Natural Language Report Builder
Describe the report you want and get a configuration you can check before it runs.
What it adds
A plain-language entry point to the app's report builder that produces a normal, editable report configuration.
What your agent is told to do
5
What your agent is told to do
5-
1
Expose the app's defined metrics, dimensions, and grouping options to the model as an explicit vocabulary, and have it return a report configuration in the same shape the builder already saves. It must not invent a metric name or a calculation.
-
2
Show the generated configuration in the builder for review and require the user to run it. Do not execute a potentially expensive aggregation on the model's first guess.
-
3
Maintain a controlled glossary mapping business terms to defined metrics — what active means, what a qualified lead is, which revenue figure counts — and resolve ambiguous language through that glossary rather than letting the model decide each time.
-
4
Validate the configuration against the user's data permissions before it can run, and reject combinations of dimensions that would expose restricted records through a small-group breakdown.
-
5
Once generated, the report must be an ordinary saved report that runs without any model involvement. Filters on a plain list view are owned by Natural Language Filters; this feature covers metric, grouping, and date-range configuration only.
Edge cases it handles
8
Edge cases it handles
8- A request for a metric the app does not define must be refused by name. Restricting the model to the declared vocabulary is what prevents a plausible-looking report built on a number nobody computes.
- The configuration must be visible and editable before the query runs, because a misread grouping can turn a cheap report into one that scans the whole dataset.
- Business terms mean different things in different workspaces. Route them through the glossary and show which definition was used, rather than letting the model pick a reasonable meaning silently.
- Some dimension combinations expose individuals even when each dimension alone is permitted. Block breakdowns that fall below a minimum group size or that cross a permission boundary.
- A saved report must reproduce identically later without calling the model again. Persist the resolved configuration, not the original sentence, or the report changes meaning between runs.
- If the model returns a configuration that fails validation, keep the user in the builder with their request intact and explain which part could not be resolved.
- Report runs can be slow and expensive. Estimate the cost of the generated configuration and warn before running anything unusually large.
- When the model is unavailable, the manual report builder must remain fully functional with the natural language entry point clearly disabled.
Definition of done
9
Definition of done
9- Generated configurations use only metrics and dimensions the app defines.
- The configuration is shown for review and requires an explicit run before any query executes.
- Ambiguous business terms resolve through the glossary and the definition used is visible.
- Combinations that would expose restricted data or fall below the minimum group size are blocked.
- A saved report reruns identically with no further model involvement.
- Validation failures return the user to the builder with an explanation and their request preserved.
- The manual builder remains fully usable when the model is unavailable.
- The feature matches the existing design system.
- No existing functionality is broken.
Related features
AI Insight Cards
AI Insight Cards
Turn a report into a few grounded cards, each tied to a number the app calculated.
What it does
A small set of cards summarising notable movements in a report, each linked to the query and figure that produced it.
How it works
- 1 Detect candidate movements in the app first, using the report's own aggregations. The model ranks and phrases the candidates it is given; it does not go looking for them and it does not produce the arithmetic.
- 2 Bind each card to the query, metric, and period that support it, and make the card link through to the filtered view so a reader can check it in one click.
- 3 Set a materiality threshold before generation, in both relative and absolute terms, so a swing on a metric with three events does not outrank a real change on a metric with thousands.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/ai-insight-cards
AI Trend Analysis
AI Trend Analysis
Explain what a chart is actually showing, in sentences backed by computed numbers.
What it does
A written summary attached to a time series or metric set, describing direction, magnitude, and comparison period.
How it works
- 1 Compute every number in ordinary code before the model is involved: the change, the rate, the comparison period, the baseline, the seasonal adjustment. The model writes prose about figures it is given and must never produce a figure of its own.
- 2 State the comparison window and the time zone in the summary text, resolved from the workspace's configured zone rather than the server's. A change described as week over week is meaningless without saying which weeks.
- 3 Instruct the model to describe what the data shows and to stop there. Do not let it assert causes, attribute movement to campaigns or releases, or predict what happens next.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/ai-trend-analysis
AI Key Point Extraction
AI Key Point Extraction
Reduce long content to the decisions, facts, and requests that actually matter.
What it does
A short list of the salient points in a long document or thread, each tied back to the passage it came from.
How it works
- 1 Apply this where length is the problem: long threads, meeting notes, call transcripts, lengthy tickets, multi-page documents. Content that already fits on a screen does not need extracting.
- 2 Classify each point by kind — decision, fact, risk, request, open question — so a reader can scan for the category they came for instead of reading a flat list.
- 3 Carry a reference to the source passage with every point and let the reader jump to it. A point nobody can verify is a claim, not a summary.
Copy the prompt
No account needed
Add this feature to my app:
https://addthisfeature.com/x/ai-key-point-extraction
How it works
-
1
Copy the link
Grab the Markdown instruction URL for this feature.
-
2
Give it to your AI
Paste it into Claude Code, Cursor, v0, Lovable — whatever you build with.
-
3
It inspects, then implements
Your agent reads your existing app first, then adds the feature to fit it.
Works with your stack
These instructions are written to adapt. They tell the agent to detect your framework, match your existing design system, and reuse what you already have — rather than assuming a particular stack.
Need it tighter than that? Customize the feature and tell it exactly what you're running.