Over the past two weeks, we’ve been at work picking through ELLA with a fine-toothed comb before we get back to new feature development — over 100 individual bugfixes, tweaks, and improvements. Some of these fixes have been rolling out over the last few weeks already, while others were included in a big push today.
We’ve been focusing heavily on the experience advisors feel when using sensemaking. In particular, we’ve been focusing on:
Speed — research across business factors and documents should hang less and finish sooner.
Reliability — chat send/stop/mic states are more consistent, long replies scroll more cleanly, and interrupted streams are more likely to resume cleanly if your connection hiccups.
If Sensemaking used to feel “stuck” mid-turn or the send button flickered the wrong state, this release targets that. The experience using Sensemaking continues to be a focus, so please share your feedback with how you’re using AI chat.
You can now thumbs-up or thumbs-down Sensemaking replies, and optionally leave a short reason. That feedback helps us improve answer quality without training on your client content.
There’s also a workspace setting to allow manual quality review of anonymized AI responses. When it’s off, vote buttons stay disabled with a clear path to settings. When it’s on, you can rate answers in context — and the quality-review prompt no longer sits as a permanent banner once you’ve already consented.
AI inputs and outputs are never used for model training, with or without this setting.
General UI
Migrated icons from Lucide to Phosphor across the application for visual consistency
Added truncation for breadcrumbs in the top-nav
Simplified the new client animation and removed an unused "Goal" icon from business factor cards
Fixed positioning of the trial limit banner on the Add Client button
Removed an invalid CSS rule that broke brand color styling
Page subheaders (including Fact Finding) are more consistent
Editing client business info no longer fails when industry is unset.
Client logos: higher size limit, no infinite flash on failed image lookup from our logo provider, actions hidden when there’s no logo to show.
Fact Finding
Added an "Edit" button to the template preview drawer
Added a scope filter for fact-finding templates on the templates library page
Owner-guides sidebar polish: clearer naming, add-owner-guide, better empty states
Improved performance when switching between fact-finding guide screens
Settings
Fixed a spurious "saving" indicator on load for About This Engagement in settings and standardized AI Preferences and AI Memory on the same autosave pattern
Deliverables
Advisor cursors no longer show anonymous animal names during deliverable collaboration
Multiple improvements to Deliverables editing stability
Trust & Reliability
Further hardened document storage access so workspace files are more tightly scoped in how long a download link can be accessed, and by whom
Clearer error handling when something goes wrong server-side (including support-friendly error references)
