Executive Report

opportunity intelligence engine

Document version: 1.1 · Prepared: 2026-05-03

Audience: founders, investors, and strategic partners

At a Glance

Peephole helps service providers see through to the opportunities others miss. A door peephole gives you a focused, framed view of who's on the other side; Peephole gives consultants, agencies, nonprofit leaders, and economic development professionals a focused, framed view of the opportunity moving through their local economy — and tells them exactly what to do about it.

Where ZoomInfo serves enterprise sales teams and Crunchbase tracks venture activity, Peephole serves the services economy — the consultants, agencies, nonprofit leaders, business owners, and economic development professionals who operate where most Birmingham-, Mobile-, and Huntsville-scale opportunity actually happens.

The product transforms a continuous stream of local market signals — restaurant openings, hiring surges, mergers, RFPs, infrastructure changes, redistricting events — into specific, actionable next steps grounded in a 217-service marketplace catalog. The user opens the app, sees the opportunities that matter to them today, and is told exactly what to do about each one — which service responds to the signal, what to lead with, who to contact, and which work product to produce.

We have a working iOS application, a tested AI matching pipeline running on Google's Gemini 2.5 Flash, end-to-end RLS-enforced data infrastructure on Supabase, and the first 217-service marketplace catalog populated. We are pre-revenue and pre-public-launch, raising $2.5M seed to convert this technical foundation into a regional then national market leader in opportunity intelligence for the services economy.

The Problem

The U.S. services economy is enormous — over $10T in annual GDP contribution — but the tools sold into it are designed for someone else's job.

• Sales intelligence platforms (ZoomInfo, Apollo, Cognism) are built for enterprise SaaS reps cold-emailing director-level buyers. Their data ontology assumes large, predictable buying centers. A marketing consultant in Birmingham doesn't have a "named accounts list."
• Market intelligence tools (Crunchbase, PitchBook, CB Insights) track venture-funded startups. The bulk of real local opportunity — a food hall extending hours, a manufacturer choosing a site, a county receiving an EDA grant — never appears.
• News aggregators (Feedly, Google Alerts) surface raw articles but stop there. The user still has to do the synthesis: *"OK, but what should I actually do about this?"*
• CRM tools (HubSpot, Salesforce, Pipedrive) start at the deal — by the time something is in your CRM, you've already done the hard work of finding it.

The gap between "a news article about something interesting happening in your market" and "a concrete actionable next step grounded in your skills" is exactly where most service providers spend their pre-sale time. They read, they think, they Google, they ask their network. It's slow, it's idiosyncratic, and most opportunity dies in this gap.

Peephole closes that gap. The product is named after the door fixture for a reason: a peephole gives you a focused, framed view of what's on the other side before you decide whether to open the door. That's exactly the user experience — a small, focused window onto local market signal, sharp enough to act on, narrow enough to triage in a coffee break.

The Product

What the user sees

A solo consultant opens Peephole on her phone in the morning over coffee. Her Daily Focus card is at the top of her Home tab — three opportunities the system thinks are worth her attention today, ranked by signal strength + fit with her stated services. She taps the first one — a piece on a long-time Birmingham restaurant returning after a hiatus.

The detail screen tells her, in three glance-sized sections:

• The opportunity — what's happening, why it matters as a business signal, framed in the article's own emotional valence.
• The recommended service — pulled from Peephole's 217-service catalog, with a confidence score, a "why this fits" paragraph grounded in the article's specifics, and an offer angle she could lead with on outreach.
• The next move — one tactical sentence she could act on today, plus a button that scaffolds a Workspace artifact (e.g. a Stakeholder Map, a Brand Audit Report, a Competitive Brief) with the opportunity already populated.

If she likes it, two taps add it to her Pipeline with a tracked status (Researching → Outreach → Proposal → Won/Lost). The Pipeline tab shows everything she's pursuing across the week with revenue projection. Her Workspace tab holds the actual deliverables in progress.

What's beneath the surface

Every component of the user-visible product is backed by deliberate technical architecture:

The dual-mode AI matching engine

The most differentiated piece of the product is how matching actually happens. We run two distinct AI prompts depending on what we know about the user:

Personalized Mode — fires when the user's profile is rich enough (profession + industry + services + industries). The prompt injects the full profile, and Gemini is required to ground every narrative in concrete profile elements ("As a senior brand strategist working in healthcare…").

Opportunity-First Mode — fires when the profile is sparse, or when the user explicitly toggles the in-app Match Mode chip. The prompt has zero profile coupling. Instead, it walks Gemini through an explicit 8-step reasoning chain:

1. What opportunity signal is present?
2. What economic or business need does this signal suggest?
3. What service from the marketplace best responds to that need?
4. Why does that service fit the situation?
5. What offer angle could someone use?
6. What type of person or business could act on this?
7. What first action should they take?
8. What output should they create in the Development Workspace?

The result is opportunity-grounded analysis — actionable for any reader, not anchored to one user's identity. Every recommendation surfaces five new structured fields per service: signalType, detectedNeed, actorTypes (multiple personas), suggestedFirstAction, and suggestedWorkspaceOutput. This is the explainable AI layer that turns Peephole from a "smart feed" into a reasoning system.

Live verification on staging shows 100% Gemini success on Opportunity-First mode and 91.7% on Personalized mode, with 100% of clean rows producing all five reasoning fields populated meaningfully — e.g. signalType: "redistricting", suggestedWorkspaceOutput: "Economic Impact Assessment".

What's already built and tested

The core product is built and verified. What separates the current state from a real launch is the operational layer — a dedicated production environment, billing, App Store listing, observability, and the marketing fly-wheel — not the product itself.

The Market

Who Peephole serves

Peephole's user is anyone whose income depends on identifying opportunity in a local services economy and converting it into engagement. Concretely:

• Solo consultants and boutique agencies — marketing, brand strategy, business consulting, organizational development, financial advisory, real estate, education, workforce development.
• Nonprofit leaders and economic development professionals — chamber of commerce staff, city planners, EDC directors, foundation program officers tracking grant-relevant signals.
• Small business owners — service-based businesses (legal, accounting, design, IT services) needing market intelligence without enterprise tooling cost.
• Service providers entering a new market — anyone expanding their footprint who needs a structured read on what's happening in a metro before deploying capital or time.

Market sizing (initial geography)

Birmingham metro alone has roughly 35,000 small-to-mid services-economy businesses per BLS QCEW data, with an average of 1.5 owners or principals per business — call it 50,000 individual potential users in metro Birmingham alone. At a target $49/month subscription, Birmingham is a $30M ARR total addressable market. Mobile, Huntsville, Montgomery, and the broader Alabama metro footprint roughly triple the in-state TAM.

National TAM at the same conversion model — limited to top-50 metros and the same ICP — is $1.5B-$2.5B ARR. International expansion (UK, Canada, Australia commonwealth + EU markets with similar SMB density) adds another estimated $3B-$5B ARR addressable.

Why now

Three forces converge:

1. AI-cost collapse for high-context reasoning. Gemini 2.5 Flash and equivalent models from Anthropic and OpenAI have brought the per-call cost of catalog-grounded reasoning to fractions of a cent. Two years ago, the same prompt would have cost 10-100x more and made Peephole's unit economics untenable.

1. Mobile-first market intelligence is unfilled. Every existing market-intelligence tool was designed for the desktop and the seller. Peephole is designed for the browser of the consultant's morning coffee — a feed-shaped product on the device people actually use to triage their day.

1. The services economy is increasingly disaggregated. Solo consultants, fractional executives, agency-of-one operators are the fastest-growing segment of the U.S. workforce. They have buying power, but they don't fit any incumbent's ICP. Peephole is built for them from day one.

Traction & Validation

We are pre-revenue and pre-public-launch. What we have is:

• A working product end-to-end on staging infrastructure, capable of supporting friends-and-family beta today.
• Verified AI matching at the unit-economics level — a single enrichment call costs ~$0.002-$0.005 and produces a five-section actionable analysis grounded in a 217-service catalog.
• Architectural rigor — RLS-enforced data isolation, diagnostic capture for failure modes, dual-mode prompt architecture surviving real-world prompt regressions, 270+ automated tests across the stack.
• Domain depth — the 217-service catalog is the artifact of months of work systematizing how services-economy work is actually packaged. It's not a list; it's an ontology with scaling tiers, use cases, ideal client profiles, and pre-mapped reasoning context for the matching engine to ground in.
• Initial geographic specialization — Birmingham source atlas, RSS feed coverage, sources tier classification (verified municipal sources vs. unverified blogs), pre-validated against local market signals.

This is a founder-led pre-seed product with the technical risk burned down and the go-to-market and business risk still ahead.

Roadmap & Scale

Year 1 (post-funding) — establish Birmingham, prove unit economics

• Q1: Production environment, App Store launch, first 100 paying users in Birmingham, billing infrastructure (Stripe), observability (Sentry, PostHog).
• Q2: Reasoning trace UI surfaces the 5-field Opportunity-First context to users. Marketplace gap surfacing turns hallucinated-service signals into catalog-expansion candidates. Onboarding refinement based on real funnel data.
• Q3: Mobile, Huntsville, Montgomery rollout. Source atlas expansion. Geographic personalization. Reach 1,000 paying users.
• Q4: Atlanta + Nashville rollout (regional adjacent metros). API for partners (e.g. local chambers of commerce embedding Peephole intelligence). Reach 3,500 paying users = ~$2M ARR.

Year 2 — multi-region, multi-tier

• Top-15 U.S. metro coverage, vertical-specific source atlases (real estate, healthcare, nonprofit, government).
• Tiered pricing (free with cap → $29 individual → $99 team → $399 enterprise/agency).
• B2B distribution channel through chambers of commerce, EDCs, SCORE chapters, regional accelerators.
• Reach $10M ARR by end of Year 2 with positive contribution margin.

Year 3 — national + adjacent products

• Top-50 metros + international expansion to UK + Canada.
• Adjacent products: Peephole for Investors (deal-flow signals), Peephole for Real Estate (location-bound opportunity intelligence), Peephole API (white-label opportunity intelligence for industry verticals).
• Reach $30M+ ARR, Series B from growth-stage capital.

Cost Structure & Unit Economics

Direct cost per active user per month

A typical Peephole user generates ~20-40 enrichment calls per month (opening cards, refreshing the daily focus, adding to pipeline, generating workspace artifacts). At ~$0.003 per Gemini call:

• Direct AI cost per active user: $0.06 - $0.12/month
• Hosting, ingestion, push, infrastructure: ~$0.20/month/user (Supabase + Firebase + storage)
• Customer support + ops amortized: ~$2/user/month at 1,000 users; declines materially with scale
• Total cost-to-serve per active user: ~$2.30/month at 1,000-user scale, dropping to <$0.80/month at 10,000-user scale

Pricing and gross margin

Initial pricing structure:

Blended ARPU at 1,000 users: ~$32/month (assumes 60% Free, 30% Individual, 8% Team, 2% Enterprise).

Gross margin at scale: ~92% (industry-standard SaaS, supported by the AI cost collapse described above).

LTV:CAC pathway

Assumed Year 1-2 economics:

• CAC: ~$60 blended (mostly content + community-led + chamber partnerships, with paid taking a smaller share)
• Average tenure: 18-24 months at maturity (early data will refine)
• LTV at $32 blended ARPU and 92% margin: ~$530-$700
• LTV:CAC: ~9-11x

The economics are healthy because the product is mobile-native, discovery-friendly (consultants share insights they find), and the underlying AI cost is a fraction of subscription price.

The Ask

$2.5M Seed Round. Standard SAFE or priced round. Targeting close in Q3 2026.

We have $X already committed from [founder + angel commitments — to fill in], with the remaining round open to lead and follow investors with services-economy or AI-product domain expertise.

Use of funds — 24-month runway

                       USE OF FUNDS — $2.5M / 24 months
   ┌────────────────────────────────────────────────────────────────────┐
   │                                                                    │
   │ Engineering & Product       ████████████████████████  $1.25M  50%  │
   │ Go-to-Market & Sales        █████████████             $625K   25%  │
   │ AI / Infrastructure         ██████                    $300K   12%  │
   │ Operations & G&A            ████                      $200K    8%  │
   │ Reserve / Contingency       ███                       $125K    5%  │
   │                                                                    │
   └────────────────────────────────────────────────────────────────────┘

Where the money goes — line by line

Engineering & Product — $1.25M (50%)

• 3 senior engineers (mobile, backend, AI/ML) at fully-loaded $200K each = $600K/year × 2 = $1.2M
• 1 senior product designer at fully-loaded $150K × 12 months bridge = $150K
• Tooling and equipment for the team — $25K
• Supports: production-grade observability, Reasoning Trace UI, marketplace gap surfacing, multi-metro source atlas expansion, billing + tier enforcement, API layer for Year 3 platform play.

This is the largest line because product velocity is the moat. The dual-mode AI matching architecture, the catalog ontology, and the mobile-first feed are all things competitors will copy within 12 months once we're publicly visible. Sustained product velocity by a small senior team is what keeps us ahead of that.

Go-to-Market & Sales — $625K (25%)

• Founding head of GTM (player-coach role) — $200K fully loaded × 18 months = $300K
• 1 BDR / partnerships rep focused on chamber + EDC channel — $150K × 12 months = $150K
• Performance marketing budget (paid acquisition, content amplification) — $100K
• Conference presence + community sponsorship (Birmingham, Mobile, Atlanta SMB & consulting events) — $50K
• Content production (case studies, founder-led editorial in trade press, podcast circuit) — $25K

Distribution is the wedge. Birmingham's services economy is community-knit; the cheapest distribution is via the community organizations that already serve our ICP — chambers, EDCs, SCORE, business journals, industry associations. The GTM hire is responsible for converting that recognition into an organized partnership channel.

AI / Infrastructure — $300K (12%)

• Gemini API spend — $150K (supports ~5,000 active monthly users at projected enrichment volume; represents real cost-of-goods, not engineering overhead)
• Supabase enterprise tier + dedicated infrastructure — $50K
• Firebase, push, storage, CDN — $25K
• Security audit + compliance (SOC 2 Type I prep) — $50K
• AI evaluation infrastructure (golden-set datasets, automated quality monitoring) — $25K

The AI evaluation line is critical and frequently under-funded at seed. We need real, ongoing measurement of recommendation quality across modes, geographies, and verticals — not just gemini_pct but did the user act on it. This becomes the inner loop of product improvement.

Operations & G&A — $200K (8%)

• Founder salary (below-market, conservative) — $120K × 12 months bridge to A
• Legal, accounting, banking, taxes — $40K
• Insurance + corporate compliance — $20K
• Office / coworking / travel — $20K

Reserve / Contingency — $125K (5%)

• Buffer against AI pricing changes, unexpected infrastructure costs, or extension of runway by 1-2 months in pursuit of better Series A signal.

Milestones this $2.5M produces

By the time this round is fully deployed (Month 24):

• 3,500+ paying users across 4-6 metros
• ~$2M ARR with positive contribution margin
• Second-product prototype (Peephole for Investors or Peephole API) demonstrating platform optionality
• Established channel partnerships with 8-12 regional chambers / EDCs / industry associations
• SOC 2 Type I certified, ready for enterprise tier sale
• Series A ready — with 12+ months of growth data and a credible $20M+ ARR-by-Year-4 narrative

Why this size round

A $2.5M seed gives 24 months of runway with milestone-credible Series A signal at the end. Smaller ($1M-$1.5M) compresses the GTM line and forces premature commercialization before the product-Birmingham fit is proven. Larger ($4M+) at this stage either dilutes too aggressively or implies promised scale that the staging-tier infrastructure can't deliver against — better to raise the upsize at Series A from a position of demonstrated traction.

Risk & Honest Caveats

We believe in disclosing what could go wrong:

• AI provider dependency. We currently route through Google Gemini. The 35-second timeout fix shipped this quarter exposed how much our quality-of-output depends on the underlying model. Mitigation: we maintain clean abstraction at the prompt-builder layer, allowing migration to Anthropic Claude or OpenAI GPT-4 with engineering effort but without rebuild. Our test suite catches prompt regressions.
• Geographic specialization vs. expansion tension. Birmingham depth is what makes the product actually useful today. Premature national expansion would dilute that. The roadmap deliberately compounds metro-by-metro before going national.
• Service catalog maintenance cost. 217 services is the right starting point. Maintaining the ontology as the marketplace evolves is real work, partially mitigated by the "marketplace gap surfacing" feature which crowdsources catalog-expansion signal from Gemini's hallucinations.
• Competition. ZoomInfo, Apollo, Crunchbase will not enter our market — their unit economics don't work at our pricing. The realistic competitive risk is a vertically-focused incumbent (e.g. a real estate intelligence tool) horizontally expanding. Our defense is integrated mobile-native UX + the services-marketplace ontology, both of which take real time to replicate.

Closing

Peephole is a working AI-native opportunity intelligence platform built for the people who serve their local economies. We have built the technical foundation — verified, tested, and architecturally rigorous — and now need capital to convert that foundation into a regional then national market category.

The unit economics are strong. The market is real. The product works. What's missing is the operational scale-up that converts a working beta into a category-defining business.

We are raising $2.5M seed to do exactly that.

Peephole Business

opportunity intelligence engine

A market intelligence platform built where

the services economy actually lives.

Brand Identity Notes

Prepared with assistance from the Peephole engineering team. Document under continuous revision; latest version reflects the state of the product as of 2026-05-03. For most current technical state, see the codebase's `HANDOFF.md` and PR #6 (`feat/ai-service-matching-v2`).