Talk to your Brand.
Not your files.

Most AI tools store your documents and retrieve them. CrawlQ Studio does something structurally different: it converts your brand knowledge into 28 typed entity nodes and their relationships, then uses that Brand AI Memory as the reasoning surface itself. When Athena answers a question about your brand, it does not search your files — it traverses entity nodes, follows typed edges, and weights answers by confidence scores accumulated from everything you have ever defined. Brand AI Memory is not a source the AI consults. It is the substrate the AI thinks through.

• 28 entity types across 6 dimensions (Brand, Audience, Voice, Market, Content, Strategy) are extracted from your foundation documents and held in Brand AI Memory as typed nodes — not chunks.
• Each entity has a confidence score gated at ingestion: ≥0.70 auto-accepted into Brand AI Memory, 0.40–0.69 queued for your review, <0.40 rejected. Nothing enters the reasoning substrate without earning its place.
• Ask Memory (live, June 2026): POST a natural-language question against your Brand AI Memory and receive an answer constructed by traversing entity relationships — not by embedding lookup. The reasoning path is traceable.
• When the AI generates content, it traverses Brand AI Memory from your defined personas, through their pain points, across to your vocabulary nodes, checking against your anti-pattern list — all in a single multi-hop pass. Not retrieval. Reasoning.

Your brand intelligence becomes the reasoning surface. Not the context window.

Retrieval-Augmented Generation works by cutting documents into chunks and pulling the closest match to your prompt. For brand intelligence — where context, relationships, and voice rules must all be honoured simultaneously — this approach has five structural failure modes that no prompt engineering can fix.

• Fragmentation: your brand strategy brief is 2,400 words of connected reasoning. At 400-character chunks, the logic that flows between paragraphs is severed. Brand AI Memory keeps every connection as a typed relationship.
• Cross-reference blindness: your persona's core pain point and the vocabulary your brand uses to address it live in separate chunks. They are never retrieved together. In Brand AI Memory, they are a typed edge — always traversed together.
• Temporal staleness: RAG has no mechanism for 'this supersedes that.' Your 2024 positioning and your 2021 positioning carry equal embedding weight. Brand AI Memory has versioned, confidence-weighted nodes.
• Context collapse: five foundation documents at once exceeds any practical context window. Brand AI Memory traversal activates only the nodes relevant to the current question — no information missed, no context overflowed.
• Hallucination fills gaps: when RAG cannot find the right chunk, the model generates from its training data — which is the internet average, not your brand. When Brand AI Memory traversal reaches a confident node, it uses your actual knowledge.

The difference is not better retrieval. It is a different reasoning architecture.

Every competitor that shows you a 'brand score' is asking an LLM to estimate it. That score drifts with the model version, the prompt phrasing, and the random seed. CrawlQ Studio's Brand Maturity Score is calculated by a formula engine. Zero LLM calls. Sub-millisecond. The same inputs always produce the same number. And each output earns one of four official quality gates.

• Foundation Alignment (FA, 25%): how many of your named entities — specific values, personas, differentiators — appear in the content versus the total claims made. Evidence ratio, not vibe.
• Coverage Depth (CD, 20%): does the content address all key conceptual areas — audience, positioning, evidence, action — detected through synonym group matching, not keyword counting.
• Audience Alignment (CI, 20%): readability, structural formatting, and action language. Content that cannot be navigated or acted on scores low regardless of brand accuracy.
• Source Balance (SB, 15%): Shannon entropy across your foundation documents. Content drawing 80% from one source scores low even if the voice is technically correct.
• Voice Coherence (VC, 20%): every defined brand term used raises the score. Every named anti-pattern lowers it by 30 percentage points. Three violations and Voice Coherence approaches zero.

85+ = Brand Certified. 70–84 = Brand Verified. 50–69 = Draft Quality. Below 50 = Needs Attention. The gates have names, not just numbers.

Every piece of intelligence the TAMR+ pipeline extracts from your documents — every entity, every relationship, every brand claim — is scored before it enters Brand AI Memory. This is the confidence gate: a three-tier admission system that keeps your Brand AI Memory honest. It is not a feature you configure. It is the architectural guarantee that your reasoning substrate only contains knowledge that has earned its place.

• ≥0.70 confidence: auto-accepted. The entity enters your Brand Memory immediately and becomes available as a reasoning node for all future generation and scoring.
• 0.40–0.69 confidence: pending review. The entity is surfaced to you for approval before it becomes part of your Brand Memory. You decide, not the AI.
• <0.40 confidence: auto-rejected. The extraction was not confident enough to trust. It is discarded, not stored. Brand AI Memory stays clean.
• Why this matters: Brand AI Memory that accepts every extraction is not a reasoning substrate — it is noise. The confidence gate is what makes traversal trustworthy. Every node in Brand AI Memory passed a threshold.

Other tools show you everything the AI found. We show you only what the AI is confident about.

Most brands think they have a content problem. Many have a foundation problem. They have been optimising output without investing in input — asking the AI to score 90 when the ceiling imposed by their foundation is 55. CrawlQ Studio calculates the maximum Brand Maturity Score your current foundation makes possible, and tells you exactly what to add to raise it. No other brand AI tool does this.

• The ceiling formula: 70% from how many foundation documents you have uploaded, 30% from how many entities those documents produced. It is deterministic — no estimation.
• Five foundation documents and fifty accepted entities gives you a ceiling of 100%. Two documents and fifteen entities gives you a ceiling of roughly 42%.
• A ceiling of 42% means: even the most perfectly structured content cannot score above 42. Not because the writing is bad — because the foundation is incomplete.
• The ceiling is a mirror. It shows the gap between the brand you think you have built and the brand you have actually defined in a form the reasoning substrate can traverse.

You cannot score what you have not defined. And you cannot defend what you cannot measure.

Every scored piece of content generates a SHA-256 cryptographic fingerprint of three inputs: the dimension scores, a hash of the content itself, and the weighting schema used at the moment of evaluation. If any of those three change after the fact — score altered, content modified, weights adjusted retroactively — the hash changes and the tampering is visible. The tamper evidence is structural, not policy.

• Every AI-generated document carries an EU AI Act Article 50 provenance badge: which model generated it, which brand data it was grounded in, which decision produced it, and when.
• Any AI decision can be replayed three days or three years later under the same conditions. If your content is later modified and attributed to you as liability, you hold cryptographic evidence of what you actually generated.
• The defensibility threshold is explicit in code: aggregate score ≥70 AND Coverage Depth ≥60 AND no gaps in the required foundation document set — all three conditions simultaneously. A high aggregate alone does not make content defensible.
• No other brand AI tool in this market ships this. It was built because one day someone will ask for it, and 'the AI said so' will not be a sufficient answer in a regulated market.

Keep the receipts. Before they are required.

Every brand knows what it is. Almost no brand has written down what it refuses to be. In CrawlQ Studio's Brand AI Memory, ANTI_PATTERN is a typed entity — the same structural weight as a PERSONA or a VALUE. Every time generated content uses language your brand has explicitly named as off-brand, the Voice Coherence dimension drops by 30 percentage points per violation. The system does not just reward what you do right. It penalises what you said you would never do.

• Using your brand vocabulary correctly is table stakes — it means the AI remembered who you are. That is basic functionality. Any tool can do it if you upload enough examples.
• Allowing an anti-pattern through means the AI violated an explicit brand rule — a failure of a different order. The asymmetric penalty (−30pp per occurrence) reflects that asymmetry.
• Anti-patterns are not the opposite of your patterns. They are the specific, named ways your brand fails under pressure — the jargon that appears when you are trying to sound credible, the vague aspiration that appears when you have nothing specific to say.
• The most structurally valuable thing you might do for your brand this week is not writing better content. It is naming three things your brand will never say — and adding them as ANTI_PATTERN nodes to Brand AI Memory.

What you refuse to be defines you more precisely than what you claim to be.

Every document you upload, every question you ask Athena, every piece of content the system scores — adds accepted entities and relationships to Brand AI Memory. Brand AI Memory does not reset between sessions. The reasoning substrate deepens with use. This is not a feature of the product. It is a consequence of the architecture: Brand AI Memory accumulates intelligence in a way that a retrieval system structurally cannot.

• Early stage: Brand AI Memory contains the entities extracted from your first foundation documents — personas, values, voice attributes, key differentiators. The reasoning substrate is grounded but narrow.
• As you use the product: every question asked, every asset scored, every new document uploaded feeds the confidence-gated ingestion pipeline. Brand AI Memory grows in precision, not just volume.
• Over time: Brand AI Memory begins to surface Strategic Insight signals — cross-entity connections that span your audience nodes, your market nodes, and your voice nodes simultaneously. These are visible as high-confidence relationship paths, traceable to the evidence that produced them.
• The switching cost builds automatically. Your entire brand intelligence — every accepted entity, every confidence-weighted relationship, every historical score — lives in Brand AI Memory. No competitor can reconstruct that history.

The longer you use it, the smarter it gets about your specific brand. Not brands in general. Yours.

The large AI platforms have a business model built on one thing: the more you use AI, the more tokens you burn, the more you pay. Every session resets by design. That is not a bug. It is the revenue model. CrawlQ Studio's Brand AI Memory makes that model obsolete for brand work — and that is precisely why no large player will build it. A system where intelligence accumulates in Brand AI Memory and token consumption decreases over time is structurally opposed to a per-token business.

• The advanced reasoning layer runs multi-hop traversal over Brand AI Memory — the same class of technology used in financial crime investigation and intelligence analysis at firms like Palantir. Your brand intelligence is reasoned over, not retrieved from a vector index.
• Three European patent filings (pending) on the Brand AI Memory reasoning methods and the confidence-gating architecture. The algorithms are proprietary. The integration depth cannot be replicated by reading how the product works.
• Small players cannot copy it — the confidence gate mechanics, multi-hop entity traversal, and the deterministic scoring engine together represent years of proprietary research. The moat is in the mathematics, not the interface.
• Large players will not build it — it requires deliberately designing a system that reduces per-session token consumption as the knowledge base matures. That directly cannibalises a per-token revenue stream. The incentive structure prevents it.

One founder. Three patent filings. Five years. The architecture is the moat.

CrawlQ launched on AppSumo in 2021. The platform broke. It was rebuilt entirely — 104 days, 721 pull requests, 1,829 commits, one founder — and 10,000+ accounts were migrated before the announcement was made. Not one tier was downgraded. Not one feature was withdrawn. Not one account was asked to start over. This is not marketing positioning. It is a verifiable engineering record.

• Your lifetime access does not expire and does not lapse for inactivity. Sign in with your AppSumo email at studio.crawlq.ai and your tier, credits, and workspace limits activate automatically.
• The add-ons (Brand Radar $19/mo, Strategy Engine $49/mo, Brand Command Center $69/mo) cover real Brand AI Memory compute and frontier model inference cost — costs that grow with reasoning depth and complexity. Stated plainly. The base product stays whole. Upgrades are a choice, not a condition.
• Lifetime cohorts get new capabilities first. Onboarding redesign, BYOK, the Governance tab, and Ask Memory all went to LTD holders before general availability. That sequencing is deliberate and continues.
• The reasoning substrate described in these guides is the same infrastructure your lifetime deal runs on. When Brand AI Memory gets richer, your deal benefits without renegotiation.

The deal is permanent. Brand AI Memory compounds. The founder reads every message.