The Two Independent Signals
CoDomain tracks two signals for each pattern. Neither signal informs the other before the divergence is computed — they are built independently so the gap between them is real.Market spend
Market spend is an annualized run-rate estimate for each AI agent subpattern and industry vertical. It is triangulated from three source tiers: primary financial disclosures, analyst reports, and procurement signal. Spend figures are always expressed as ranges with confidence bands, never as point estimates. The width of the band reflects the quality and tier of the underlying sources.Field sentiment
Field sentiment is drawn from practitioner discourse across public channels: LinkedIn, community forums, industry newsletters, Substack publications, and conference proceedings. Each statement is sourced and dated before it enters the corpus. Statements are then classified along two axes:- Sentiment cluster — enthusiastic, cautious, or skeptical
- Directional indicator — rising, flat, or falling
How the Signal Is Built
CoDomain’s signal passes through six stages before a human editor reviews it for publication. Each stage handles one domain of the evidence.Financial signal
The financial layer processes primary financial disclosures, analyst reports, and procurement signal to produce spend estimates per subpattern and industry. Confidence bands are derived from source tier — tighter for primary disclosures, wider for analyst estimates. Spend is never expressed as a single number.Market discourse
The discourse layer processes practitioner statements from the cleared corpus. It computes the distribution of sentiment clusters (enthusiastic, cautious, skeptical) and directional indicators (rising, flat, falling) per pattern. Reads below the publication threshold are flagged as directional only rather than validated.Technical fingerprint
The technical fingerprint stage identifies the dominant architecture for each subpattern — the integration surface, execution model, and behavioral boundary that defines what qualifies as that pattern versus an adjacent one. This is what produces the qualifying boundary statements in each frame: the line between an agent and a chatbot, between orchestration and sequential chaining.Operator signal
The operator signal stage processes deployment evidence — case studies, operator disclosures, and practitioner accounts — to identify structural patterns in what is working and what is stalling. This produces the mechanism language in the Pattern section of each published frame.Technology infrastructure signal
The infrastructure stage tracks deep technical design decisions, architectural implications, research directions, and development patterns that shape how deployment categories form and evolve. This includes execution models, memory architectures, orchestration primitives, evaluation frameworks, safety infrastructure, and the research-to-production pipeline. Infrastructure signal is treated as a leading indicator. Where the technical substrate is maturing faster than deployment conviction has formed, CoDomain notes the gap. Where deployment is outrunning infrastructure readiness, CoDomain notes that too.Divergence synthesis
The synthesis stage aggregates outputs from all prior stages, resolves conflicts between signals, computes a divergence score for each pattern, and assigns the direction label and verdict. When signals disagree, the more cautious read is applied unless the evidence clearly supports otherwise. The narrative prose in each frame — Signal, Pattern, Decision, Risk — is generated from synthesis output and reviewed by a human editor before publication.Source Standards
CoDomain applies a tiered source standard. Spend confidence bands are derived directly from the tier of the underlying sources.| Tier | Category | Examples |
|---|---|---|
| Tier 1 | Primary financial disclosures | SEC filings, disclosed ARR, funding round valuations with revenue multiples, earnings call data, direct company financial statements |
| Tier 2 | Analyst reports | Grand View Research, Mordor Intelligence, MarketsandMarkets, Precedence Research, Fortune Business Insights, Menlo Ventures State of AI, and equivalent research firms with documented methodologies |
| Tier 3 | Procurement signal | CIO surveys, enterprise technology spending surveys, procurement pattern data from credible institutional sources |
| Discourse | Practitioner channels | LinkedIn posts, industry newsletters, community forums, Substack publications, conference proceedings — anonymized to role and company type |
Publication Standards
CoDomain publishes only frames that meet the methodological standard. A refresh may generate candidate frames for every pattern and industry in the lens. Only frames where the evidence supports a responsible read are approved for publication. Frames that do not clear the threshold are held as draft or marked Insufficient signal. This is intentional. CoDomain is not trying to fill every slot for visual completeness. A map with 11 published frames and 9 held frames is more trustworthy than a map that publishes 20 frames with manufactured directions. The publication thresholds are:- Field sentiment — n≥20 sourced statements within a 90-day rolling window
- Spend — ≥2 Tier 1 or Tier 2 sources per subpattern
Key Caveats
The map refreshes periodically. Each refresh produces a new versioned publication with a changelog. Where a direction changes materially between versions, the changelog notes the mechanism behind the shift.Questions about the methodology? Contact us at hello@cosentriq.com