Chat, analysis and named features

Answered on 5 October 2026: both recommendations accepted. The earlier proposal grouped similar names as possible synonyms. The definitions and examples show that some of them answer different questions. I recommend preserving those differences while making them easy to browse together.

The recorded decisions keep chat and analysis tasks separate from evidenced AI participation. This does not approve automatic changes to existing classifications.

Question	Recommendation
Does Chat Interaction mean the same thing as AI Chatbot Interaction?	Keep Chat Interaction broad. Record an AI participant when evidence supports it, and offer AI Chat as a narrower browsing choice.
Does Data Analysis mean the same thing as AI Analysis?	Keep the task separate from AI involvement. Make AI-assisted data analysis findable through both facts.

No new answer is needed for either choice. The product-feature example below applies an existing rule; the feature-category and naming requirements you added are recorded on the Products page.

1. A chat exchange and its participants

Already decided: a brief activity inside a video describes that moment. It does not turn the entire video into that kind of piece. A product's appearance and its role also remain separate. This choice concerns the narrower meaning of two proposed moment descriptions.

Recommendation: Chat Interaction describes a visible exchange of messages. Whether an AI assistant participates is another fact, supported by the source or the content. Someone looking for chat references can see the broader set, then narrow to AI Chat. Unknown participants stay unknown. A conversation can involve both AI and people.

Checked example: Intercom's chat guide describes Fin handing a conversation to a human support team. The conversation remains a chat even though its participants change. This is a publisher-documented example; I have not tested that handover in our library.

In our collected HiBob video, HiBob turned AI into a flywheel for product and team growth, the selected stored interval is labeled Chat Interaction. An inspected frame shows an exchange discussing AI use cases. That supports a chat-interface example, but the appearance alone does not prove who produced a response. Keep the existing assignment pending interval verification; do not turn it into an AI assignment just because the video's title mentions AI.

Counterexample: a pictured conversation between two people can be Chat Interaction without being AI Chat. An empty chat input or a marketing claim about an assistant does not establish a visible exchange. These are illustrative boundary cases.

Alternative: keep Chat Interaction and AI Chatbot Interaction as separate, related descriptions, with broader searches explicitly including the narrower one. That can preserve the distinction too, but the relationship and search behavior need maintaining. Treating the names as interchangeable would lose whether AI participation was actually evidenced.

What your answer affects: definitions, prompts and validation must ask both what is shown and what the source supports about the participants. Results and trackers must preserve the same distinction. A chat label must not invent a product, company, campaign or maker relationship. The storage design and conversion of old assignments follow the technical audit.

2. The analysis task and AI involvement

Already decided: the library keeps descriptions at the scope supported by evidence, with defined filters and richer explanatory text. The remaining question is how these two proposed descriptions should relate.

Recommendation: Data Analysis describes examining data to produce findings, comparisons or conclusions. Record AI involvement separately when supported. AI-assisted data analysis can then be found both by its task and by AI involvement. An AI label alone does not tell us what task occurred.

Checked example: OpenAI's data-analysis guide describes analyzing uploaded spreadsheets and producing tables or charts. It documents a case where the task and AI involvement coexist. This source is a text example of the distinction, not a newly verified video classification or a test of the service.

The selected library video Replying to @internal_glowup: How is ChatGPT making your life better? has a stored AI Analysis interval whose description says it concerns makeup coverage. The frame inspected within that interval instead presents translation. That discrepancy makes this assignment unsuitable as evidence that AI Analysis means Data Analysis. Its cause and the rest of the interval still need checking.

Counterexample: a person examining sales data can perform Data Analysis without demonstrated AI involvement. A recorded AI conversation that only translates a phrase does not become Data Analysis merely because an AI system is involved. These examples define the intended boundary; they do not certify our existing labels.

Alternative: retain a separately defined AI Analysis description for presented interpretation of supplied material, alongside Data Analysis. This may be useful for image or document interpretation, but its inclusion rules would still need review. Neither option should infer hidden reasoning from an interface animation or merge the two names as synonyms.

What your answer affects: collection must retain supporting source claims, and analysis must identify the task actually shown. Search, saved trackers and cited answers should distinguish observed activity from a publisher's capability claim. Human corrections and original evidence must survive later reanalysis. This meaning choice does not approve exact fields, a new universal filter list or a bulk migration.

A named product feature — apply the existing rule

The Products page already separates named features from the work demonstrating them. You do not need to approve that relationship again.

Nike's LeBron 17 page describes Max Air in the heel and Zoom Air in the forefoot. In sampled frames of the collected Introducing the LeBron 17 video, the caption names the Heel Max Air Unit. The specific name helps someone find that feature. A reusable description of how it is presented helps someone find comparable creative references across products.

The old proposal asks whether to promote Heel Max Air Unit as a general moment description. Applying the existing product model points toward retaining it as feature information with an evidenced product connection. The exact feature identity, interval and any replacement moment description still need verification. This does not authorize deleting the old label or claim that the product's performance has been independently tested.

What happens next

The evidence note records the prior proposal, selected database checks, inspected sources, media discrepancies and the consequences for processing. Sampled frames do not establish complete motion or audio behavior. No old assignments will be converted on that basis.

The product-use follow-up is answered: Hands-on Interaction includes deliberate handling and adjustment as well as operation. The sports-related descriptions remain in preparation. Framing and the Offer distinction are now approved in their moment and promotion records. Exact definitions and mappings still need verification; those approved meanings will not be asked again.

Your answer settles the two explicit recommendations. The earlier prototype and Brand Events discussion must be reconciled before specifying how reusable feature categories, narrower capabilities and product-specific names are stored or searched. Historical proposals are not automatically current vocabulary. The review plan retains the full decision pass and subsequent schema, pipeline and workflow audit.

Prepared 5 October 2026.

Last reviewed 5 October 2026.