A team of AI agents that prospects, qualifies, and closes for you.Try it free →
✦ AI AGENTS

A team of digital SDRs that never sleeps

4 agents that prospect, qualify, score, and close for you. Train them in 5 minutes and let them run.

Try it free →See the agents
Try it free for 30 days
Active
Screener
Filters out what doesn't fit
Claude Opus 4.8
Active
Scorer
Scores by fit
GGPT-4.1
Active
Prospector
Writes 1-to-1
Claude Opus 4.8
Active
Closer
Books the meeting
Claude Opus 4.8
The pipeline

Four agents, one flow

Each agent is a step: filter, score, write, and close. Chained together, they turn a cold list into booked meetings.

01 · SCREENER

Filters out who doesn't fit, before spending a message

Clean your lists against your ICP. Every lead, one decision: accepted or discarded, with the reason.

accepted/discarded decisionCriterion by criterionEnriches if data is missing
✦Screener AgentActive
BISH1H2•</>{}
Preview
You are a B2B lead qualification engine for a
LinkedIn outbound campaign.
You receive an ICP and a lead profile, decide if
the lead matches, and explain why.
DECISION RULES
- Any mandatory criterion that does NOT match
→ "discarded".
- All mandatory criteria match → "accepted".
OUTPUT (JSON only):
{ "decission": "accepted" | "discarded",
"criteria_breakdown": [...] }
Changes save automatically · ~383 tokens · 1,530 characters
Live scoringby ICP fit
H
Holded
new CRO · +3 signals
9HOT
F
Factorial
hiring 9 AEs · Snowflake
8HOT
C
Cabify
medium signal · 1 match
6WARM
02 · SCORER

Knows where to start before you do

Scores each lead from 1 to 10 by fit with your ICP, so you go after what converts most first.

Score 1–10Weighting by criterionPrioritizes your queue
03 · PROSPECTOR

1-to-1 messages that read like they were written by hand

It reads the lead's real profile and writes a unique, one-of-a-kind message. No templates, no "hope you're doing well."

Personalization from the profileDynamic variablesMulti-touch
Draft · Marta Pérez
Head of Mkt · Landbot

Marta, I saw your launch of the new automation suite at Landbot. Building multichannel outbound without growing the team is exactly what we solve. Want me to show you how a team your size does it?

{{lead_name}}{{lead_headline}}{{lead_company}}{{icp_definition}}
Tools:ProfilePostsCompanyGoogleWeb Scraper
I'm interested, but right now we're short-staffed…
That's exactly why it fits: an agent builds it, not your team. Does a 15-min call on Thursday at 3pm work for you?
Perfect, Thursday works for me 👍
Meeting booked → Thursday 3pm
InactiveTrainingAutopilot
04 · CLOSER

Replies to inbound and books the meeting

When the lead replies, the Closer keeps the conversation going and books the meeting. In Training mode (you review) or Autopilot (replies on its own).

Training / AutopilotHandles objectionsMeeting booked
NATIVE MCP

Your agents and your prospecting tools, inside Claude

WaLead exposes a native MCP server (mcp.walead.ai): connect Claude and trigger searches, enrichment, and campaigns with your own prospecting tools, without leaving the chat.

Search and enrich leads from the chat
Launch campaigns without opening the app
Your native tools, exposed as functions
ClaudeMCP · walead
Find me 20 SaaS CMOs in Madrid and enrich their email.
walead.search_leads()
icp: "CMO · SaaS · Madrid"
enrich: ["email"]
20 leads · 18 verified emails
Launch the campaign to the HOT ones.
walead.launch_campaign()
Campaign active · 42 leads
Models

A different model for each job

Assign each agent the model that performs best, managed by us or with your own API key (BYOK).

Managed by WaLeadBYOK · your own keyCredits per run
Agent brain
Claude Opus 4.8
Anthropic · Managed
✦10 cr. /run
GPT-4.1
OpenAI · Managed
✦0.5 cr. /run
Grok 4 Fast
xAI · Managed
✦2 cr. /run
Your credentials (BYOK)
sk-ant-•••••••••••••••••3f2Connected
No code

Full control, in 5 minutes

Train each agent from a four-tab panel. No code, no waiting on anyone.

# Role
You are the Screener on a B2B GTM team in Spain.
# Task
Evaluate each lead against the ICP and decide accepted / discarded.
# ICP
Sector: {{icp_definition}}
Size: 50-500 · Role: C-level / VP
{{icp_definition}}{{lead_company}}{{lead_headline}}Markdown + variables
FAQ

Frequently asked questions

What does each agent do?+

They're four agents chained as a pipeline: the Screener filters out who doesn't fit your ICP, the Scorer scores those who do from 1 to 10, the Prospector writes personalized 1-to-1 messages, and the Closer replies to inbound and books the meeting.

How long does it take to train them?+

Less than 5 minutes and no code. You define the instructions in Markdown with variables, turn on the native tools you want, and, if needed, upload context documents. The agent is ready to run.

What is native MCP in Claude?+

WaLead exposes a native MCP server (mcp.walead.ai) that connects directly to Claude. From the chat itself you can trigger searches, enrichment, and campaigns with your prospecting tools, without leaving the conversation or opening the app.

Can I choose the model for each agent?+

Yes. You assign each agent the model that performs best for its job (Claude Opus, GPT-4.1, Grok…), managed by us with a cost in credits per run, or with your own API key (BYOK).

Deploy your team of agents today

They prospect, qualify, and close while you focus on selling.

Try it free →
Try it free for 30 days