The platform
One agent. Four paths. Two kinds of user.
Peeve reads your live interface, decides how to resolve each question, and acts. For humans as motion, for their agents as tools. This page is the whole system; the links go deeper.
How a question moves through the system
The dashed line is the whole argument: every session scores the route it used and returns that to the Brain. Peeve is the only agent in this category that measures its own interface, because it's the only one that acts inside it.
How a question is resolved
Four steps, every time. And the last one is why it keeps working.
01
Invoke
A user asks in their own words, from inside your product. No keyword menu, no list of canned intents to pick from.
02
Climb in ~150ms
The Brain scores the question against your live route map and climbs the ladder to the lowest rung that resolves it (Answer, Show, Do, or Escalate), printing the competing confidence scores as it goes.
03
Act on the path
Answer with a source, show by moving the cursor, do it on your backend after confirming the exact effect, or hand off carrying the full session.
04
Measure
Every session scores the route it used and feeds that back to the Brain, so drift is caught early and the next answer is truer than the last.
The platform · six parts, one loop
Nothing here is a separate product.
The Brain
Every element stored by role and accessible name, re-verified every session, self-healing on rename.
Read more →Agent channel
The same capabilities as typed MCP tools, scoped per user, with the effect stated before it runs.
Read more →Voice
Spoken narration timed to the cursor, so the user learns the path instead of reading a caption.
Read more →Analytics
Real intent in your users' own words, and which route they stalled on.
Read more →Intent detection
Read more →Connectors
Capabilities map to your endpoints; hand-offs land in the helpdesk you already run.
Read more →Security
What Peeve reads, stores, and never sees, including your users' credentials, which it never holds.
Read more →Nothing here is a separate product. One registry feeds every surface, and one measurement loop keeps all of them true.
Install Peeve ↗One agent, many jobs
One install spreads across the work your team is drowning in.
The same agent deflects how-do-I tickets, onboards new users, runs self-serve account actions, captures and qualifies buying intent, serves your users' agents, and surfaces where people stall. No separate chatbot, tour builder, sales bot and analytics stack to wire together.
Why teams choose Peeve
One agent, not five tools
Answer, show, do, capture leads and hand off from one install. No separate chatbot, tour builder, sales bot and analytics stack to wire together.
Every session makes it better
The measurement loop keeps the whole system true, so it improves as your product changes instead of decaying.
Proof, not a sales call
Public pricing, a no-card trial, and a working demo: the product is the pitch.
14-day trial, no cardInstalls in an afternoonRead-only until you widen itKill switch in one commandCancel anytime, no lock-in