells you which happened.
Retranscribe
Re-runs the audio at replay quality so the saved transcript beats the live one.
Intel extractor
Pulls durable account/contact/deal intelligence out of the transcript into structured notes.
Coaching scorecard
Grades the call against the meeting mode's rubric — MEDDIC, demo, kickoff, QBR.
CRM write-back
Drafts the CRM updates as proposals; you confirm, it files — to any of eight CRMs or the built-in one.
Review analyst
The deeper post-call read — patterns, risks, and what to do next on this deal.
Loss-reason classifier
When a deal closes lost with no reason, classifies it from the evidence — and never overwrites what the rep wrote.
Knowledge ingest
Feeds the transcript into your private, per-tenant memory so future calls get smarter.
Guardians — reliability4
Agents that watch the system itself. A meeting is unrepeatable; these exist so nothing silently fails.
Run guardian
Watches every wrap-up stage with a heartbeat; a stalled stage gets one recovery attempt and an honest status, never a spinner.
Struggle guardian
Notices a rep fighting the UI — rage clicks, error loops — and offers help, consent-gated and rate-capped.
Ambient guardian
Anticipates what you'll need next — warming briefs and context before your next meeting.
Health canaries
Continuously probe the LLM path, STT fleet, and golden paths; three failures page a human.
Pipeline + market12
Working your book between the calls.
Forecast category assigner
Assigns commit / best case / pipeline from evidence — MEDDIC completeness and meeting recency, not rep optimism — with the reason on every row. You can always override.
Signals discovery
Finds live buying triggers in your market from news, filings, hiring, and the open web.
Signals QA judge
Reviews every lead slate before it reaches a customer and sends the pipeline back to fix what it flags.
Auto-tuner
Proposes bounded improvements to the signal pipeline from real outcomes — a human applies or rejects every one.
Intent scoring
Blends external signals into a per-account intent score with the contribution of each signal shown.
Factual signal engine
Watches hard numbers — filings, job velocity, permits, traffic — and refuses to narrate a number it can't cite.
Competitor watch
Diffs competitor sites daily against our positioning and surfaces what changed.
Person brief
The 5-minute relationship brief on any contact — your history, their public footprint, the hooks.
Today brief
One synthesis of your day — meetings, tasks, signals — on the dashboard and in a morning email.
Local briefing
Pre-call local-market context for accounts with a meeting in the next 24 hours.
Cadence drafter
Drafts outreach in your voice, adapted to the recipient's personality — draft-only, nothing auto-sends.
Stalled-deal sweep
Flags deals sitting past your own historical pace, to you before your manager.
How to read this page
- Every card is a shipped system you can see working in the product — not a roadmap item, not a renamed feature counted twice.
- Agents that write anywhere (CRM, email, outreach) are draft-first: a human confirms before anything leaves the building.
- Recording and voice features are consent-gated, enforced server-side.
- The count at the top is computed from this list at render time. When we ship an agent, the number moves; it is not a marketing round number.
Run your next meeting with a wingman.
Free to start. Your own accounts, your own market, your next real call.
Start free now