AI Consulting Services | Novel Cognition

AI Optimization Consulting Since 2017

Novel Cognition was founded in 2017, years before the terms “AIO” or “GEO” entered the marketing lexicon. We are not an agency that pivoted to AI in 2023. Our consulting practice builds on a database career spanning 42 years and a decade of entity-architecture research, formalized in frameworks like Distributive Authority Networks (DAN), Hidden State Drift (HSD), and the NovCog Atlas semantic index.

Every engagement produces artifacts—audit reports, dashboards, schema libraries, agent configurations—that internal teams can operate long after our sprint ends. We are not a vendor selling a recurring license; we are a practitioner team teaching clients how AI retrieval systems see their brands.

A clear distinction: we are not Cognition Labs (the maker of Devin AI), nor are we associated with Ben Goertzel’s AGI work. Novel Cognition is a Denver-based AI optimization consultancy, operating a 22-domain network and three dedicated Surface Forge zones.

Tier 1: AI Brand Audit — $5,000, 2 weeks

Most brands have never measured what an LLM “knows” about them. The AI Brand Audit delivers a quantitative and qualitative baseline.

Deliverables:

• Entity-graph audit: a crawl of Wikidata, DBpedia, and Google Knowledge Graph entries for your brand and key executives, mapped to known gaps and contradictions.
• Perplexity citation baseline: 50 branded queries executed across Sonar, Sonar Pro, and GPT-4o search; we record every citation, its domain, and its position.
• Bing and Google retrieval check: same queries, scored for your owned-surface vs. third-party coverage.
• Action plan: a prioritized, 90-day roadmap for fixing entity-graph errors, improving schema, and increasing owned-surface density.

At the end of 2 weeks, you’ll have a 20-page report (sample available on request) and a clear picture of your brand’s AI footprint. This is the diagnostic step before any buildout.

Ideal for: companies with a defensible knowledge graph (products, people, events) and a suspicion that AI platforms misrepresent them. Also appropriate for agency groups who need an entity audit for a portfolio brand.

Tier 2: AIO Buildout — $25,000–$75,000, 90 days

When the audit reveals structural visibility gaps, the AIO Buildout deploys our owned-surface methodology at scale. This is a full implementation engagement.

What’s included:

• Surface Forge zone creation: we build (or strengthen) three canonical publishing zones—typically a research subdomain, a lexicon, and a newsroom—each architected for LLM retrieval rather than human traffic alone. See our Surface Forge methodology for the technical design.
• Schema rewrite: all entity-describing pages (Organization, Person, Product, Event, Article) receive JSON-LD aligned to the Schema.org vocabulary, extended with sameAs links to Wikidata, Crunchbase, and verified social profiles.
• Atlas backlink injection: our NovCog Atlas (245,000+ indexed articles) provides topical adjacency links to reinforce entity associations across our network and partner properties.
• Wikidata Organization reconciliation: we edit and stabilize your Wikidata item with valid citations backing every claim, reducing the risk of drift.
• Measurement infrastructure: a Perplexity citation dashboard updating nightly, plus event-tracking pixels on all Surface Forge pages, fed into a Looker or GDS instance you own.
• Verification sprint: at day 80, we re-run the original audit’s 50 queries and publish a delta report showing exact citation improvements.

The fee range reflects scope: number of brands, languages, and whether we’re building from scratch or hardening an existing surface. A typical B2B SaaS brand falls near $45,000.

Ideal for: venture-backed startups losing share of voice to aggregator sites; public companies preparing for AI-search implications in the next earnings cycle; professional services firms where partner-bio and case-study pages are the primary conversion surface.

Tier 3: Embedded Fractional CAIO — $10,000+/month

For organizations that need ongoing AI-search operations without the overhead of a full-time CAIO. We embed as part of your executive team, typically 4–10 hours per week.

Core responsibilities:

• Ongoing AIO operations: monitoring the Hidden State Drift framework for entity drift, updating schema, and managing the Perplexity citation pipeline.
• Content infrastructure: directing internal content teams to produce LLM-optimized assets, not to please Google’s core updates, but to increase entity recall across multiple AI interfaces.
• Model selection and architecture: advising on local vs. API model deployment, retrieval-augmented generation (RAG) patterns, and MCP server design, with production-grade experience on pgvector, Qdrant, Pinecone, and Weaviate.
• Agent architecture review: evaluating autonomous agent designs (e.g., OpenClaw, CrewAI, Autogen) for safety, cost, and retrieval-quality trade-offs, drawing on our NovCog Dev experiments.
• Quarterly board-ready reports: a 3-page summary of AI-search shifts, citation trends, and recommended budget adjustments.

We typically begin with a 90-day AIO Buildout before the retainer starts, ensuring baseline infrastructure exists. The monthly retainer scales with scope; $10K is the minimum for a single brand in English. Multilingual support, multiple brands, or deep RAG integration increase the fee.

Ideal for: companies with $50M+ revenue who cannot justify a full-time CAIO; tech firms that want an independent AI architecture reviewer not tied to a cloud vendor; PE/VC portfolio support where multiple brands need fractional oversight.

How We Work: Evidence, Not Promises

Every engagement follows a 4-phase cycle that prioritizes measurement over motion.

Phase 1: Baseline (Week 1). We capture pre-engagement metrics: LLM query responses, citation rankings, entity-graph snapshots, and share-of-owned-surface. No action is taken until we’ve quantified the starting position.

Phase 2: Intervention (Weeks 2–10). Changes are rolled out in controlled batches—schema updates, new surface pages, Wikidata edits—each tracked with a “before/after” delta.

Phase 3: Re-measure (Weeks 11–12). The identical query suite is re-run. A variance report documents every shift, annotating which interventions likely caused which changes. We don’t claim credit for uncorrelated movements; LLM inference is stochastic, so we report confidence intervals from multiple probe runs.

Phase 4: Hand-off (Day 90). All dashboards, schema files, editorial processes, and agent configurations are transferred. The client owns the operational stack—we maintain no lock-in.

This process is derived from our 8-year track record with clients in B2B SaaS, professional services, publishing, and deep tech, documented in select research papers and our case studies (NDA-permitting).

Artifacts You’ll Receive

To avoid abstraction, here are sample deliverables from real engagements (anonymized):

• Entity Graph Report: 30-page PDF mapping every Wikidata and DBpedia inconsistency for your brand, plus a side-by-side comparison with top competitors.
• Perplexity Citation Dashboard: A Looker Studio report refreshing daily, showing your domain’s citation position across 50+ queries in Sonar Pro and GPT-4o search.
• Schema Library: A validated JSON-LD file set for Organization, Product, Person, and Article types, hosted in your CMS or CDN with version control.
• Owned-Surface Scorecard: A weighted metric (0–100) combining citation frequency, entity appearance, and retrieval position across four AI platforms—updated monthly during retainers.
• CAIO Monthly Brief: For retainer clients, a 3-page PDF summarizing retrieval changes, emerging threats (e.g., new competing Wikidata entities), and recommended actions.

These are not “reports that gather dust.” They are operational objects that your internal team, agency partners, or board can act on immediately.

Who Should Not Engage Us

This consulting is not for everyone. A careful filter saves both sides time:

• If your primary goal is ranking higher in traditional Google SERP, you need an SEO agency. We focus on entity authority for AI interfaces—Google, Perplexity, ChatGPT, Bing Copilot—where the retrieval mechanism is fundamentally different.
• If you need a fast, cheap “prompt engineering” fix, we are not the right fit. Our work is structural: schema, knowledge graph, owned-surface density. It compounds over months, not days.
• If your brand lacks any meaningful entity-graph presence (no Wikidata item, no structured data, no verified Google Knowledge Panel), we can help but the baseline cost will be higher. We won’t take on engagements where the likely outcome is a low-impact delta report.

We are selective: we typically accept 3–4 new engagements per quarter to maintain quality. If you’re unsure, a 30-minute discovery call is zero-pressure.

Next Steps

If you recognize your brand’s situation in the tiers above, the most productive first step is getting in touch for a short, structured call.

We’ll ask for a few things beforehand: your domain, your current AI-search concerns (specific mis-citations help), and whether you’ve run any internal audits. No pitch deck, no heavy presentation—just a conversation between practitioners.

From there, we can recommend an engagement path, provide a sample brand audit (NDA-ready), and set expectations for timeline and investment. Email engage@novelcognition.ai to start.