← ALL BLUEPRINTS / AGENT BLUEPRINT
Speed-to-lead prospecting agent for inbound demo requests at a small ecommerce software company on Salesforce
The Prospector answers every inbound demo request within minutes, separates real buyers from shoppers and existing customers, enriches and scores each brand against your ICP, and books qualified prospects straight onto a rep's calendar in Salesforce. It stops the overnight decay that is costing you half your inbound and turns demand you already paid for into held meetings and pipeline.
| ROLE | SALES | PLATFORM | CUSTOM |
|---|---|---|---|
| INDUSTRY | ECOMMERCE / DTC | ||
| DRAWN BY | HACKING DEMAND | ISSUED | OCTOBER 20, 2026 |
What The Prospector does
A 11 to 50 person company selling software to ecommerce and DTC brands gets demo requests through its website form, but a rep usually replies the next morning, and by then roughly half of those brands have stopped responding or booked with a competitor. The form is also noisy: shoppers looking for order status, existing customers with shipping or returns questions, agencies, and vendors all land in the same queue, which slows reps down even more. The Prospector works for the sales team and owns the first hour of every inbound request. It classifies the submission, enriches the brand and the person, scores fit against your ICP and personas, sends a personalized reply with a booking link, routes non-sales requests to the right team, writes everything back to Salesforce, and runs short follow-ups on prospects who do not book. Reps own every live conversation, every demo, and every pricing discussion.
Custom build
This team runs Salesforce, so HubSpot Breeze is not the right fit, and Salesforce's own agent tooling is heavy for an 11 to 50 person company that needs one focused workflow working fast. A custom Claude build orchestrated in n8n handles the hard part well: reading messy free-text form submissions, telling a buyer apart from a shopper asking where their package is, and writing a reply grounded in the brand's actual storefront. n8n connects Salesforce through its API (or an MCP connector), calls FullEnrich and Apollo, and triggers Chili Piper routing, all in one place that is easy to monitor and adjust. Because the team wants it built and kept running, this setup also makes ongoing tuning straightforward: prompts, scoring rules and templates live in version-controlled workflows rather than scattered across tools.
What starts it
- A new demo request form submission creates a Lead in Salesforce (24/7, including nights and weekends)
- A high-fit brand identified by website visitor tracking views pricing or integration pages without submitting the form
- A prospect who received a booking link has not scheduled after a set number of hours
- A booked meeting is cancelled or the prospect does not show
- A seasonal window opens (for example, the run-up to peak holiday trading) and the agent switches to its peak-season messaging and routing rules
What it reads
- Salesforce Leads, Contacts, Accounts and Opportunities, including past closed-lost reasons and existing customer records
- Demo request form fields and free-text message
- Contact and company enrichment via FullEnrich waterfall lookups, with Apollo as a fallback for firmographics
- Storefront signals gathered from the brand's public site: ecommerce platform, approximate catalog size, apps and tools in use, shipping and returns policy pages
- Warmly website visitor and intent data (pages viewed, repeat visits, company identification)
- Help desk system for checking whether the sender is an existing customer with an open ticket
How it works, step by step
- Classify the submission in under a minuteClaude reads the form fields and message and sorts it into one of five buckets: prospective buyer, shopper asking about an order, existing customer, agency or partner, or vendor/spam. Shoppers and customers never enter the sales flow. Anything below a confidence threshold goes to a human review queue instead of being guessed.
- Enrich the brand and the requesterFor buyers, n8n runs FullEnrich to confirm email, title and LinkedIn, pulls firmographics from Apollo, and scrapes the storefront for platform, catalog size, and visible tech stack. It checks Salesforce for an existing Account, open Opportunity, or past closed-lost deal so the reply does not treat a returning brand like a stranger.
- Score fit and pick the pathThe agent scores the lead against your written ICP (platform, estimated order volume band, category, team size) and persona (founder, ecommerce manager, ops or CX lead). High fit goes to instant booking, medium fit gets a qualifying reply with one or two questions, low fit gets a polite self-serve path such as a recorded demo or help docs.
- Send a personalized reply within minutesUsing a framework you approve (one line on what they asked, one line tied to something specific about their store, one clear next step), the agent sends from the assigned rep's mailbox with a Chili Piper link routed to that rep's calendar. High-fit leads can also see the scheduler directly on the form thank-you page, so many book before any email is needed.
- Route and write back to SalesforceThe Lead is assigned by your round-robin or territory rules, the classification, fit score, enrichment, and a five-line research brief are written to Lead fields and an activity note, and the rep gets a Slack or email alert. Order-status and returns requests are forwarded to the help desk with the original message, and existing-customer requests go to the account owner.
- Follow up until booked or closed outIf no meeting is booked, the agent sends two or three short follow-ups over several days, each adding one useful point (a relevant integration, a peak-season consideration). Replies with real questions stop the sequence and hand to the rep. No-shows get a same-day reschedule message.
- Review and tune weeklyEach week the agent produces a summary in Salesforce reporting: response times, booking rates by fit tier, misclassified submissions corrected by humans, and reply themes. The operating owner adjusts the ICP rules, prompts and templates based on what reps flag.
When a person takes over
- A prospect replies with a pricing, contract, or custom integration question: the agent stops, tags the rep, and the rep answers personally, because commercial terms and technical commitments are never the agent's call.
- A multi-brand group, a large retailer, or an agency representing several stores submits a request: it routes to the head of sales or a founder for a personal reply, since these deals are worth more and need judgment on how to engage.
- The classifier is unsure whether the sender is a buyer, a shopper, or an existing customer: the submission sits in a human review queue with the agent's best guess, and a person decides within business hours.
- An existing customer uses the demo form, often to ask about an add-on or a problem: the account owner takes it so the relationship stays with the person who knows the account.
Rules it follows
- The agent never answers order status, shipping, refund, or returns questions, never asks shoppers for order numbers or payment details, and forwards these straight to the help desk with a short acknowledgment.
- No promises about go-live dates, implementation timelines, or readiness before peak trading periods. If a prospect asks whether they can be live before the holiday season, the agent books the call and flags the timing question for the rep.
- Every outgoing message uses only facts verified from the form, enrichment, or the prospect's public storefront. If a detail cannot be confirmed, it is left out rather than inferred.
- Messages come from the assigned rep's mailbox, follow your approved templates and tone, cap at three follow-ups, and honor unsubscribes and any data privacy requests immediately.
- Salesforce writes are limited to defined Lead fields, tasks and notes. The agent cannot convert leads, create Opportunities, change ownership outside routing rules, or delete records.
How you know it works
- Median time from demo request to first personalized reply, tracked separately for business hours, nights and weekends
- Demo request to meeting booked rate, split by fit tier
- Meeting held rate (booked meetings that actually happen)
- Pipeline created from inbound demo requests per month, compared with the same season last year
- Share of non-sales submissions correctly routed away from reps, measured by human corrections in the review queue
From pilot to live
- Phase 1: Shadow mode on live inboundConnect Salesforce, the form, enrichment and Chili Piper. The agent classifies, enriches, scores and drafts replies for every request, but reps send them. Compare the agent's classifications and drafts against what reps would have done, and lock the ICP rules and templates.
- Phase 2: Auto-reply for clear casesTurn on automatic replies and instant booking for high-confidence buyers and automatic routing for shoppers and existing customers, starting with nights and weekends where the delay hurts most. Medium-confidence leads still get rep-approved drafts. Review booking and misroute data weekly.
- Phase 3: Full coverage and ongoing operationExtend auto-reply to all hours, add follow-up sequences, no-show recovery, and visitor-intent triggers from Warmly, and load seasonal playbooks for peak trading. A named owner monitors errors, reviews the weekly report, retunes prompts and scoring each month, and re-tests before every seasonal spike.
Tools that fit
Chili Piper
Demand conversion platform
n8n
Workflow automation platform
FullEnrich
Waterfall contact enrichment
Warmly
Reveal site visitors and buying signals
Apollo
Sales intelligence & engagement 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.
Half our demo form submissions are shoppers asking where their order is. Will the agent email them a demo pitch?
No. Classification happens before anything is sent. Shoppers asking about orders, shipping or returns get a short acknowledgment and are forwarded to your help desk with their original message. They are never added to sales sequences, and anything the agent is unsure about goes to a person instead.
What happens during peak season when demo requests spike and reps are buried?
The agent's response time does not change with volume, so every request still gets a reply in minutes. Before the busy period you load seasonal rules: messaging that acknowledges code freezes and peak trading, stricter fit thresholds so reps only see the best leads, and an overflow round-robin. Questions about going live before peak are always flagged for a rep rather than answered.
Will prospects feel like they are talking to a bot?
Replies come from the assigned rep's mailbox, reference something specific about the prospect's store, and stay short. The agent does not try to carry a sales conversation: the moment a prospect replies with a real question, the rep takes over. Most high-fit prospects simply book through the scheduler and their first live conversation is with your rep.
Every good build
starts with a conversation.
New builds, remodels, and additions, powered by AI. Let’s plan the work.