Traditionally, a business user opens a dashboard, applies filters, compares trends, and tries to understand what changed. With conversational AI and natural-language analytics, users can increasingly ask a question and receive a summary directly.That raises an interesting question about the role of dashboards. They may still be essential for exploring trends, checking figures, and investigating exceptions, even if they are no longer the first place users look.
For those building or maintaining BI systems, are users actually interacting with dashboards less because of AI, or is conversational analytics mostly an additional layer on top of existing reporting?

Many organizations build dashboards around KPIs that were defined years ago.

Over time, teams align targets, incentives, and reporting processes around those metrics. The KPI becomes part of how performance is measured.

The challenge is that businesses evolve.

Customer behavior changes.
Products change.
Markets change.
Business priorities change.

Yet many KPIs remain untouched.

As a Business Analyst, how do you determine when a KPI is no longer driving the right decisions?

What signals tell you that a metric has become a reporting exercise rather than a meaningful measure of business performance?

Have you ever had to retire a KPI that leadership was heavily invested in, and how did you justify the change?

I’m trying to fetch data from the Facebook Ads Insights API, but every request returns the following error:

{
  "error": {
    "message": "(#3) Application does not have the capability to make this API call",
    "type": "OAuthException",
    "code": 3
  }
}

The app has a valid access token, and I can successfully make requests to other Facebook Graph API endpoints.

Example request:

GET https://graph.facebook.com/v2.8/act_<AD_ACCOUNT_ID>/insights
  ?fields=impressions,clicks,spend
  &access_token=<ACCESS_TOKEN>

Has anyone encountered this issue with the Insights API specifically?

I’m wondering if this is related to:

• Missing app permissions
• App review requirements
• Ads API access approval
• Business Manager configuration

What was the actual fix in your case?

I’m working on an AI application that uses a large language model for question answering over internal documentation. While retrieval augmentation has improved factual accuracy, the model still occasionally generates confident but incorrect responses when the retrieved context is incomplete or ambiguous.

A simplified version of the inference pipeline looks like this:

retrieved_docs = retriever.search(query, top_k=5)

prompt = f"""
Use ONLY the information below to answer the question.

Context:
{retrieved_docs}

Question:
{query}
"""

response = llm.generate(prompt)

I’ve experimented with increasing retrieval depth, adjusting chunk sizes, and rewriting prompts, but there’s always a trade-off between factual accuracy, latency, and response quality.

For those building production AI systems:

• How do you measure and mitigate hallucinations beyond prompt engineering?
• Have you found techniques like reranking, verification models, or multi-agent validation to be effective?
• What evaluation metrics do you rely on to determine whether changes actually improve factual reliability?

I’m particularly interested in approaches that have worked well in production rather than benchmark experiments.