Identify it. Validate it. Protect it.


iCONECT AI uses Continuous Active Learning (CAL) learning based on targeted human input to deliver a ranked population which allows users the ability to prioritize their review to zero-in on the most relevant information faster than any other tool on the market.*

Finding EVERY relevant document is critical to a thorough investigation. iCONECT AI’s Exhaustive Sequential Classification Technology identifies and prioritizes documents and provides results that may never have been found using keyword searching alone.


iCONECT’s Artificial Intelligence engine delivers results in real-time, regardless of the size of the overall document population. iCONECT AI identifies the information that is needed today, not three weeks from now. Simple graphs and charts communicate project progress.


Missing key documents due to reviewer misjudgments can undermine an investigation before it starts. After identification of example documents, iCONECT AI harnesses that knowledge and quickly returns relevant information in real-time without the need to do rounds of review.


By reducing the number of people necessary to comb through documents, iCONECT AI helps lower overall costs associated with review by as much as 80% over traditional methods.


How this process works for you

iCONECT’s CAL method is analogous to a music streaming service “choosing” which songs a user will enjoy based on their previous song selections. Users review and tag a small portion of the data set which is used to train the iCONECT prediction engine (a mathematical model) to order data from most to least (likely to be) relevant. A model is built for each tag. With each new document tagged, the model becomes more accurate for each tag. After a model reaches a certain quality measure, the model can be applied to the entire data set.

iCONECT AI sits between Processing and Review in the traditional eDiscovery workflow and relies on the document content so it can handle all types of data, including email and office files. The system allows for the review of multiple issues in a single pass.

iCONECT can seamlessly integrate additional data on a rolling basis during the training of the model. This is because the iCONECT prediction engine ranks the entire information set together, rather than a limited number of randomly selected documents. New documents added to the population simply join the ongoing ranking process. After the model is set, it can be applied to any additional documents to generate further results ‘with one click’.

Benefits of iCONECT Predictive Review

  • Effortless Setup – Utilizing folder templates and an intuitive wizard, begin your predictive review project with confidence, and stay in sync with client and project needs.
  • Streamlined Issue Identification and Review – Easily classify and review a document for multiple issues, saving the need for multiple reviews of the same document per issue.
  • Intuitive Review Feedback – Reduce and simplify review with smart document identification. Use iCONECT’s unique confidence bar to help determine relevance, and soft checks that guide predictive coding suggestions.
  • Built-In – Predictive Review is built right into the iCONECT interface, minimizing user training and enabling you to take advantage of the platform’s intuitive workflow without utilizing a different interface.
  • Leverage XMPLAR® – Using the power of Xmplar, iCONECT’s “custom find similar” functionality, identify only the portions of a document that matter: create a sample document from documents outside the collection, or from anecdotal input.
  • Comprehensive Reporting – Monitor how expected metrics and scores are met or exceeded using iVIEW Data Visualizer, which displays charts and graphs that can be shared with team members to monitor progress.
  • On Demand – iCONECT Predictive Review is licensed on demand. Available as enterprise or per project pricing, you have the flexibility to include Predictive Review in projects that can most benefit.

Powered by Sentio


In a Foreign Corrupt Practices Act investigation with a dataset of 804,661 documents written in English, Spanish, and a combination of both languages, 808 records were issue coded for responsiveness. The iCONECT Analytics engine was able to identify 382,237 as irrelevant. This saved 14,332 hours in review time and $1.5 Million in attorney time.

Want to learn more? Download the AI Bake Off White Paper below.

*some additional charges may apply.