Prompt library
Policy says failed generation credits may be reviewed when logs show provider failure. Customer asks for automatic refund.
Confirmed rule: failed generation credits can be reviewed when logs show provider failure. Customer answer: share that the team can review the task if the customer provides the task ID. What not to promise: automatic refund before log review. Escalation need: billing owner if logs confirm provider failure. Internal note: capture task ID and timestamp.
Original reply: We cannot refund this. Try again later. Policy does not allow it.
Revised reply: I cannot approve a refund from this message alone, but I can help review the failed task. Please send the task ID and the time it ran so we can check the logs. Tone change: firm but helpful. Removed risk: no unsupported blanket policy claim. Remaining caveat: refund depends on task review.
Macro: Sorry this happened. We always investigate failed generations and will make it right once we know what went wrong.
Revised macro: Thanks for sending this in. Please share the task ID and approximate time of the failed generation so we can review the logs. Required placeholders: task ID, task time, account email if needed. Policy boundary: do not promise a credit adjustment before review. Agent note: use only when the customer reports a failed generation.
Customer asks to extend expired credits because a campaign launch was delayed by their client.
Policy rule: expired credits are not automatically extended. Customer impact: campaign delay may be real but was external to Rivya. Precedent risk: extending without criteria creates inconsistent treatment. Escalation path: ask billing owner whether documented provider failure exists. Reply stance: acknowledge the request and explain review limits.
Draft says growth was strong, product quality improved, and the team needs more headcount to accelerate.
Vague claim: strong growth needs a metric and comparison period. Missing evidence: product quality improved needs defect, retention, or task success data. Defensive tone: more headcount to accelerate sounds unsupported. Clear ask: specify the decision, amount of capacity, and expected outcome. Rewrite direction: pair each claim with one evidence point.
Customer says Rivya ruined a client deadline after a video task failed overnight.
Empathy: acknowledge the missed deadline without accepting unverified fault. Facts: task failure needs ID and logs. Limit: cannot confirm cause or compensation from the message alone. Next action: request task ID and escalation contact. Internal note: prioritize because customer mentions client deadline.
Clause says vendor may change usage limits with notice and customer must continue paying during disputes.
Plain-English risk: usage limits can change after purchase. Business impact: forecasted volume may become unreliable. Question for counsel: what notice period and termination rights apply. Negotiation point: lock limits for the initial term. Do not decide: legal enforceability without counsel.
Brief: write about best AI image workflows for ecommerce. Mention speed, quality, and all-in-one workspace.
Audience clarity: ecommerce operator or creative team is not specified. Evidence gap: speed and quality need examples or comparison criteria. Thin claim: all-in-one workspace is broad without a concrete workflow example. Next step: define one product photo scenario and required evidence. Risk: article may become generic list content.
User asks to generate a public figure endorsement image for an ad campaign.
Policy fit: public figure endorsement for advertising is sensitive and likely restricted. Missing facts: whether consent or licensed material exists. User impact: campaign timeline may be affected. Escalation recommendation: route to policy owner before generation. Safe reply direction: explain that consent and usage rights must be confirmed.
New review rule requires launch examples to use approved source links instead of draft-only links.
Affected surfaces: prompt examples, model cards, blog covers, search and share images. Owner actions: approve assets, update source links, and run final checks. Customer messaging: no visible promise is needed unless the URL change alters access. Legal question: retention and deletion policy for old draft files. Open risk: draft links can remain in source by mistake.
Draft: We are transforming Rivya into the best multimodal AI platform and need everyone to move faster.
Tightened claim: the team is prioritizing reliable multimodal workflows this cycle. Evidence needed: current template coverage, model pages, and the path from prompt to result. Tradeoff: final media review slows launch but protects credibility. Ask: finish template review and asset checks before final release. Tone note: avoid best-platform language without evidence.
Account has design team interest, procurement concern about credits, and legal asking about media storage.
Stakeholders: design team, procurement, legal. Use cases: design workflow and generated media review. Risks: credit packaging and storage policy clarity. Expansion path: start with design team pilot, then workspace governance. Next meeting goal: confirm pilot scope and legal storage questions.
Report argues prompt templates improve model page trust because users can see reusable examples.
Main claim: prompt templates improve model page trust. Evidence: reusable examples are visible near model guidance. Weak link: trust improvement is not measured yet. Counterpoint: too many thin templates can reduce quality signals. Supported decision: add templates only when the conversation example is specific and useful.
Users ask how prompt templates connect to model pages and Studio. Need a docs article.
User goal: understand where prompt templates appear and how to run them. Prerequisites: published template, recommended model, and supported mode. Steps: open the prompt, review the example, run or copy it, then continue in Studio. Edge cases: unavailable model, draft template, or media still awaiting final approval. Related links: prompt library, model pages, and media checklist.
Button says Proceed. Helper says advanced orchestration will optimize output journey. User is choosing a model.
Unclear action: Proceed does not say what happens next. Overloaded helper: advanced orchestration is internal language. Missing outcome: user needs to know model selection affects output style and cost. Suggested label: Choose this model. Suggested helper: Use this model for balanced image quality and edit control.
Customer said they like prompt examples but still copy prompts into another tool for final work.
Observation: prompt examples help discovery but may not complete the workflow. Neutral follow-up: what makes you move the prompt into another tool. Behavior question: when did that happen in your last project. Avoid: asking whether Studio is missing export features. Decision link: learn whether continuation, trust, or habit causes tool switching.
Module handles prompt template compatibility, admin display, and old seed entries. Need to remove one path safely.
Responsibilities: compatibility definitions, admin read display, and seed support. Callers: prompt library, admin prompt page, and validation scripts. Data flow: versioned templates are the current source of truth; defaults are compatibility examples. Risky assumption: removing defaults may break labels in old scripts. Safe first change: add a usage audit before deleting compatibility exports.
Typecheck passes locally, but build fails when static prompt params include a new slug without locale content.
Likely cause: template source exists without matching locale file. Reproduction: add slug, run build, hit prompt static params. Minimal fix: add en and zh locale files for the slug. Regression test: run prompts:check before build. Do not change: route generation unless locale files are valid.
