Analyst Notes, daily editions, briefings, signal interpretations, watch conditions, quotables, and other editorial Content displayed through the Service are produced by the Company's Generation Pipeline, which includes automated detection algorithms and large language model ("LLM") inference. This Content is AI-Generated Content and is subject to the limitations, disclaimers, and liability carve-outs set forth in this Section.
A. How Analyst Notes Are Produced
The Service produces Analyst Notes through a multi-stage Generation Pipeline that may include:
- Deterministic signal-detection algorithms that fire when observable on-chain metrics cross configured thresholds
- Large language model inference that produces natural-language interpretation of detected signals, including summaries, why-it-matters commentary, actionability suggestions, watch conditions, and quotables
- Automated publication gates that filter low-confidence, low-novelty, or low-corroboration outputs
- Outcome evaluation that compares prior signals to subsequent on-chain state to populate the public call log
The Company may use one or more third-party AI providers and may change providers, models, prompts, evaluation logic, or other components of the Generation Pipeline at any time without notice. Methodology information is published at https://rollupradar.com/methodology for informational purposes; the methodology page does not constitute a warranty of accuracy or fitness for any particular purpose.
B. No Per-Note Human Review
You acknowledge that AI-Generated Content is produced and published programmatically at scale. The Company does not warrant that any individual Analyst Note has been reviewed, verified, fact-checked, or edited by a human prior to publication. While the Company maintains quality controls, prompts, evaluation gates, and editorial guidelines designed to improve Analyst Note quality, those controls operate at the system level and do not constitute per-Note review.
C. Known Limitations of AI-Generated Content
You expressly acknowledge that AI-Generated Content has known limitations, including but not limited to:
- Hallucinations: AI systems may produce statements that sound authoritative but are factually incorrect, including but not limited to incorrect protocol names, incorrect chain deployments, incorrect historical references, incorrect addresses, incorrect attribution of activity to specific entities, and incorrect claims about external systems.
- Stale Training Data: The AI models used in the Generation Pipeline have training data cutoffs and may not reflect current protocol deployments, partnerships, governance changes, contract addresses, security incidents, depegs, migrations, or other facts that have changed since training.
- Pattern-Matching Errors: AI systems may incorrectly apply patterns observed in training data to current data that has different underlying dynamics, leading to misleading interpretations or attributions.
- Confidence Calibration: Confidence scores, signal strength indicators, and similar metadata are heuristics produced by the Generation Pipeline and do not constitute reliable predictors of accuracy. A high-confidence Note may be incorrect; a low-confidence Note may be correct.
- Speculation Presented as Analysis: AI-Generated Content may include speculative interpretations, attributions, or hypotheses presented in declarative language. Hedging language such as "typically," "historically," "worth verifying," "in our prior coverage," "based on prior patterns," "worth checking," or similar epistemic markers signals that the underlying claim may not be directly observable from the data; the absence of such hedging language does not warrant accuracy.
- Outdated Information: Notes are time-stamped at generation but remain accessible after their underlying data may have changed. The Company is not obligated to update Notes as conditions evolve.
- Inconsistency Across Notes: Different Notes generated by the same Pipeline may use different framings, attributions, or interpretations even where the underlying data is similar.
D. User Responsibility to Verify
You expressly acknowledge that you bear sole responsibility for verifying any factual claims, protocol references, market observations, attributions, historical comparisons, or other statements made in Analyst Notes before relying on them for any purpose. The Company does not warrant the accuracy of any specific claim in any specific Note. Independent verification against canonical sources (block explorers, official protocol documentation, primary data sources) is the User's responsibility.
E. Editorial Modifications, Withdrawals, and Corrections
The Company may, in its sole discretion and at any time, modify, withdraw, retract, or correct any Analyst Note, signal, briefing, watch condition, or other AI-Generated Content without prior notice and without obligation to preserve prior versions. The Company makes no commitment to correct errors in any specific timeframe, to notify Users of corrections, to maintain historical accuracy across Note revisions, or to provide an audit log of changes. Users should not assume that any AI-Generated Content is current, accurate, or unmodified at the time of access. The Company may publish corrections at https://rollupradar.com/corrections or similar surfaces in its sole discretion.
F. No Warranty of Accuracy, Completeness, or Fitness
The Company makes no representations or warranties of any kind, express or implied, regarding the accuracy, completeness, reliability, currentness, or fitness for any particular purpose of any AI-Generated Content. AI-Generated Content is provided "AS IS" and "AS AVAILABLE" without warranties of any kind, and you use such Content entirely at your own risk.
G. AI-Generated Content Is Not Investment Advice
Without limiting Section 4 (No Financial Advice), AI-Generated Content — including Analyst Notes, watch conditions, historical analogs, signal interpretations, quotables, and any other editorial output — expressly does not constitute investment advice, trading recommendations, financial planning, market forecasting, or any other form of professional advice. AI-Generated Content is editorial commentary on observable on-chain data and should not be used as a basis for financial, trading, treasury, allocation, or other decisions.
H. No Liability for AI-Generated Content
You expressly agree that the Company shall have no liability whatsoever for any losses, damages, missed opportunities, foregone gains, costs, or other consequences arising directly or indirectly from AI-Generated Content, including but not limited to:
- Errors, hallucinations, inaccuracies, or omissions in any Analyst Note or other AI-Generated Content
- Incorrect protocol references, deployment statuses, contract addresses, entity attributions, or historical claims
- Misinterpretation of on-chain data by the AI system
- Stale, outdated, or superseded information presented as current
- Investment, trading, treasury, allocation, hedging, or other decisions made in reliance on AI-Generated Content
- Failure of AI-Generated Content to identify risks, opportunities, security incidents, depegs, exploits, or material events
- The AI system's failure to recognize or properly interpret novel or unusual patterns
- The AI system's tendency to produce confident-sounding language for uncertain or speculative claims
- Modifications, retractions, or corrections to Notes that affect prior decisions
- Any other consequence of the inherent limitations of AI-generated content
This limitation applies regardless of whether the Company knew or should have known of the limitations of its AI systems, and applies in addition to (and not in limitation of) the disclaimers and liability caps set forth in Sections 11 and 12.
I. User Feedback on AI-Generated Content
The Service may provide mechanisms for Users to flag, rate, or provide feedback on Analyst Notes (including but not limited to upvote/downvote affordances and error-flagging functionality). Such feedback is provided to the Company on a non-confidential basis and may be used by the Company without restriction to improve the Generation Pipeline, train or fine-tune models, or for any other purpose. Submitting feedback does not create any obligation on the Company to respond, correct, or act on such feedback. The Company may publicly display aggregate feedback metrics in its sole discretion.
J. No Reliance on Methodology Disclosures
The Company may publish methodology, prompt structures, evaluation criteria, signal-detection logic, or other descriptions of the Generation Pipeline. Such disclosures are provided for informational and transparency purposes only and do not constitute warranties that the Pipeline will perform as described, that any specific Note was produced in accordance with the disclosed methodology, or that the methodology will not change without notice.