Reports, Journey Templates & Rebuilt Brand Detection

Features

• Reports Page — Reporting moved out of a modal and onto its own page. Pick the sections you want, preview the report inline before you commit to a download, and read a month-over-month strip of the headline numbers above it.
• Report History — Generated reports are kept rather than streamed once and discarded. Every report you have run is listed with its period and sections, and can be previewed, downloaded again, or deleted.
• Scheduled Reports — Put a report on a weekly or monthly cadence and have it generated and emailed without anyone opening the app.
• Prompt List & Change History in Reports — Reports now carry the exact prompt text each execution ran against and a dated record of every prompt you added, edited, paused, or removed during the reporting period, so a number can be traced back to the question that produced it.
• AI Platform Panel in Reports — A section stating which AI platforms, models, and model versions the report's numbers actually came from, and how much of the period each one covered.
• Cited Brands Per Prompt — Reports list the brands each platform actually named in its answer, prompt by prompt, instead of only reporting your own.
• Journey Template Gallery — The Agent Journeys create page opens on a browsable gallery of ready-made journey briefs grouped by category, rather than a blank task field.
• Suggested Journey Templates — Templates suggested from your own site, so the first journey you run is about a page you actually have.
• AI Conversions in Analytics — The Google Analytics summary band now surfaces conversions attributed to AI referral traffic alongside the sessions figure.

Improvements

• Rebuilt Brand Detection — The detector behind mention and citation counts was rebuilt. It no longer reports your brand as present in answers that never named it, it recognizes far more of the genuine mentions that were previously missed, and where a mention is genuinely ambiguous it now abstains rather than guessing. Accuracy is measured against a held-out evaluation set rather than asserted.
• Redesigned Report PDF — The generated PDF was rebuilt on the product's typography and ink palette, so a report you send to a client looks like the product it came from.
• Provider Naming in Prompt Tracking — Provider identifiers that arrive with a vendor prefix now resolve to the right vendor instead of falling through to a generic label, and provider logos sit centered in their rows.
• WordPress Drafts by Default — Content created through a WordPress connection now lands as a draft, so nothing reaches a live site without a deliberate publish.
• Retired Model Removed — Prompt execution on the ChatGPT side moved onto a current model and off two that OpenAI has retired, so tracked prompts stop failing against models that no longer answer.

Fixes

• Signed-In Requests — Fixed a regression that caused authenticated requests to fail, and hardened the code path so the same shape of mistake cannot recur silently.
• Spurious Sign-Outs — A rejected request now verifies the session is genuinely dead before signing you out, instead of treating any authorization failure as an expired login.
• Publishing Confirmation — The write-confirmation step works again on every publishing surface, not just the ones it was originally wired into.
• Journey Errors No Longer Masked — A journey that stops safely now reports the step error that caused it instead of presenting the stop as an ordinary finish.
• Fan-Out Answer Quality — Duplicate prompt variants are removed before execution, and variants that leaked your brand name into the question they were meant to ask neutrally are now rejected.
• Report Presets — Fixed executive-preset options that silently did nothing, and section descriptions that described a different section than the one they sat under.
• Share of Voice Reliability — A partial read now degrades visibly instead of failing the page, and two analyses created at the same moment no longer collide.
• Google Analytics Sync — Overlapping syncs can no longer write over each other; a sync now takes an exclusive lease and its writes are rejected if that lease has moved on.
• Journey Runs on Existing Data — Fixed a case where accounts with existing journey history could not receive the template update at all.