Everything you need to know.

Luna Signal Intelligence is an autonomous competitive intelligence platform that predicts competitor moves 6 to 10 weeks before they are publicly announced. It monitors over 50 public sources continuously — including job postings, patents, SEC filings, GitHub releases, G2 and Trustpilot reviews, podcasts, arXiv, Reddit, and Bluesky — without requiring any paid API or manual effort. Luna uses a multi-LLM engine (Claude, Gemini, Groq, Mistral) to cross-reference weak signals across sources and generate structured alerts with probability scores, estimated timelines, and recommended actions. It also auto-drafts battlecards, performs win/loss analysis, and produces Mythos deep-research briefs for high-stakes decisions.

Luna solves the competitive blind spot problem: most companies learn about competitor moves only when they are publicly announced, leaving no time to respond strategically. Luna provides 6 to 10 weeks of advance warning by detecting weak signals — hiring patterns, patent filings, code releases, review sentiment shifts — before they become public announcements. This early warning allows teams to adapt roadmaps, brief sales, protect key accounts, and make proactive strategic decisions instead of reactive ones.

Luna is a predictive competitive intelligence platform, whereas Crayon and Klue are retrospective — they collect and organize what competitors have already announced. Luna predicts what competitors are about to do, 6 to 10 weeks before any public announcement. The key architectural differences: Luna monitors 50+ sources autonomously every 6 hours (vs. daily refresh), uses a multi-LLM engine with automatic failover (vs. single proprietary model), cross-references external signals with your internal documents for context-specific alerts (unique to Luna), starts at $49/month (vs. $25-40K/year), and supports bring-your-own LLM API keys for full cost transparency.

Luna delivers competitive analysis at a fraction of the cost and time of traditional consulting. A McKinsey-style competitive analysis typically costs $15,000 to $50,000 and takes 6 weeks to deliver. Luna produces a comparable strategic brief in 4 hours for under $100 using its multi-LLM engine. Unlike consulting engagements, Luna operates continuously — monitoring competitors 24/7, updating analysis automatically every 6 hours, and never missing a signal due to analyst availability. The Mythos deep-research mode uses Claude Opus extended thinking to deliver board-ready strategic briefs on demand.

Luna is designed for mid-market and enterprise companies with 50 to 10,000+ employees operating in competitive markets. The primary users are VP Strategy, Head of Competitive Intelligence, VP Product, CEO, CFO, and Chief Revenue Officers — professionals who need advance warning of competitor moves to make strategic decisions. Luna is particularly valuable for organizations that have experienced at least one costly competitive surprise in the past 12 months. It is not designed for solo entrepreneurs or pre-revenue startups.

Luna covers six industry sectors with calibrated signal detection: SaaS and B2B Technology, Fintech and Financial Services, Pharma and Biotechnology, Industrial and Manufacturing, E-commerce and Retail, and Energy and Infrastructure. Each sector has its own signal sources, detection thresholds, and prediction patterns optimized for industry-specific dynamics. For example, patent filings are weighted heavily in Pharma and Industrial but less so in SaaS, where hiring signals and GitHub activity are more predictive. Luna automatically adjusts its signal weighting based on your sector.

Luna generates competitive predictions using a four-step autonomous process. First, it collects signals from over 50 public sources every six hours. Second, it deduplicates items by URL hash across the entire network and scores each signal for relevance using statistical anomaly detection against each company’s 90-day baseline. Third, it clusters correlated signals across multiple sources — for example, connecting a job posting, a patent filing, and a GitHub repository to the same strategic initiative. Fourth, it cross-references these clusters through a multi-LLM router (Claude, Gemini, Groq, Mistral, OpenRouter) to produce a structured prediction with a probability score, estimated timeline, and recommended action. Every prediction cites the specific signals that triggered it — nothing is invented or hallucinated.

Luna cross-references your internal documents with external competitive signals to produce context-specific alerts. When you upload a business plan, roadmap, or financial projection, Luna extracts your target markets, strategic objectives, planned product launches, and expansion timelines using semantic vector embeddings (pgvector). It then runs similarity searches against incoming signals on every scan cycle. If a competitor starts moving into a market you are targeting, Luna alerts you specifically — not generically. This internal-external cross-referencing is unique to Luna and is the key reason its alerts are relevant to your situation rather than generic industry noise.

Luna monitors over 50 public data sources grouped into eight families. News and media includes Google News, TechCrunch, sector publications, YouTube channels, and Apple Podcasts. Hiring signals covers LinkedIn, Indeed, Welcome To The Jungle, Greenhouse, and Lever RSS feeds. Patents and IP covers USPTO, EPO, Google Patents, and INPI. Regulatory and financial includes SEC EDGAR full-text search, EUR-Lex, FTC, FDA, ANSM, CNIL, and Companies House. Customer reviews tracks G2, Capterra, Trustpilot, and Product Hunt. Code and products monitors GitHub releases, GitLab, npm, PyPI, and Dev.to. Science covers arXiv, PubMed, and HAL. Social and community monitors Reddit, Hacker News, Bluesky, Mastodon, and Wayback Machine. All sources are 100% public — no private data is accessed. Users choose which sources to enable per workspace.

Luna collects competitive intelligence data every six hours, autonomously and server-side, completely independent of whether any user is logged in. The scan engine runs on a pg_cron schedule (0 */6 * * * UTC) and processes all active competitors across all users in a single sweep. Critical alerts are dispatched immediately upon generation via webhook, Slack, and email — there is no delay between detection and notification.

Luna uses a multi-LLM router that automatically selects the best available AI provider for each task. The routing works as follows: Anthropic Claude Sonnet 4.5 handles cross-referencing analysis, Claude Haiku 4.5 handles deduplication, and Claude Opus 4.1 powers Mythos deep-research briefs. Google Gemini, Groq (Llama 3.3 70B), Mistral, and OpenRouter serve as automatic failover providers. If one provider fails or hits a rate limit, the router falls through to the next — every attempt is logged. Users can also bring their own API key from any supported provider in Settings, and Luna will use it for their workspace scans, chat, and battlecard generation at no additional platform cost.

Yes, Luna supports bring-your-own API keys for five LLM providers: Anthropic (Claude), Google (Gemini), Groq, Mistral, and OpenRouter. Users paste their key in Settings, then click Test and Save — Luna validates it with a minimal API call. Once active, the user’s key takes priority over Luna’s system keys for all workspace operations: scans, chat, battlecard generation, and Mythos reports. Keys are stored with column-level security restrictions so they are never exposed to the browser or any client-side query. This is one of the features that distinguishes Luna from competitors like Crayon and Klue, which do not offer LLM cost transparency or user-controlled AI spending.

Luna exclusively analyzes publicly available information and is 100% legal in all jurisdictions where it operates. Sources include public job boards, patent databases (USPTO, EPO), public GitHub repositories, published SEC filings, public review platforms (G2, Trustpilot, Capterra), and public advertising transparency portals. Luna never accesses private systems, internal databases, or proprietary information. The platform is GDPR compliant by design, with data stored regionally (EU data stays on EU servers) and encrypted at rest with AES-256.

Yes. Private companies often generate richer signal data than public ones — job postings, GitHub activity, and patent filings are all public regardless of whether a company is listed on a stock exchange. Private competitors are often easier to read because their teams are less media-trained.

No single source represents more than a small fraction of Luna’s intelligence. With 50+ sources enabled, if any individual one becomes unavailable the remaining sources continue uninterrupted and the cross-reference step automatically reweights. Source diversification is a core architectural principle.

Luna collects signals every six hours via its autonomous scan engine. Reports and alerts reflect data from the past 24 to 90 days depending on the analysis type. Every signal is timestamped with its original publication date and the scan cycle that captured it, so users always know exactly how recent the information is. The deduplication system ensures no signal is counted twice even if it appears on multiple sources.

Not yet. Video and audio intelligence (YouTube conference talks, earnings call recordings) is planned for Phase 2 when Claude native video understanding becomes available.

The most valuable documents are:

• Business plans

• Financial projections

• Product roadmaps

• Geographic expansion plans

• Market studies you have commissioned

• Win/loss analysis

• Strategic planning documents

The more context Luna has about your objectives, the more relevant every alert becomes.

Luna does not store raw files after processing. When you upload a document, Luna extracts the content, converts it into semantic vectors (pgvector), and stores structured data in an isolated database. The original file is deleted after extraction. Your data never leaves your isolated environment.

Never. Each client has a completely isolated database with Row-Level Security. No data crosses between accounts. Your documents are used exclusively to generate analysis for your company.

Not without your explicit consent. Luna may use aggregated, fully anonymized patterns (never identifiable data) to improve prediction accuracy across the platform. This requires your opt-in. You can decline at any time with no impact on service. As an incentive, clients who opt in receive bonus LunaUnits each month.

Yes. Instantly. Go to Settings → Documents → Delete. Deletion is immediate and permanent. The extracted vectors are also removed from the database.

Our minimum target is 65% accuracy during beta, rising to 70%+ as the model accumulates validated data. Our live accuracy rate is always visible in your dashboard. Every prediction is dated, archived, and tracked to its outcome — confirmed, not confirmed, or pending. We display this publicly. We have nothing to hide.

The prediction is marked as not confirmed in your dashboard and in our public accuracy tracker. Luna learns from every incorrect prediction. Each outcome — right or wrong — improves the pattern matching model over time. Your 30-day money-back guarantee covers situations where you feel Luna did not deliver useful predictions.

No. Luna generates probabilities — not certainties. No competitive intelligence tool can guarantee outcomes. Any vendor that claims otherwise is misleading you. What Luna guarantees: full transparency on accuracy, sourced predictions, and honest confidence scores.

Every prediction and every chat response shows a confidence indicator based on the quality and quantity of underlying data.

• High confidence 75%+ — multiple sources, recent data, strong historical pattern match

• Medium confidence 50–74% — partial data, some uncertainty

• Low confidence below 50% — limited data, Luna recommends verifying with additional sources

When data is insufficient, Luna says so directly. It never invents an answer.

Yes. Luna’s M&A Intelligence module detects patterns that precede acquisitions — VP Corporate Development hires, M&A law firm mentions, CFO departures, investor relationship activity, and unusual capital movement signals. M&A predictions consume premium LunaUnits due to the analytical depth required (Claude Opus).

Three levels:

• CRITICAL (threat score above 75): immediate Slack and email. Actionable in 30 seconds. Includes recommended action.

• WARNING (score 50–75): included in daily digest.

• INFO (score 30–50): included in weekly report.

You configure the thresholds. You decide what matters.

• Executive summary of the week

• Top 8 signals ranked by strategic impact

• Active predictions with updated probabilities

• Threat matrix showing all competitors ranked

• One confirmed or updated prediction

• Intelligence quality KPIs (accuracy rate, signals analyzed)

Delivered every Monday at 08:00 in your timezone.

Yes. Through the Intelligence Chat or the Reports tab. Describe what you need. Luna estimates the LunaUnit cost before generating. You confirm before it runs. Analysis types available: competitor deep dive, market entry analysis, M&A intelligence, sector trend report, battlecard, board report.

A conversational interface where you ask Luna any strategic question in natural language.

“Should I enter the US market in Q3?”

“Is my competitor preparing a product launch?”

“What is happening in my sector this month?”

Luna answers using your internal documents plus real-time external signals. Every response cites sources. Every response shows a confidence level. Luna never generates an answer without data backing it.

• Weekly and monthly reports: email HTML + PDF download

• Quarterly board reports: premium PDF, white-labeled with your company logo, ready for board presentation

• All data is also available via API and exportable to Excel and Power BI

A LunaUnit is Luna’s unit of consumption for competitive intelligence operations. Every action — generating an alert, producing a report, running a scan, making an API call, or drafting a battlecard — consumes a defined number of LunaUnits. Standard units cost $0.20 each and cover alerts, digests, weekly reports, API calls, battlecards, and exports. Premium units cost $1.00 each and cover board-level reports, M&A intelligence, deep-dive analyses, and Mythos research briefs that require Claude Opus. The model is similar to AWS credits or OpenAI tokens — you pay for exactly what you consume, and unused units roll over indefinitely.

Because a company tracking 3 competitors in one country should not pay the same as one tracking 15 competitors across 4 continents. Usage-based pricing is fair. It also means Luna’s revenue grows proportionally with the value delivered. When you get more value, we earn more. Aligned incentives.

Five packs:

• Analyst — $149/mo — 500 standard + 50 premium units

• Manager — $499/mo — 2,000 standard + 200 premium units

• Director — $1,499/mo — 8,000 standard + 500 premium units

• C-Suite — $3,999/mo — 25,000 standard + 2,000 premium units

• Enterprise — $12,999/mo — 100,000 standard + 10,000 premium units

Unused units roll over to the next month. Never lost.

• Standard units ($0.20 each): alerts, digests, weekly reports, API calls, battlecards, exports

• Premium units ($1.00 each): board-level reports, M&A intelligence, deep-dive analyses, market studies

Premium actions use Claude Opus — the most powerful model — for the depth and quality they require.

The broader the geographic scope, the more sources are activated, the more data is processed, and the more LunaUnits are consumed.

• Local scope: ×1.0 base cost

• Regional scope: ×1.8

• National scope: ×3.2

• International scope: ×6.5

You choose the scope before each analysis. Luna shows the estimated cost before you confirm.

Never. Unused units roll over to the following month indefinitely. This is a fundamental trust rule. We will never delete units you have paid for.

Yes. Configure a threshold (example: 500 units remaining) and a pack to automatically purchase when reached. Continuous service. No interruption. No manual monitoring.

Yes. 30-day money-back guarantee on all plans. If Luna does not deliver a prediction you find genuinely useful in the first 30 days, we refund you in full. No questions asked.

• Starter: 3 users

• Growth: 10 users

• Business: 25 users

• Enterprise: unlimited

Additional users beyond plan limits: $100–$150/month each.

• Admin (one per company): full access, billing, user management

• CEO/Strategy: predictions, threat matrix, board reports, chat

• VP Product: competitor signals, patents, roadmap intelligence

• VP Sales: battlecards, pricing alerts, win/loss signals

• CFO: M&A signals, financial reports, quarterly board reports

• Analyst: raw signals, exports, API access

• Read-only: receives reports and digests, no dashboard access

Yes. The admin configures which alerts go to which roles. A pricing change alert goes to VP Sales. An M&A signal goes to the CFO and CEO. A product launch signal goes to VP Product. Each user sees a view tailored to their responsibilities.

SSO (Okta, Azure AD, Google Workspace) is available on Enterprise plans.

No — and this is intentional. Luna does not connect to your CRM, ERP, or HRIS. Your systems push data to Luna — Luna never pulls from your infrastructure. This protects your security, keeps the responsibility clear, and means Luna never depends on your system uptime. Direct API connection to client systems is available only for Enterprise clients with a specific contract.

• Slack and Microsoft Teams (alert delivery)

• Microsoft Excel (MCP native connector or .xlsx export)

• Power BI (native data connector with auto-refresh)

• Salesforce (push alerts into CRM records)

• HubSpot (sync competitive opportunities)

• Zapier and Make (generic webhook for any tool)

• REST API (full data access via Bearer token)

• PDF export (automated, white-labeled, schedulable)

• CSV export (raw signals and historical predictions)

Yes. Luna exposes a complete REST API documented with Swagger (auto-generated by FastAPI). Available endpoints: signals, predictions, reports, threat scores, market trends, competitor profiles. API keys are managed in Settings. API calls consume standard LunaUnits (1 unit per call).

Yes, Luna is fully GDPR compliant. European client data is stored exclusively on EU West servers and never crosses regions without explicit consent. All data is encrypted at rest using AES-256 and in transit using TLS 1.3. Each client’s data is fully isolated using PostgreSQL Row-Level Security — no data can cross between accounts at the database level. Clients can request full data export or permanent deletion at any time through Settings.

• North America: US East servers

• Europe: EU West servers

• Asia-Pacific: Singapore servers (Phase 3)

Your data never crosses regions without your consent.

Luna is a decision-support tool. All strategic decisions remain the client’s responsibility. This is clearly stated at registration and in our Terms of Service. Luna generates probabilities from third-party public data that may be incomplete or inaccurate. We are transparent about this. Luna does not provide professional consulting services.

SOC2 Type II certification is planned for Phase 2 (after $500K in recurring revenue). Current security posture includes AES-256 encryption, TLS 1.3, Row-Level Security, isolated client databases, and regular security audits. Enterprise clients requiring SOC2 today should contact us to discuss current security documentation.

24 hours from registration to first intelligence briefing.

• Registration and company profile: 8 minutes

• Document upload (optional but recommended): 15 minutes

• Personal onboarding call: 30 minutes

• First weekly briefing: delivered the following Monday at 08:00

No 7-week onboarding. No dedicated analyst. No training.

No. Luna is designed for strategy and business professionals, not engineers. The dashboard requires no technical knowledge. If you want API access or Power BI integration, basic technical familiarity helps — but your IT team can handle the connection in under an hour.

Yes. Every new client receives a personal 30-minute onboarding call. Not a Loom video. Not a help article. A real conversation where we configure your competitor list together, set your alert thresholds, and make sure your first predictions are relevant to your situation.

Everything is editable at any time with no fees and no delays. Sector, competitors, alert thresholds, documents, integrations — all updated within 5 minutes. Luna recalibrates automatically after each change.

Settings → Billing → Cancel subscription. Monthly plans end at the current billing period. Annual plans include a 30-day full refund window. No retention calls. No complicated process. We do not believe in trapping customers.