Skip to main content

Documentation Index

Fetch the complete documentation index at: https://help.daylit.com/llms.txt

Use this file to discover all available pages before exploring further.

Daylit’s AI continuously analyzes your customer payment history, open balances, communication threads, and behavioral patterns to surface actionable intelligence for your AR team. Rather than reviewing every account manually, your team can focus on the customers the AI has flagged — and act on a recommended plan instead of starting from scratch.

What AI signals are

An AI signal is a pattern detected on a customer account that warrants attention. Signals are generated by analyzing a combination of payment behavior data and communication history. Common signal types include:
  • Payment behavior worsening — The customer is paying materially slower than their own historical baseline.
  • Severe or chronic late pattern — The customer has a persistent history of paying late across multiple invoices.
  • Accelerating deterioration — Lateness has worsened month over month in recent periods.
  • Balance growing — Open balance is increasing while payment velocity is flat or declining.
  • Ghosting — The customer is unresponsive to outreach despite having an overdue balance.
  • Churn risk — Open balance has dropped sharply, the customer has gone quiet, and remaining balance is deeply aged.
  • Frequent short payer — The customer routinely pays less than the invoiced amount.
  • Healthy account — No collection action needed; the account is current or carries a zero balance with no behavioral risk flags.
  • Payment behavior improving — The customer is paying materially faster than their historical baseline.
Each signal also carries one or more tags from a standardized taxonomy so your team knows exactly what pattern was detected (for example, payment_behavior_worsening, ghosting, or high_portfolio_impact).

Signal severity levels

Every signal is assigned a severity to help your team prioritize:
SeverityMeaning
HighImmediate attention required. The AI has detected a significant risk pattern or the account represents a material portion of your AR.
MediumAction recommended soon. The pattern is concerning but not yet critical.
HealthyNo collection action needed. The account is in good standing.
NoneNo signal generated — usually because there is insufficient data or no material change from baseline.
The Customers table defaults to sorting by signal severity (high first) so the accounts that need the most attention appear at the top.

Conversation summary

The AI reads every email and logged communication in a customer’s thread and generates a concise summary of where things stand. The summary answers:
  • What is the current status of the collection conversation?
  • What is the most recent thing that happened?
  • Who needs to take the next action — your team (AR, Sales, Operations, or Executive) or the customer?
  • Is there a commitment date or a predicted payment date in the conversation?
The summary also includes a tone assessment (cooperative, neutral, frustrated, avoidant, defiant, or amicable) based on the language used in the thread.
Conversation summaries are generated from synced email communications and manually logged calls and notes. The more complete your communication history in Daylit, the more accurate the summary.

Collection strategy plan

For each customer, the AI generates a recommended next step — either send an email, make a call, or escalate internally. The strategy plan includes:
  • Recommended action — The specific action type (email, call, or internal follow-up) the AI recommends right now.
  • Reasoning — A brief explanation of why this action is recommended given the account’s current state.
  • Urgency — A score and bucket (Critical, Today, This Week, Upcoming, or Info) indicating how time-sensitive the action is.
  • Email draft or call script — If the recommended action is an email, the plan includes a pre-written draft with the appropriate tone and template. If it is a call, it includes talking points and anticipated objections.
  • Anticipated outcome — What the AI expects to happen if you take the recommended action.
You can act directly on the strategy plan from the AI Insights tab or from the Action center.

How to find insights on the customer detail page

1

Open the customer detail page

Click the customer’s name in the Customers table.
2

Select the AI Insights tab

Click AI Insights in the detail page navigation. You will see three sections: Signal, Conversation Summary, and Strategy Plan.
3

Review the signal

The Signal section shows the active signal title, severity, tags, and the evidence behind each tag — the specific metric or behavioral observation that triggered it.
4

Review the conversation summary

The Conversation Summary shows the AI’s read of the full communication thread: current status, tone, who holds the next action, and any commitment or predicted pay date.
5

Review the strategy plan

The Strategy Plan shows the AI’s recommended action, urgency, draft email or call script, and the reasoning behind the recommendation.
You can also see a condensed signal indicator on each row in the Customers table — the AI Signal column shows the severity badge and the primary signal tags.

Giving feedback on an insight

You can rate any AI insight to help improve signal quality over time. On the AI Insights tab, use the thumbs up or thumbs down buttons next to any insight section. A thumbs up confirms the insight was accurate and useful. A thumbs down flags it for review and optionally lets you leave a short note explaining why it missed the mark. Feedback is shared with Daylit’s AI team and used to calibrate the models for your account.

How insights are refreshed

AI signals, conversation summaries, and strategy plans are refreshed automatically on a scheduled basis as new payment data and communications are synced. You do not need to trigger a refresh manually under normal circumstances. If you want to force a refresh — for example, after logging an important call or receiving a payment — click Run AI analysis from the customer detail page. Daylit queues the analysis and updates the insights within a few minutes.
Insights improve as more historical data accumulates. A customer with only one or two closed invoices may show a “None” signal simply because there is not yet enough data to establish a baseline. Signals become more reliable over time.

Managing customers

Search, filter, organize, and act on your customer accounts in the Customers table.

AI insights

Learn how Daylit’s intelligence layer generates signals, summaries, and plans across customers and invoices.

Action center

Work through AI-recommended collection actions prioritized by urgency.

Invoice payment predictions

See how the AI predicts when each open invoice will be paid.