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.

The intelligence API surfaces AI-generated insights produced by Daylit’s agent layer. Each insight has a target_type (customer, invoice, or portfolio), an insight_type key that identifies the model that produced it, and a payload_json object whose shape varies by type. You can list, retrieve, and submit thumbs-up/down feedback on any insight.

​
List insights

GET /api/intelligence/insights
Returns a paginated list of active insights. company_id is required — the endpoint returns an empty result set if you omit it or if your account does not have membership in the specified company.

​
Query parameters

​
company_id
string
required
UUID of the company. Membership is enforced server-side.
​
target_type
string
Filter by target scope: customer, invoice, or portfolio.
​
target_id
string
Filter by the specific target’s identifier (UUID or string ID).
​
insight_type
string
Partial match on the insight type key (e.g. signal matches all signal types).
​
customer_id
string
Filter to insights linked to a specific customer UUID.
​
invoice_id
string
Filter to insights linked to a specific invoice UUID.
​
active_only
boolean
default:"true"
When true, only returns insights where valid_to is null (still active).

​
Response

​
results
array
List of insight objects.
curl --request GET \
  --url "https://your-instance.com/api/intelligence/insights?company_id={company_id}&target_type=customer&insight_type=signal" \
  --header "Authorization: Token {token}"

​
Get insight

GET /api/intelligence/insights/{pk}
Returns a single insight including the full payload_json.

​
Submit feedback

POST /api/companies/{company_id}/intelligence/insights/{pk}/feedback
Submits or updates your thumbs-up/down rating on an insight. Each user can have at most one feedback record per insight; subsequent calls overwrite the previous rating.
​
rating
string
required
up or down.
​
comment
string
Optional free-text comment explaining your rating.

​
Response

curl --request POST \
  --url "https://your-instance.com/api/companies/{company_id}/intelligence/insights/{pk}/feedback" \
  --header "Authorization: Token {token}" \
  --header "Content-Type: application/json" \
  --data '{"rating": "up", "comment": "Accurate signal — customer did pay late."}'

​
Payload schemas by insight type

The payload_json field on each insight has a type-specific structure. The following sections document the shape for each supported insight_type.key.

​
customer_conversation_summary

A concise summary of all customer-level communications.
​
summary
string
Up to 18 words, clause style. Empty string if no communications exist or content is unsafe.
​
tone
string
Conversation tone: cooperative, neutral, frustrated, avoidant, unknown, amicable, or defiant.
​
active_court
string
Which team currently owns the next action: AR, Sales, Operations, Executive, or Customer.
​
action
string
One concrete next step to advance the conversation. null if no communications.
​
action_breakdown
object
​
last_touch
object
​
commitment_date
string
YYYY-MM-DD promise-to-pay date, or null.
​
predicted_pay_date
string
YYYY-MM-DD model-predicted payment date, or null.
​
evidence_trace
array
At most one evidence/reasoning line supporting the summary. null when no communications exist.
​
unrelated_items
array
Topics mentioned in the conversation that are not related to the main collection issue.

​
invoice_conversation_summary

A summary of invoice-level communications using a three-question framework.
​
stress_level
integer
1–10 stress level. 1 = Proactive/Routine, 10 = Total Silence/Potential Loss.
​
active_topic
string
The immediate to-do or active talking point in the conversation.
​
desired_outcome
string
What resolution looks like — the north star of the conversation.
​
blocker
string
Why the desired outcome has not been achieved yet.
​
last_touch
object

​
customer_signal / invoice_signal / portfolio_signal

A detected material change signal at the customer, invoice, or portfolio level.
​
entity
object
​
signal
object
Present when a material change was detected. null when no change was found (in which case reason is populated).
​
reason
string
Explanation for why signal is null. Only present when no material change was detected.

​
invoice_pay_date_prediction

An AI prediction of when an invoice will be paid, with supporting evidence.
​
predicted_date
string
YYYY-MM-DD predicted payment date. null if insufficient data is available.
​
basis
string
Primary basis for the prediction: promise (explicit commitment), historical (past payment patterns), pattern (statistical behavior), or judgement (analyst inference).
​
evidence_sources
array
Data sources used (e.g. communication_thread, payment_history).
​
alternative_scenarios
array
Other possible payment dates with probabilities.

​
customer_strategy_plan

A recommended collection strategy for a customer, including email, call, and internal follow-up action options.
​
strategy_plan
object
The recommended strategy.