Unified customer profile · Customer 360 for AI agents

A unified customer profile that AI agents actually read.

Behavior, conversations, CRM state, enrichment. One profile per lead, updated in real time. Qualifier, Support, Engagement, and Booking agents read from it before they type the first message. No context hunting, no re-asking.

+15ppDemo conversion uplift (40% → 55%)
−70%Sales prep time per lead
<5%Repeated questions by agents
Sources
W
Website
JS · pages · forms
C
Chat
live messages
M
Messengers
WhatsApp · Telegram
@
Email
opens · replies
{}
API / CSV
custom events
CRM
HubSpot · Pipedrive
Lead Profile
LW
Lindsey Welch
CMO · Tier A · MQL
pricingHubSpotdemo
Live · 3ms ingest
Consumers
Q
Qualifier
AI agent
S
Support
AI agent
E
Engagement
AI agent
B
Booking
AI agent
Sales
Lead Card UI
CRM sync
User Summary
The problem

Most AI agents are working blind

An LLM hooked up to a chat widget has no memory of your lead. Your data lives in silos: pageviews in analytics, emails in the CRM, chats in another tool. Without a single source of truth, the CMO who visited pricing five times gets the same opener as a first-time visitor. Here's what that costs you every day.

01 · Repetition

Agents ask the same questions you already answered

The lead typed her role into your demo form yesterday. Today your agent asks her again. She closes the tab.

GENERIC AGENT SAYS"Hi! Could you tell me your role and company size?"
02 · Context blindness

A 5× pricing visitor is treated like a new arrival

Your agent has no idea the lead already evaluated your product for two weeks. Instead of closing, it opens with a generic welcome.

GENERIC AGENT SAYS"Hi 👋 What brings you to our site today?"
03 · Sales scramble

Your reps waste 10 minutes hunting for context

CRM, GA, chat history, email threads. Five tabs. Constant context switching. Ten minutes. Every single call.

SALES REP BEFORE THE CALL"Hold on, let me pull up the CRM… and the chat transcript… one sec…"
04 · Cold transfer

Hot leads cool down during handoff

The agent qualified them five minutes ago. By the time Sales reads the transcript and calls back, the lead has moved on.

GENERIC AGENT SAYS"Thanks! Someone from our team will reach out shortly."

Every one of these is a data problem, not a model problem. Lead Profile fixes all four.

See the solution
The solution

One customer profile. Every agent. Real-time.

One profile. Read by every agent. Dashly Lead Profile is the unified customer profile your Qualifier, Support, Engagement, and Booking agents all read from. A full customer 360 view: behavior, conversations, CRM state, and ICP match from every touchpoint, merged into a single source of truth. When one agent learns something, every agent knows it.

LW
Lindsey WelchONLINE
lindsey@wowinfluencer.com+1 (212) 482-6632
💬 Chat• Pop-up✉ Email⚡ Push
Properties
Location & local time
📍 New York, USA
13:45
Current page
/integrations/hubspot
For AI agentsUser Summary
Intent
🔥 High
Stage
ready_to_buy
Lead Tier
A · 92
MQL Source
AI Qualifier
Role
CMO
Industry
SaaS · 100 empl.
Budget
$50k+
Goal
Lead qualif.
Suggested approach
Skip qualification, already MQL. Focus on HubSpot integration ROI. Push to demo this week.
Events
TimelineGrouped+ Add event
22.04.202613:42:08
Page Viewed
/integrations/hubspot
22.04.202611:08:42
Demo: form completed
email · phone · company
22.04.202611:05:17
Marked as MQL
source: AI Qualifier
22.04.202610:58:11
Quiz finished
role=CMO · budget=$50k
22.04.202610:52:04
AI conversation closed
12 messages · webchat
21.04.202616:14:33
Page Viewed
/pricing
19.04.202609:30:01
First session
utm: linkedin · cmo_q2
Conversation history
Only conversations with the user
W
webchat
12 messages · 14-22 Apr
Hi! Does Dashly integrate with HubSpot? Q1 timeline.
11:09
Yes. Bi-directional sync, scoring fields included.
Great. Pricing for ~50k visitors/mo?
11:10
Let's jump on a call. I'll walk you through ROI for your size.
Open conversation →
What it's made of

Three pillars. One context layer

The Lead Profile stands on three distinct engineering systems. Each is production-grade on its own. Together they give your agents the one thing most AI chatbots never have: real memory.

Pillar 01

CDPB2B customer data platform

The B2B customer data platform underneath everything. A CDP for B2B teams that collects and normalizes every signal from every touchpoint, from a page view to a CRM paid status, into a canonical schema.

Sources
JS, API, CRM, chat, email, WhatsApp, Telegram, CSV
Schema
30+ standard events, 20+ standard properties
Examples
$mql · $crm_paid · $quiz_finished
Freshness
Real-time CDP, updated on every event
Role
Source of truth. Fuels everything else.
Pillar 02

AI User SummaryWhat the lead did

On every key action, a dedicated agent re-analyzes the lead's full history and rebuilds a structured summary: intent, stage, readiness signals. Synced into your CRM and read by every agent.

Source
Dedicated analyst agent, watching CDP for trigger events
Contains
Interests · Readiness signals · Source · Journey stage
Distribution
Synced into CRM · Read by every other agent
Example
Pricing 3× + demo requested → stage: ready_to_buy
Role
The active analyst inside your data.
Pillar 03

AI Conversation InsightWhat the lead said

Structured memory of every chat (topics, questions, explicit signals) across web, WhatsApp, and email. Agents always know what was already discussed.

Source
Transcripts across all channels (AI + human)
Contains
Topics · Questions · Explicit signals · Last context
Updates
Keywords real-time + LLM on $ai_communication_closed
Fallback
Regex keyword extraction if LLM times out
Role
Conversational memory layer of the agent's context.
CDP is the fuel. User Summary is what the lead did. Conversation Insight is what the lead said. The full profile is in the agent's context from the first message.
What changes for AI agents

From generic openers to the right message, every time

Same lead. Same chat widget. With Lead Profile in context, every agent opens the conversation at the right level, with the right next step.

QUALIFIER AGENT
Q

Skips quiz re-runs for already-qualified leads

quiz_completeddemo_requestedmql = truerole = CMO
"I see you already completed the quiz and requested a demo. Let's find a time that works?"
SUPPORT AGENT
S

Remembers last conversation's topic

topics: webhooksquestions_asked: setupreturning_visitor
"Welcome back. Continuing with webhook setup, or a new question?"
ENGAGEMENT
E

Personalizes the opener with real context

pricing_visits: 3source: LinkedIn Adsinterest: integrations
"Saw you're comparing pricing. Want a 2-min ROI walkthrough for teams your size?"
What changes for Sales

From 10 minutes of context-hunting to 3 seconds

Your reps stop living in five tabs. The Lead Card shows what matters: who the lead is, what they've done, what they've asked, and the recommended approach for the next call.

01
Tier scored automatically. A/B/C based on role, industry, company size, budget.
02
Full behavioral timeline. Pages visited, quizzes completed, forms submitted.
03
Conversation excerpts. Exactly what the lead asked in chat, surfaced and structured.
04
Suggested approach. A one-liner that tells the rep how to open the call.
LW
Lindsey Welch
CMO · WOW Influencer · SaaS · 100 empl.
TIER A
Interests
PricingHubSpot integrationIntegrations
Readiness signals
Demo requestedQuiz completedMQL
Topics discussed
IntegrationPricingTimeline: Q1
Last activity
Asked about HubSpot integration · 2h ago
Suggested approach
High-intent lead, ready for demo. Focus on HubSpot integration ROI. Target Q1 start.
NameEmailLast activity
DWDiana Walsh
UnknownSep 19, 2020
JMJames Mitchell
Sep 19, 2020
SCSarah Chen
Sep 19, 2020
RTRyan Torres
UnknownSep 19, 2020
LHLaura Henderson
Sep 19, 2020
MKMarcus Klein
UnknownSep 19, 2020
ATAshley Turner
UnknownSep 19, 2020
NBNathan Brooks
UnknownSep 19, 2020
OWOlivia Ward
UnknownSep 19, 2020
CEConnor Evans
Sep 19, 2020
Segmentation

Flexible customer profile segmentation

Every lead lives in one place. Segment users into any group by website data, CRM state, or external services. Segments update automatically in real time, so your campaigns, chat blasts, and agent routing always hit the right people.

  • By website behavior: pages, sessions, time on site
  • By CRM data, payment systems, product usage
  • By UTM tags, traffic sources, campaigns
  • Any condition combinations with AND / OR
The honest part

You could build this. Here's why most teams shouldn't

We get it. Your team is technical. A weekend with Claude, some glue code, a webhook or two, and you've got a working prototype. That prototype is 10% of the work.

"An LLM is 10% of an AI agent. The other 90% is the data infrastructure that makes it smart."
Production requirement
DIY with Claude + glue code
Dashly Lead Profile
Real-time profile freshness under load
Design an event bus. Handle race conditions across millions of daily events. Make event ingestion fast enough that new behavior shows up in the profile within milliseconds, not seconds. In a live conversation, every millisecond between the lead's action and the agent's next message is noticeable lag.
Built-in. 3ms event ingestion, profile updated in real time.
Conversation analysis with fallback
Build an LLM analysis pipeline. Handle timeouts. Extract topics, questions, signals. Write a regex fallback when the LLM fails.
Triggered on conversation close. Regex fallback included.
CRM data enrichment, bidirectional
Write connectors for HubSpot, Pipedrive, Salesforce. Handle field mapping, rate limits, webhook retries, and API versioning forever. Then build the reverse flow: enriched profile attributes back to CRM.
22 production integrations, maintained by us. HubSpot and Salesforce lead enrichment included out of the box.
Unified schema across channels
Design an event taxonomy. Enforce it across web, WhatsApp, Telegram, email. Version it. Document it. Debug it.
30+ standard events, one schema, every channel.
Profile delivery to agents
Build a getUserProfile equivalent. Add caching. Guarantee consistency. Keep the tool schema stable for the agents calling it.
Native tool. <500ms cached. Stable contract.
Observability at production scale
Dashboards for coverage, latency, accuracy, LLM usage, event lag. Alerts. Runbooks. On-call rotation.
Included. Dashboards out of the box.
Time to production
6–12 months with 2–3 senior engineers. Ongoing maintenance forever.
2–4 weeks. No dedicated infra team needed.

Your team's edge is your product, not data infrastructure. We spent 7 years building this so you can spend 2 weeks shipping agents.

Book a demo
Data foundation

Your customer profile is only as good as the data feeding it.

Four ingest paths cover every source you've got. Website behavior, product events, CRM state, historical contacts. Everything lands in a single canonical schema. CRM data enrichment runs bi-directionally: MQL status, tier, role, and industry write back to HubSpot or Salesforce the moment an agent resolves them.

JS

JS snippet

One line on your site. Basic visitor data starts collecting immediately. Extra events via wizard, no developer needed.

22+ integrations

HubSpot, Pipedrive, Salesforce, WordPress, Zapier, Make and more. Ready-made connectors, no code required.

{}

REST API & SDK

Send custom events and properties directly from your product or backend. Full control over your schema.

CSV / database import

Bring in historical contacts with full attribute history. Seed the profile from day one.

FAQ

Questions about Lead Profile

Data sources, how the profile feeds agents, setup time, and how this is different from a CDP or CRM.

Book a demo

A unified customer profile combines behavior, conversations, CRM state, and enrichment data from every touchpoint into a single record per lead. Dashly Lead Profile is built as the customer profile your AI agents read at conversation start, so every Qualifier, Support, and Booking agent works off the same real-time context.

Every AI agent in Dashly (Qualifier, Support, Engagement, Booking) reads from the same Lead Profile via a native tool call. The full profile lands in the agent's context at conversation start: properties, behavioral events, prior conversation topics, and a suggested approach. Profile reads are cached under 500ms.

Lead Profile ingests data from a JS snippet on your website, 22+ native integrations (HubSpot, Pipedrive, Salesforce, Zapier, Make), a REST API and SDK for custom events, and CSV or database import for historical contacts. Every source writes into the same canonical schema of 30+ standard events and 20+ standard properties.

Lead Profile is built on a B2B customer data platform layer and extends it. A CRM stores deal state. A CDP for B2B stores raw events. Lead Profile sits on top of both and acts as the context layer AI agents read from. It combines the CDP layer (raw data), AI User Summary (what the lead did and why they matter), and AI Conversation Insight (what the lead said in prior chats) into a single structured object. Think of it as a customer 360 view shaped for AI consumption, not a dashboard for humans.

Yes. HubSpot lead enrichment and Salesforce lead enrichment are built in, bi-directional. Field mapping covers contacts, companies, deals, and custom properties. CRM data enrichment writes back on every state change: role, industry, lead tier, MQL status, and conversation summary all land in your CRM so sales reps see the same data the AI agents saw during qualification.

Event ingestion takes 3ms. The profile is updated in real time on every event, so agents have the freshest data even mid-conversation. If a lead visits /pricing while the agent is already chatting, the next message can reference that visit.

Yes. This is what the Qualifier agent is built for. AI lead qualification runs against the full profile: role, budget, goal, fit. The agent writes MQL status back, then hands off to the Booking agent or a sales rep. Tier scoring (A/B/C) is automatic based on role, industry, company size, and budget. You can layer custom scoring on top through the property API. Compared to standalone lead qualification software, you get the qualification, the enrichment, and the booking in one chat.

For most B2B teams, yes. Dashly combines AI lead scoring with qualification and booking in one layer. Every lead gets an automatic tier (A/B/C) the moment the profile has enough signal. You can replace separate lead scoring software by defining your own rules and thresholds through the property API, or let the AI agent derive scores from conversation and behavior. The key difference from point solutions: the score lives in the same profile the Qualifier and Booking agents already read from, so there's no separate sync to set up.

A customer 360 platform is built for human analysts: dashboards, cohorts, segment exports. Dashly Lead Profile is shaped for AI consumption. The same underlying data (behavior, CRM state, enrichment, conversation history) but delivered as a structured object an AI agent reads in 500ms at conversation start. You still get segmentation and dashboards. The difference is that the profile is the primary interface, not a report.

Most teams go live in 2 to 4 weeks. Day one covers data source connection (JS snippet, CRM, API). The profile starts building immediately. From there it's one to three weeks to configure and ship your first AI agent. No dedicated infrastructure team needed.

Get started

Your agents will know exactly who they're talking to

Lead Profile is live for every Dashly plan. Connect your sources, deploy an agent, and watch your first context-aware conversation happen within two weeks.

No credit card required · Setup in under 10 minutes

What happens next
  1. 01
    We connect your data sources. JS snippet, CRM, API, whichever applies. Takes under a day.
  2. 02
    Profiles start building automatically. Every visit, event, and conversation enriches the lead's context in real time.
  3. 03
    We configure your first agent together. Qualifier, Support, or Engagement. Whichever moves your pipeline fastest.
  4. 04
    You ship to production. Average time from first conversation to live: 2–4 weeks.