← ALL BLUEPRINTS / AGENT BLUEPRINT
Prospecting agent for B2B financial services sales teams on Salesforce
The Prospector finds in-market accounts that match your ICP, researches and enriches the right buyers, drafts compliance-reviewed personalized outreach, and routes engaged prospects to reps with full context in Salesforce. The result is more qualified meetings booked and more pipeline created, without reps spending their week on list building and first-touch emails.
| ROLE | SALES | PLATFORM | CUSTOM |
|---|---|---|---|
| INDUSTRY | FINANCIAL SERVICES | ||
| DRAWN BY | HACKING DEMAND | ISSUED | OCTOBER 19, 2026 |
What The Prospector does
The Prospector works for the sales team at a 51 to 200 person financial services firm that sells to businesses (for example payments, treasury, lending, or wealth and benefits platforms sold to finance leaders at mid-market companies). It owns the top of the outbound funnel: maintaining the ICP and persona definitions, watching for buying signals, researching and enriching accounts and contacts, drafting personalized first-touch and follow-up sequences inside approved messaging frameworks, and handing warm prospects to the right rep with a research brief. Salesforce stays the system of record. Reps own conversations once a prospect engages, and compliance owns the approved language library the agent writes from. Because the firm wants it built and kept running, the agent ships with monitoring, a weekly tuning cycle, and a named owner on the operator side.
Custom build
The firm runs Salesforce, so a custom Claude build orchestrated with n8n is the better fit than HubSpot Breeze, which is designed around HubSpot data. Claude handles the judgment-heavy work (ICP scoring, research synthesis, persona-specific writing, reply classification) and can be constrained tightly to an approved messaging library, which matters in a regulated industry. n8n connects Salesforce, Warmly, Cognism, FullEnrich, Salesloft, and Chili Piper, and can be self-hosted to keep prospect data in the firm's own environment. Salesloft keeps sequencing, human approval, and activity logging inside a Salesforce-native workflow so every touch is recorded for audit and retention.
What starts it
- A target account shows a buying signal (repeat visits to pricing or product pages, a new CFO or controller hire, funding, expansion, or a relevant regulatory or vendor change)
- A new account is added to a Salesforce target list or moves into an active ABM tier
- A known contact from a past opportunity or closed-lost deal changes jobs or re-engages with content
- An inbound lead is created in Salesforce that matches the ICP but has not booked a meeting
- A weekly scheduled run to refresh priority accounts that have gone 30+ days without a touch
What it reads
- Salesforce (accounts, contacts, opportunities, closed-won and closed-lost history, activity, owner and territory assignments)
- Website visitor and intent signals from Warmly
- Contact and company data from Cognism, with FullEnrich as a waterfall fallback for missing emails and direct dials
- Public sources: company websites, press releases, job postings, annual reports and regulatory filings where relevant
- Approved messaging library and compliance rules maintained in a shared knowledge base
- Sequence engagement data from Salesloft (opens, replies, clicks, bounces, unsubscribes)
How it works, step by step
- Define and score the ICPAnalyze closed-won and closed-lost deals in Salesforce to confirm firmographics (industry, revenue band, employee count, region, tech stack), buying committee personas (CFO, controller, treasurer, head of finance ops, compliance lead), and the signals that preceded past wins. Encode these as a scoring model that Claude applies to every account, and review it with sales leadership before going live.
- Detect and prioritize signalsn8n workflows pull signals from Warmly, Salesforce changes, and public sources on a schedule and in real time. Claude scores each account against the ICP and signal strength, skips accounts owned by a rep with an open opportunity or recent activity, and queues the top accounts for research each day.
- Research and enrichFor each queued account, Claude builds a short brief: what the company does, likely pain points tied to your offer, recent events, and 2 to 4 relevant contacts by persona. Cognism supplies verified contacts first, FullEnrich fills gaps. Every fact in the brief carries its source, and contacts are checked against Salesforce for duplicates, existing owners, and do-not-contact flags before being created or updated.
- Draft personalized outreachClaude writes a 3 to 5 touch sequence (email plus optional LinkedIn and call tasks) using approved frameworks: a signal-based opener, a persona-specific problem statement, a proof point pulled only from the approved library, and a low-friction ask. Drafts are pushed to Salesloft as pending steps, never sent directly by the agent during the pilot.
- Review and sendReps or an SDR lead approve, edit, or reject drafts in Salesloft. Edits are captured and fed back weekly to improve prompts. Once quality is proven per persona, low-risk follow-up steps can move to auto-send while first touches stay human-approved.
- Handle replies and bookClaude classifies replies (interested, referral, objection, not now, unsubscribe, out of office). Interested replies get a Chili Piper booking link routed to the correct rep by territory and round robin. Unsubscribes and opt-outs are written to Salesforce immediately. Anything ambiguous or sensitive goes to a human.
- Hand off and logWhen a meeting is booked, the agent creates or updates the Salesforce opportunity or lead status, attaches the research brief and full outreach history, and notifies the rep in Slack or email. All generated messages and sends are logged in Salesforce for record retention.
When a person takes over
- A prospect replies with interest, a question about pricing, terms, rates, or regulatory matters, or anything beyond booking a meeting: the assigned rep takes over because the agent must not make product, pricing, or compliance claims
- The account is already owned by a rep, is an existing customer, or sits in a named strategic tier: the account owner decides whether and how to engage
- A draft would need a claim, case study, or statistic that is not in the approved messaging library: compliance or marketing reviews before anything is used
- A reply is negative, mentions a complaint, legal concern, or data request (such as a deletion request): the SDR lead or compliance handles it within the defined response window
Rules it follows
- Write only from the approved messaging and proof-point library, never promise returns, rates, savings, or outcomes, and include required disclosures and opt-out language in every email
- Respect consent and suppression rules: check Salesforce do-not-contact flags, regional rules (CAN-SPAM, GDPR, CASL), and global suppression lists before any contact is created or sequenced
- Never contact accounts with an open opportunity, active customers, or contacts touched by a rep in the last 30 days, and cap new contacts per account to avoid flooding a buying committee
- Every research claim must cite a source and be under a set age, unverifiable facts are dropped rather than guessed, and personal or non-business data is never used in personalization
- Enforce daily send limits, domain warmup, and bounce thresholds, pausing sequences automatically if bounce or complaint rates exceed set limits, and log every message for audit and retention
How you know it works
- Meetings booked per week from agent-sourced accounts
- Pipeline created and pipeline value attributed to agent-sourced meetings in Salesforce
- Positive reply rate by persona and signal type
- Draft approval rate without edits (a measure of message quality)
- Rep hours saved on research and list building per week
From pilot to live
- Pilot (weeks 1 to 4)Lock the ICP, personas, and approved messaging with sales and compliance. Connect Salesforce, Warmly, Cognism, FullEnrich, and Salesloft through n8n. Run the agent for two reps on one segment with every draft human-approved, and review output quality, data accuracy, and compliance flags weekly.
- Expand (weeks 5 to 10)Add remaining reps and personas, turn on reply classification and Chili Piper booking, and introduce job-change and closed-lost re-engagement plays. Move proven follow-up steps to auto-send while first touches stay approved. Report meetings and pipeline against the pre-pilot baseline.
- Run and optimize (ongoing)The agent runs across the full team with a named operator owning monitoring, deliverability, and a weekly tuning cycle (prompt updates from rep edits, scoring model refresh from new wins and losses, new signals). A monthly review with sales leadership and compliance covers results, exceptions, and messaging library updates.
Tools that fit
Warmly
Reveal site visitors and buying signals
Cognism
B2B contact database platform
Chili Piper
Demand conversion platform
n8n
Workflow automation platform
Some tool links are partner links. Hacking Demand may earn a commission at no extra cost to you.
This blueprint is a version of The Prospector (prospecting agent) ↗
Want The Prospector built for you?
Our experts build it inside your stack. Then we hand it off, maintain it, or operate it with you. Agents and programs get better with tuning, so most teams keep us on to run them. Bring this blueprint to a scoping call.
Scope this agent ↗ Get your own blueprint ↗Good questions.
Straight answers.
How do we keep outreach compliant in financial services?
The agent only writes from a messaging and proof-point library that compliance approves, cannot make claims about rates, returns, or outcomes, and drafts go to Salesloft for human approval before sending. Every message is logged in Salesforce, suppression and consent rules are checked before contact, and anything touching pricing, terms, or regulatory questions is handed to a person.
Will the agent step on accounts our reps already work?
No. Before researching or sequencing, it checks Salesforce for account ownership, open opportunities, customer status, and recent rep activity, and skips or routes those accounts to the owner. Strategic or named accounts are excluded unless the owner opts them in.
What does our team need to do once it is running?
Reps review and approve drafts (a few minutes per batch), take booked meetings, and flag bad research or messaging. Sales leadership and compliance join a monthly review to approve messaging changes. The operator running the agent handles monitoring, deliverability, data quality, and weekly tuning.
Every good build
starts with a conversation.
New builds, remodels, and additions, powered by AI. Let’s plan the work.