Agentic CRM · v3 closed beta · 24/7 on-duty

An agentic CRM. Agents propose the moves. You take the meetings.

Client Path runs the full deal lifecycle on a visual journey map. Research, qualification, outreach drafts, follow-ups, renewal watch. Every agent move is governed by a rule you control — auto-execute or human-gated. Every action audited.

Start free trial See how it works

14-day trialNo credit cardSOC 2 Type II

deald_4f2a · stage 3 / 5

ACAcme Corp · Sales-Intelexpansion · S. Patel (VP RevOps)4 agents on duty

0.0

Win probability

Deal value

$0,000 ARR

Agent activity · last 3h5

Drafted follow-upto S. Patel · awaits review2m

Advanced to Proposerule: score_crossed ≥ 80 · auto11m

Surfaced funding signal$40M Series B · Insight Partners47m

Re-ran researchv3 · 18 citations1h

Pinged owner on Slackrenewal · Q3 expansion3h

DiscoverQualifyEngageProposeClose

01How it works

Three moves from offering catalog to a pipeline that runs itself.

The agents aren't a sidebar widget. They have authority. They propose stage advances, draft outreach, and surface signals. You stay in the loop with a confirmation policy you control per rule.

Define what you sell

Author offerings, products, and rubrics. Map cross-sell, prerequisite, and renewal edges between them. The agents read this as their playbook.

3 offerings12 productsrubric · v2

Set agent rules

Declare triggers (score crossed, stage dwelled, signal event) and what the agent should do. Choose confirmation or auto-execute per rule.

6 rules · activeauto-execute · 2human-gated · 4

Watch the journey map

Agents surface signals 24/7, draft outreach, and propose stage moves. You see the canvas, approve from the inbox, and take the meetings.

lens A · dealslens B · clientsinbox · 3 pending

02What you get

A CRM where the agents have the authority.

No fields-and-forms dashboard. No bolt-on AI summarizer. The agents propose moves, draft outreach, manage your catalog and quotes, and audit themselves while doing it.

Two lenses on one canvas. The dashboard isn't a list of forms. It's the deal in motion.

Lens A · Deal Journey shows the next move from discover to close. Lens B · Client Trajectory zooms out to three years of activity across every offering you've touched on that account. Both render on the same xyflow canvas — a clear view of every deal, at every stage.

Lens A · Deal journey · 86 active Switch lens

Discover38

Vector Forge

$24k

Quay & Co.

$18k

Qualify21

Northwind

$54k

Ember Bio

$36k

Engage14

Acme Corp

$84k

Helix.io

$42k

Propose9

Beacon Labs

$120k

Close4

Lattice North

$66k

Acme Corp · agent dwelt 6 days at Engage, proposing advance to Propose.

03A deal in motion

One deal. Every move the agent has made. Every move it's about to.

This is the deal slide-over your team lives in. Stage progression, products, the live quote, agent activity, operator notes, all in one place. Pull it open from any card on the journey map.

Acme Corp · Sales-Intel · Expansion

d_4f2a · owner @sarah · BDR @priya · 230 FTE · Austin, TX

4 agents on duty

DiscoverQualifyEngageProposeClose

Activity timeline

Drafted follow-up email

subject: 'Series B + your Q3 hiring' · awaits review

agent · drafter

2m ago

Discovery call booked

S. Patel (VP RevOps) · Tue 3:00 PM CT · 30m

agent · scheduler

11m ago

Surfaced funding signal

Acme closed $40M Series B · Insight Partners

agent · watcher

47m ago

Advanced Qualify → Engage

rule · score_crossed · win_prob ≥ 80

rule · auto

1h ago

Re-ran research · v3

18 citations · $0.014 · Sonnet

agent · researcher

1h ago

Operator note

Sarah: 'Owner prefers Slack DM; CTO is the real blocker on technical eval.'

sarah@

Yesterday

Briefing snapshot

Acme closed a $40M Series B in March, led by Insight Partners[1], and is tripling AE headcount this year[2]. The agent surfaced three buying signals in the last 24h. Re-research and propose advance are queued pending your review.

Win probability

87.4

stage · Engage · 6 days

Forecast (weighted)

$84,000ARR

12 seats · onboarding · metered LLM

Next moves · queued

Advance to Proposereview

Send follow-upreview

Re-run research · new signalauto

04Compared

Traditional CRM, agentic CRM, AI-add-on. The honest version.

The old CRMs are a database your reps update. AI add-ons summarise the database. An agentic CRM does the work, and shows you what it did.

	Client Pathagentic · 24/7 · audited	Traditional CRMsalesforce · hubspot · pipedrive	AI add-oncopilots glued on the side
Primary surface	Visual journey map + Lens A/B canvas	Form-and-field record pages	Form-and-field with a chat sidebar
Agent authority	Proposes moves · drafts outreach · per-rule policy	No	Partial
Confirmation policy	Per-rule · auto or human-gated	N/A	Global · all or nothing
Full-lifecycle on one canvas	Discover → expand → renew	Spread across tabs	No
Product / quote management	Native · with cross-sell edges	Native (CPQ add-on)	No
Audit log for every agent action	Rule · prompt · cost · cited	Manual activity logging	Partial
Re-research on signal events	Triggered + diffed	No	No
On-prem / self-hosted LLM	Enterprise · Ollama · Bedrock	N/A	No

Primary surface

Client Path
Visual journey map + Lens A/B canvas
Traditional CRM
Form-and-field record pages
AI add-on
Form-and-field with a chat sidebar

Agent authority

Client Path
Proposes moves · drafts outreach · per-rule policy
Traditional CRM
No
AI add-on
Partial

Confirmation policy

Client Path
Per-rule · auto or human-gated
Traditional CRM
N/A
AI add-on
Global · all or nothing

Full-lifecycle on one canvas

Client Path
Discover → expand → renew
Traditional CRM
Spread across tabs
AI add-on
No

Product / quote management

Client Path
Native · with cross-sell edges
Traditional CRM
Native (CPQ add-on)
AI add-on
No

Audit log for every agent action

Client Path
Rule · prompt · cost · cited
Traditional CRM
Manual activity logging
AI add-on
Partial

Re-research on signal events

Client Path
Triggered + diffed
Traditional CRM
No
AI add-on
No

On-prem / self-hosted LLM

Client Path
Enterprise · Ollama · Bedrock
Traditional CRM
N/A
AI add-on
No

05Pricing

Pay per seat. Pay for the LLM you use. Nothing else.

Seat pricing covers the product. Pipelines, journey map, agent rules, quote management. LLM cost is metered through at provider rates with a 5% margin, fully audited in the activity log. Annual commits get a 22% seat discount and up to 18% on usage.

Starter

For founders and small RevOps teams running their first agentic pipeline.

$29/ seat · month

Billed annually · 12-month commitment

Start free trial

What's included

Up to 3 seats

1 active pipeline · 50 open deals

1 offering with full product catalog

Journey map · Lens A only

3 agent rules · human-gated

CSV intake, CSV / quote PDF export

Operations

Haiku tier default (Sonnet on-demand)

Community Slack support

+ Metered LLM cost · pass-through

Every agent action shows its model tier, token count, and dollar cost in the activity log. We bill at provider rates with a 5% margin. Enterprise plans on on-prem infrastructure are quoted separately. Per-rule and per-run cost caps are hard, not advisory.

Annual contracts: usage discount up to 18% at committed volume.

Haiku · default tier~$0.004 / agent action

Sonnet · standard~$0.02 / agent action

Opus · deep research~$0.18 / agent action

On-prem · EnterpriseCustom · contact us

06Stack

It plugs into the tools your team already runs.

LLM tier, search backend, notifications, billing. All swappable. Roll your own MCP server and connect a private data source in a config file. CRM sync and email integrations are on the roadmap.

Anthropic

LLM

OpenAI

LLM

Ollama

LLM · on-prem

Azure OpenAI

LLM

AWS Bedrock

LLM

Google Vertex

LLM

Tavily

Exa

Logo.dev

Logos

Crunchbase

Signals

BuiltWith

Tech-stack

People

Roadmap

Salesforce

CRM sync

Roadmap

HubSpot

CRM sync

Roadmap

Pipedrive

CRM sync

Roadmap

Gmail

Roadmap

Outlook

Roadmap

Google Cal

Calendar

Slack

Notify

MS Teams

Notify

Zapier

Webhook

Stripe

Billing

DocSign

E-sign

</>

MCP server

Custom

07Questions

The plain answers.

No. An agentic CRM treats the agent as a first-class actor that has the authority to advance deal stages, draft outreach, and trigger re-research. It isn't a chat panel that summarises records. Every agent move is governed by an AgentRule with its own confirmation policy (auto-execute or human-gated) and lands in your inbox if it needs review. You always see what the agent did, why it did it, and what it cost.

Those products are field-and-form CRMs with a chat copilot bolted on the side. The primary surface is still a record page. Client Path's primary surface is a journey-map canvas, and the data model is built around deals, products, quotes, activities, and agent rules from day one. Client Path is purpose-built as an agentic system, not an AI layer added after the fact.

Whatever you tell them to. Each AgentRule has a confirmation policy: 'auto-execute' lets the agent act immediately (advance a stage, re-run research, log an activity, draft a follow-up), while 'review' creates a pending AgentTask in your inbox. Mutating actions like sending external email or sending a quote are review-by-default. You can change the policy per rule, per account, or globally.

Lens A is Deal Journey, a kanban-style canvas of one deal moving from discover to close, with agent activity visible at each stage. Lens B is Client Trajectory, zoomed out to three years of activity across every offering you've ever touched on that account. Toggle between them with a lens chip; both live on the same canvas.

Yes. Offerings define what you sell at the strategy layer (ICP, rubric, methodology). Products are the SKUs that go on a quote, billed one-time, recurring, or by usage. Quotes are immutable snapshots; line items on a deal stay mutable until you send. Cross-sell, prerequisite, and renewal edges feed the agent's next-move suggestions.

Enterprise plans support on-prem Ollama, Azure OpenAI, and AWS Bedrock so every agent action stays within your infrastructure. Starter and Team plans run on Anthropic via the Vercel AI Gateway.

On any deal you can override the stage, pin a score, freeze a forecast, or pause the agent on an account. Every override is an Activity row with rationale; future agent runs respect it and show a diff against the model's current view. Operator notes on clients are fed back into planner and synthesizer prompts on every re-run.

14 days. No card required. A generous trial to wire a real pipeline end-to-end and see the agent rules in action. After trial, the lowest seat price is $29 / seat / month on an annual plan.

08 Get started

Let the agents work your pipeline tonight.

Wire one offering, set three rules, and watch the journey map light up. By tomorrow morning your AEs will walk into meetings that the agent already prepped for them.

Start free trial Talk to sales

14-day trial·No credit card·SOC 2 Type II