Cost insights
Cost insights break down what you're spending on AI by feature, widget, and project, with day-over-day deltas so you can catch spikes early.
Cost insights tell you what Informly is costing you and where the money is going. Go to Insights → Cost to open the page. Use it to plan your monthly spend, catch a runaway widget before it eats your wallet, and decide whether you need to upgrade or restructure.
What's on the page
The top of the page shows total spend for the selected period with a delta against the prior period of the same length. Below that, costs break out by feature.
| Feature | What it covers |
|---|---|
| LLM tokens | Tokens sent to and returned from the language model. |
| Embeddings | Vectors generated for document chunks and search queries. |
| Text-to-speech | Voice generation for voice-enabled widgets. |
| Storage | Document storage in your knowledge base. |
| Document processing | One-time cost to ingest a new document — parsing, chunking, embedding. |
Pick a time period
The time period control sits next to the date range picker.
Day
Best for spotting a sudden spike. A day-over-day delta is the fastest way to catch something going wrong.
Week
Best for steady reporting. Weekly numbers smooth out daily noise.
Month
Best for planning. Pair with your plan's usage limits to forecast next month's bill.
Each row shows the cost for the period plus the delta against the prior equivalent period.
Drill down
The drill-down lets you slice cost by dimension and find the cause of a spike.
Per widget
Sort widgets by spend descending. The widget at the top is the one to investigate first if total cost jumped.
Per project
Useful when widgets are organized by department or product line — you see which initiative is consuming the most AI.
Per conversation type
Splits AI-only chats from handoff conversations and voice from text. Voice conversations cost more per minute than chat, so a shift in volume here can move totals quickly.
A sudden 5–10x jump on a single widget almost always means a persona is too verbose, a skill is firing in a loop, or a data source just synced a huge batch of new documents. Drill into the widget, then into a sample conversation, before you change anything else.
Spot a runaway widget
A typical investigation looks like this.
Notice the delta
The top card on the page shows a red delta against the prior period.
Drill by widget
Open the per-widget breakdown. One widget is usually doing most of the damage.
Switch to AI usage
Click through to AI usage for that widget to see per-conversation token counts.
Read a sample chat
Open the longest conversation. If the AI is producing 2,000-token answers to one-line questions, tighten the persona.
Plan ahead
Cost insights pair naturally with the Wallet and your plan's usage limits.
Check this month's run rate
Pick the month-to-date range and look at the projected total.
Compare to plan quota
If you're trending past your included quota, decide whether to top up the wallet or upgrade the plan.
Subscribe to a usage alert
Set up an alert through a notification channel so you're warned before you hit the limit again next month.