Self Learning Release Notes

The following sections list every released version of Self Learning, newest first. Each section comes from the monthly OCP® Release Notes page named in its Released line.

Self Learning 1.22.0

Released: November 2026

New Features

• Backfill that never stalls — Recordings that cannot be studied (missing audio, failed imports) no longer block a project's initial history study; stuck projects recover on their own.

• Controlled cost and trace visibility — Usage, cost and stored-call-trace views are restricted to verified platform staff, with a separate allowlist for raw call traces.

• Gentler dashboards — Behind a flag: Mission Control boards use caching and bounded queries, so a slow database cannot be overloaded by dashboard refreshes.

Fixes

• Security fixes. Plain-text transcripts named .json are imported correctly (already shipped with 1.21.5).

This release includes security updates.

Self Learning 1.21.5

Released: November 2026

Fixes

• Faster, more reliable call imports — Calls imported from your contact centre are no longer downloaded more than once, and imports no longer slow down as the backlog grows.

• Plain-text transcripts saved as .json — Transcript files named .json that contain plain text are now imported as text instead of staying pending. Files that can't be read are marked as failed instead of staying pending, and a corrected file with the same name can be uploaded again.

This release includes security updates.

Self Learning 1.21.4

Released: November 2026

Improvements

• Voice list order — Lexis voices are now listed first, followed by Amazon Polly voices.

Self Learning 1.21.3

Released: November 2026

New Features

• Escalation tests — Concierge tests can now check that conversations escalate to a human queue when they should, and the Agents tab in Testing shows the escalation queues.

Self Learning 1.21.2

Released: November 2026

Fixes

• Voices load for projects not yet deployed — The persona voice list now loads for projects that haven't been deployed yet.

Self Learning 1.21.0

Released: November 2026

New Features

• Word boosting — A new Word Boosting section in Settings lets you preview the words each agent should recognise better, then review and accept them. The accepted lists are applied when you deploy.

• Agent testing — Testing has a new Agents tab for unit-testing concierge, routing and Q&A agents, and all test runs are shown in one combined Runs list.

Improvements

• Long test runs keep going — Test runs now use their own service credentials, so long runs no longer stop when your sign-in session expires.

• Drop files anywhere — Files can be dropped anywhere on the page, even while you're typing in a field.

• Focused QM reviews — A Quality Management review scoped to specific agents now uses only those agents' calls.

Fixes

• Security fixes in the web interface.

This release includes security updates.

Self Learning 1.20.0

Released: October 2026

New Features

• Q&A tests from your knowledge base — In Testing, you can generate Q&A test scenarios from selected knowledge-base documents and see which documents aren't covered yet, so building a test suite takes less manual work.

• Quality Management intelligence — Quality Management now adds standing findings, calibration, recurring review periods and conditions on when a rule applies, giving coaching and quality decisions more context over time.

• Mission Control diagnostics — Mission Control now shows diagnostics and operational insights for each integration, so issues surface without digging through logs.

• New workspace pages — Knowledge, Agent Assist and Project Costs pages are added, along with a first-run tour, a command palette and agent chat available on every page.

• Omilia batch transcription and redaction — Call ingestion can now use Omilia batch transcription, which also separates speakers, and Omilia redaction for those transcripts.

Improvements

• Build AI Agents from Playbooks & Knowledge — A Build AI Agents button now sits next to Upload files, so you can start an agent from the content you just added.

• Deployment progress on every page — Deployment and rollback progress stays visible while you move between Self-Learning pages and keeps running after you leave.

• More reliable Q&A evaluation — Test answers are now judged on facts rather than wording and checked against the expected source documents and for groundedness. Failed answers are re-checked by majority vote, which means fewer false failures.

• Richer agent tests — Agent tests can now check the welcome prompt, conversation history and re-ask behavior.

• Ingestion progress survives a reload — First-ingestion progress resumes after a page reload, with more accurate completed, retired and pending counts.

Fixes

• Accepted FAQs stay visible — After Accept or Accept all, approved FAQs now appear in the Accepted tab straight away.

• More reliable ingestion — Fixes duplicate use-case names, tracking of failed diarized batches, and stalls on silent recordings.

⚠️ Breaking Change: the cost-visibility settings are renamed to COST_FIGURE_EMAIL_DOMAINS, PROJECT_COST_EMAIL_DOMAINS and PROJECT_COST_EMAILS, and the old names are no longer read. Update your deployment configuration before upgrading.

⚠️ Breaking Change: uploads are now limited to 50 MB, and larger files are rejected.

This release includes security updates.

Self Learning 1.19.0

Released: October 2026

• Desktop analytics in Quality Management: for each use case, see how much desktop time, clicks and app switching handling takes. See which agents work differently from the rest of the team, and replay one of their sessions.
  

  

  
  

• No more waiting on first ingestion: it finishes as soon as discovery is done, so you can start working with the agent right away instead of watching the connecting animation.

• Uploaded calls keep their agent, queue and time: Quality Management can now rank agents, filter by queue and sample across days for uploaded calls, just as it does for a connected contact center. Before, every uploaded call landed on the upload day with no agent.

Self Learning 1.18.0

Released: September 2026

• Playbook rules are cards you can edit, so authored checks can be changed in place.

• Quality Management lets you drill into each question and script, with a two-hour grid so you can see coverage across the day.

• A project built from documents opens on the Mission Control board instead of the empty connect pitch.

• A missing agent name on one transcribed call no longer drops the rest of that file.

Self Learning 1.17.0

Released: September 2026

• Testing – Every Self-Learning tenant now has a Testing tab. It tracks pass rate over time, shows how much of the playbook the tests actually exercise, and ranks the cases that fail most often. Each failure opens the conversation behind it in Insights Monitor.

• Quality Management – Every Self-Learning tenant now has a Quality Management tab. It grades real conversations against the playbook, one question per behaviour, so a score names the behaviour that was missed, and it drafts a coaching note for the agent concerned. The AI grader can be calibrated against a human-scored sample, with agreement reported rule by rule.

• Knowledge Review – Where agents gave customers different answers to the same question, a review queue now puts the disputes, the conditional answers and the liability-sensitive ones in front of a person first. What they sign becomes the tenant's knowledge, and a price quoted wrongly once goes on a watch list rather than being lost in the average.

• Connecting a Contact Centre – Connecting a contact centre now studies a sample of past calls immediately and then keeps sampling live ones, so the first use cases appear within seconds. Progress counts what was actually sampled and assessed rather than every call in the tenant, and a backfill with nothing left to fetch ends instead of running forever.

Self Learning 1.16.0

Released: September 2026

• When connecting with Ccaas, operators can pick a Test, Evaluation, or Production ingestion profile (Production is the default).

• The connect queue list shows only active inbound voice queues, so outbound dialer skills no longer clutter the monitor list.

• First-ingestion progress now reaches 100% when some contacts were skipped (no recording) or failed, instead of looking stuck.

• Very short answered-and-dropped contacts are skipped before download, so they do not waste quota or break batch transcription.

• Amazon Transcribe as the speech-to-text provider for audio ingest

• RingCX connect no longer stalls on the credentials step when the browser reports the OAuth popup as closed.

Self Learning 1.15.0

Released: August 2026

New

• Test results can now suggest concrete improvements, which you can review as a diff and apply to the test or the agent playbook.

• Accepted learned FAQs appear as Live, while processed items awaiting publication remain Ready.

• Playbook sections that were previously empty can now be edited directly.

• Generated automations can be traced back to the calls they were built from.

Improvements

• Call processing produces results sooner and stays predictable under concurrent load.

• Sensitive call data stays protected even when transcription or redaction fails, and redaction no longer risks altering unrelated values.

• Reliability improvements across group reset, recording encryption, large knowledge-base publishing, RingCX connection completion, and incremental playbook rebuilds.