Prompt 01
01 · Opportunity & micro-niche discovery
ROLE
Act as an evidence-led micro-business opportunity analyst for the Stag Vault track “AI Voice & Chat Agent QA Studio”.
EDITABLE VARIABLES
[YOUR NICHE] = the narrow sector, product category or use case to target
[TARGET CUSTOMER] = the exact buyer role and organisation/customer type
[LOCATION] = UK region, country or global market
[BUSINESS TYPE] = productised service, digital product, subscription, licence or hybrid
[PRODUCT] = the named deliverable or offer
[PRICE] = price hypothesis to validate, not an assumed market fact
[PLATFORM] = selling, delivery, CRM, storefront or automation platform
[EXPERIENCE LEVEL] = beginner, intermediate or advanced
[AVAILABLE BUDGET] = cash available before validation
[AVAILABLE HOURS] = realistic hours per week
[BRAND NAME] = working business/offer name
[TONE] = plain-English, expert, reassuring, direct, warm or other lawful tone
[GOAL] = the measurable customer or business result to pursue
TRACK-SPECIFIC BUSINESS BRIEF
Opportunity: A recurring QA service for agencies and businesses deploying voice or chat agents. It designs realistic and adversarial scenarios, checks answers against actions, measures latency/escalation/tool-call outcomes, documents defects and reruns regression suites after every change.
Who it is for: Operators with analytical QA, customer-service or workflow skills who can test systematically without needing to build the underlying agent.
What is sold: Scenario library, human test runs, transcript/tool-call audit, safety/escalation defects, latency/results dashboard, release recommendation and monthly regression.
Buyer: Voice/chat-agent agencies, SMBs with deployed agents, contact centres, SaaS vendors and managed-service providers.
Where it is sold: Agency partnerships, LinkedIn, AI implementation communities, software integrators and direct outreach to businesses already advertising an AI assistant.
Startup cost: £25–£100 | Time to launch: 2–7 days
Revenue model: Pre-launch QA project plus monthly regression, release and incident retesting. | Activity model: ACTIVE with strong recurring revenue
Why now: AI agents are moving from demos to live operations, making resolution, tool execution, escalation, latency and regression failures more expensive and visible.
Differentiation: Independent, scenario-based outcome QA that verifies what the agent did—not another agent build, prompt rewrite or chatbot reseller offer.
First-customer route: Create a 25-scenario public demo against a consenting test bot, publish an anonymised defect report and approach 20 agent agencies with white-label regression QA.
OBJECTIVE
Select the strongest narrow buyer/use-case combination for this business without drifting into a generic agency or product.
INFORMATION TO ANALYSE
Use only evidence I paste, clearly named public sources, client-approved material and the following track-specific research plan:
Use client-approved policies, knowledge sources, intents, tool schemas, call/chat transcripts, expected actions and vendor observability. Include background noise/accent/device tests only with consent and do not treat broad benchmarks as the client's SLA.
If evidence is missing, produce a collection plan and [VERIFY] fields instead of guessing.
TRACK-SPECIFIC EXECUTION DIRECTION
Select agents that perform real actions—booking, qualifying, updating CRM, answering account questions—where polite text can hide a failed outcome. Rank by transaction risk, test access, call volume and release frequency.
STEPS TO FOLLOW
1. Generate 12 combinations of buyer × trigger/problem × deliverable. 2. Score each for urgency, access to buyer, evidence availability, frequency, budget, delivery risk and repeatability. 3. Identify UK and global variants. 4. Reject regulated or high-liability versions the operator cannot safely serve. 5. Select one lead micro-niche and two controlled alternatives. 6. Define what would disprove the opportunity within 48 hours.
REQUIRED OUTPUT
A ranked 12-row niche table; one lead niche; ideal-customer snapshot; buying trigger list; market-evidence gaps; red-flag/rejection list; and a one-sentence commercial thesis.
Present the work in copyable tables, scripts, templates, checklists and action-plan blocks. Fill known variables and leave unknown variables visibly labelled.
TRACK-SPECIFIC COMPLIANCE / QUALITY BOUNDARY
Test only systems and data explicitly authorised. Never attempt real account takeover, expose personal data or conduct uncontrolled adversarial tests in production. Use synthetic identities and route security findings privately.
KPI SET
Track: paid QA packs, scenarios executed, outcome pass rate, critical defects, retest closure, latency/escalation compliance, monthly regressions, release changes caught and data incidents
STAG VAULT CROSS-SELLS
Reference these existing resources where useful rather than recreating them: Track 18 Voice Agent Reseller, Track 76 Chatbot-as-a-Service, Track 118 AI Consulting and Track 55 FAQ Response Vaults.
NON-NEGOTIABLE RULES
- Never invent live demand, traffic, sales, conversion rates, prices, laws, platform rules, testimonials or customer evidence. Mark unknowns [VERIFY] and give the exact source or experiment needed.
- Treat First £100 / £500 / £1,000 figures as operating milestones, not forecasts or guarantees. Separate revenue, costs, tax, refunds and owner time.
- Use public, permissioned or client-supplied information only. Do not scrape behind logins, expose personal data, impersonate professionals or bypass platform terms.
- Keep a human approval gate for legal, privacy, safeguarding, health, finance, security, regulatory and customer-facing decisions. This system organises and drafts; it does not certify compliance or replace a qualified professional.
- Make every deliverable specific to the stated buyer, niche and evidence. Reject generic filler, copied competitors, fake proof, spam outreach and vanity metrics.
- Prioritise a cheap validation test before a full build. Stop or revise when the pre-agreed evidence threshold is not reached.
Finish with one 30-minute next action, the evidence needed to unlock the next stage, and a short “What to avoid” list specific to this business.