Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
Muse
Dots+22 moreClaude CodeCursorCodex CLIMuseDots
OpenClaw
Manus
Grok
Grok Bot+18 agentsAI Sales Architect — End-to-End Sales Agents
Design evidence-led AI sales agents across prospecting, qualification, outreach, pipeline, proposals, and handoffs.
$9.99
AI Sales Architect — End-to-End Sales Agents
Example session with this skill installed
Design an end-to-end AI sales agent system for a B2B SaaS company selling an annual analytics platform to mid-market operations teams. We use inbound demo requests and a small, research-led outbound motion. Our CRM has Lead, Account, Contact, and Opportunity records; sales stages are Qualify, Discovery, Evaluation, Proposal, and Decision. We have no approved discounts or current win-rate data to share. Produce the architecture, buyer/ICP criteria, qualification, responsible outreach, CRM rules, human approvals, and launch tests. Label assumptions and unknowns. Do not invent prospect data or claim actions were performed.
- Read your context and instructions
- Compiled the ai sales architect
AI Sales Agent Architecture — Mid-Market Analytics SaaS
1. Executive summary
Design a governed revenue copilot that prioritizes evidence-backed inbound and outbound opportunities, supports consistent qualification and follow-up, and keeps CRM records truthful. This is a proposed architecture based only on the brief; no prospect research, CRM access, messaging, or implementation has occurred.
2. Scope and open assumptions
Known: Annual B2B SaaS offer; mid-market operations teams; inbound demo and limited research-led outbound; CRM entities are Lead, Account, Contact, Opportunity; stages are Qualify, Discovery, Evaluation, Proposal, Decision.
Unknown: Product capabilities and proof, target regions, pricing and margins, supported integrations, data sources, consent rules, service capacity, stage exit criteria, and discount authority. Confirm these before automation. Do not infer win rates or buyer intent.
3. ICP and exclusions
Build an ICP from validated customer evidence. Candidate dimensions to confirm: operations complexity, team size, analytics maturity, data availability, urgency, supported systems, and ability to implement. Exclude or route for review accounts outside supported regions, with unsupported security or integration needs, no relevant problem, or economics that cannot work. Treat any score as a prioritization aid, not proof of intent.
4. Workflow and controls
| Stage | Agent support | Gate / evidence |
|---|---|---|
| Inbound | Acknowledge the stated request, classify questions, suggest qualification topics | Send only through an authorized channel; route complex or sensitive requests to a person |
| Research | Summarize first-party and approved sources; label fact, signal, inference, unknown | Store source and retrieval date; verify freshness before using a current signal |
| Qualify | Capture problem, impact, desired outcome, stakeholders, timing, constraints, and next step | Unknown stays unknown; salesperson confirms stage transition |
| Discovery | Prepare role-specific questions and an evidence-based value hypothesis | No invented pain, metrics, authority, or ROI |
| Evaluation | Track agreed criteria, stakeholders, dependencies, and open risks | Human owner validates customer commitments and technical claims |
| Proposal / Decision | Assemble approved scope and flag pricing or contract questions | Authorized person approves price, discounts, terms, and final proposal |
| Handoff | Pass goals, commitments, configuration needs, and unresolved risks to delivery | Customer-facing handoff reviewed by the account owner |
Outbound messages require a legitimate basis, account-specific evidence, relevant value, approved channel and cadence, and a working opt-out path. Stop sequences on opt-out, rejection, or ambiguous identity. The agent drafts; it does not send unless explicitly authorized and technically confirmed.
5. CRM and next-action rules
Use existing records; define duplicate matching before writes. Record source, timestamp, evidence link, confidence, owner, stage, last meaningful interaction, and one dated next action. Distinguish confirmed facts from indicated needs, hypotheses, and unknowns. Do not create activity history or pipeline amounts from generated text. Log proposed changes for owner review; only an authorized CRM connector may write after approval.
6. Launch tests
Test synthetic scenarios for duplicate accounts, stale research, missing consent, opt-out, wrong-person replies, unsupported claims, unknown budget, discount requests, stage regression, failed CRM writes, and retries. Verify that the system pauses for human approval at customer-facing, pricing, contract, and record-write boundaries. Measure response quality, evidence coverage, opt-out handling, stage accuracy, and time saved against a baseline before expanding automation.
Next decisions
Confirm ICP evidence, product proof, regions and communication rules, approved CRM fields, sales-stage exit criteria, owner/approval matrix, and what the agent may draft versus execute. Then run a small supervised pilot with synthetic or properly authorized data.
Connects securely to your tools. The creator never sees your data.
About this skill
Architect a complete, governed AI sales system
AI Sales Architect turns your offer, sales motion, team, CRM, and constraints into an end-to-end revenue workflow. It helps founders, sales teams, revenue operations, agencies, and SaaS businesses plan AI support from market selection through customer handoff.
What it covers
- Sales motion, offer intelligence, ideal and negative customer profiles, and buyer roles
- Account research, buying signals, evidence quality, prioritization, and qualification
- Inbound response, responsible outbound, relevant messaging, reply handling, and discovery
- Stakeholder maps, value cases, objections, opportunity stages, next actions, and deal risks
- Proposals, pricing and discount authority, negotiation support, procurement, and close plans
- CRM objects and truth rules, permissions, human approvals, escalation, and sales-to-success handoffs
- Analytics, attribution cautions, experiments, quality checks, failure handling, cost, and maturity
How to use it
Share the product or service, target market, sales motion, buyer roles, sales stages, CRM, available evidence, pricing rules, team responsibilities, communication permissions, and desired outcomes. The skill asks for information that materially affects the design, separates facts from signals and assumptions, and identifies unknowns instead of inventing them.
Built-in safeguards
The architecture is designed around the REVENUE PATH framework: Right Market, Evidence, Value, Engagement, Need, Unit Economics, Execution, Pipeline, Authority, Timing, and Handoff. It emphasizes truthful CRM state, permission-aware outreach, opt-out handling, authorized actions, human review for consequential decisions, and verification before claiming a tool action occurred.
Boundaries
This is planning and decision support. It does not independently source prospects, send messages, change a CRM, negotiate, approve discounts, or close deals unless compatible tools are connected and explicitly authorized. It never fabricates buyers, contact details, intent, pipeline values, proof, pricing, or sales results. Validate workflows against your actual policies, applicable rules, tools, and customer evidence before launch.
How to use
Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scanned
Verified clean 10 days ago
- Passed all security checks, Safe to install