Control delivery
Define the result and the audience that should receive it.
Feature flags
Your agent controls feature delivery by percentage, property or cohort. Exposure is recorded on the person, so the behavior and failures around your rollout are one question away.
Choose boolean or multivariate decisions and targeted rollout rules.
Record which decision a customer received.
Follow captured behavior, errors and release context around exposure.
01Targeted delivery
Ask your agent to define a flag with percentage, property or cohort rules. Check matching and non-matching customers, and give clients a safe default before you widen the rollout.
02Exposure and experience
A recorded exposure shows which result a customer received. Your agent follows the events, replay and technical evidence around it and reads rollout health, without treating a correlation as a proven cause.
The workflow, connected
Connect the delivery rule, the recorded decision and the customer evidence around your change.
Ask your agent for the flag, its variants, its audience rules and an application default.
Test intended matches and record exposure with the same customer identity.
Your agent reads exposure health and compares the behavior and failures around the change.
The capabilities behind it
Choose who receives a feature and keep the evidence that helps your team understand the rollout.
Define the result and the audience that should receive it.
Inspect recorded decisions and the customer activity around exposure.
Bring release registration, comparisons and regressions into the investigation.
Put it to work
Use percentage or property targeting and test the expected decisions before widening the rollout.
Explore this workflowUse cohorts or a supported saved audience generation for a deliberately selected group.
Explore this workflowFollow the flag decision into the customer journey and compare the surrounding technical evidence.
Explore this workflowYour first useful workflow
Keep client-visible flag configuration free of secrets, and record exposure using the intended customer identity.
A few useful answers
Flags support percentage, property and cohort targeting, with boolean or multivariate decisions.
An exposure records which decision a customer received. Compare the surrounding evidence before concluding that a change caused the problem.
Product analytics contains the separate product-experiment workflow. Flag evaluation does not assign experiment treatments.
Feature flags control delivery. Product experiments use a separate two-arm, fixed-horizon measurement workflow; evaluating a flag does not assign experiment treatments.
Keep a safe application default and surface the evaluation failure in diagnostics. Review the SDK guide for evaluation, caching and exposure capture.
Anectico / Private preview
Explore targeted delivery and exposure-linked investigation with your agent in private preview.