Product Newsletter — August 2026
August was a big one. You can now choose which AI model powers your AI Colleagues. Analytics will tell you what your employees keep asking about, including the questions you are quietly failing to answer. Knowledge Management stopped letting syncs die in silence. Helpdesk got much better at triage, in 25 languages. And Settings grew up and became the Admin console.
The short version
Your Orchestrator and every AOP can now run on a model you choose.
Query Clusters turns thousands of requests into themes, ranked by volume.
Start a big one, close the tab, get a link when it lands.
Connectors nag you before a token expires, instead of dying quietly.
Helpdesk category prediction moved onto an LLM and hit around 90% accuracy.
Restructured navigation, four job-based roles, and your bookmarks survive.
🧠 AI Colleagues Studio
Pick the brain behind your AI Colleagues
One platform default used to power everything, whether it suited the job or not. Now you choose.
| Provider | Models |
|---|---|
| OpenAI | GPT 5.5, GPT 4.1 |
| Azure OpenAI | GPT 5.5, GPT 4.1 |
| Anthropic | Sonnet 4.6, Opus 4.5, Opus 4.6, Opus 4.7 |
| Amazon Bedrock | Sonnet 4.6, Opus 4.5, Opus 4.6, Opus 4.7 |
Each AOP has a new model field that quietly follows your Master Orchestrator unless you tell it otherwise. Set it once, or fuss over every AOP.
Effort decides how hard the model thinks before it opens its mouth, and is available everywhere except GPT 4.1. Priority processing is a fast lane for when latency matters, on OpenAI GPT 5.5 for now. Whichever model actually ran is stamped onto every run, so Run History, the Debugging console and evaluation runs stay honest even after you change the setting.
Nothing moves until you move it. Leave the default alone and every AOP keeps running exactly as it did yesterday.
How model configuration works →
Let your identity provider handle the door
Consent screens irritate a lot of people. There is now an Enterprise-Managed Authorization section under Settings → MCP server & A2A. Switch it on and Leena trusts tokens issued by your own identity provider, so users are signed in silently. Three things to set:
- Your JWKS URL, which we use to check token signatures.
- Your Issuer, so tokens from anywhere else are turned away.
- Claim mappings for subject, email and employee ID, if you want them. Optional.
Switch it off and everyone goes back to the standard sign-in screen, with your configuration saved for later. Needs MCP and A2A enabled.
Papercuts, if you were affected by any of these
- Find & Replace in the AOP instructions editor behaves itself now: highlighting, overlapping dialogs, scroll-to-match, jump-to-top
- Tools scoped to an AI Colleague appear in its Tools tab even before you have used them in an AOP
- MCP actions come with descriptions, plus a searchable View all side sheet
- Slack-triggered runs show the channel conversation the AI Colleague actually picked up, summarised and expandable, in both the Debugging console and Run History
📚 Knowledge Management
The month we stopped syncs from failing in silence.
An expired token used to take a sync down without a word. Set an expiry date, a warning window between 1 and 90 days, and who gets nagged. Previously only Confluence and SharePoint offered this.
"Unknown error" was doing a lot of heavy lifting. Click a Failed status now and you learn whether it is your credentials, the source system, or us, with the next step attached.
Articles used in a test run vanished from the picker forever, which made "fix it and re-test" impossible. Prior use is now just an icon and a filter.
General Settings now asks one question: fast and reliable, or slower and highly accurate? The model underneath both options also got an upgrade, so answers improved everywhere.
New default OCR engine. Scanned PDFs, rotated pages, right-to-left languages and complex tables all behave, and password-protected files say so instead of producing an empty draft.
HRSD-style scripted user criteria are now honoured. ServiceNow evaluates its own rules, we enforce the outcome at query time, and deny always wins. We never read or store your script contents.
The full fix list
- Connector setup only warns about unsaved changes when there are, in fact, unsaved changes
- UAT and Staging bots show production-inherited settings as read-only rather than throwing an error
- Article expiry dates stopped showing a day early in US time zones
- One enormous sync no longer blocks every other bot's sync, and several stuck-in-syncing cases are cleared
- Source document URLs are captured across every connector, so View source and View Policy show up regardless of origin
- Webview highlighting scrolls the cited passage into view, and tables highlight only the rows that matter
- Content no longer arrives out of order or goes missing, and low-confidence blocks are kept rather than binned
- Special characters survive in non-English content, and hidden HTML stays out of your articles
- Images in answers are better chosen, with exact caption matches winning
🎫 Helpdesk
Triage moved onto an LLM
The old NLP service missed a lot and needed constant keyword-mapping to stay useful. It is gone. Category and sub-category prediction on email tickets and live chat now runs on an LLM, at up to 90% accuracy.
You can add free-text instructions per department to steer predictions towards how you actually work. It layers on top of the standard logic, so you are nudging behaviour rather than rebuilding it. And the keyword-mapping chore largely goes away.
It cannot go rogue, and it cannot block a ticket. Prediction only chooses from your department's active category tree, so it can never invent a category you never created. If it fails or comes back empty, the ticket is still raised with the department's fallback category.
Everything else in Helpdesk
| What changed | Why you care |
|---|---|
| 25 dashboard languages, up from 8 | Arabic, Hindi, Japanese, Tagalog, Haitian Creole and 20 more. Full list → |
| Inactive User Report | One list of everyone missing from the Employee Master or marked Terminated, instead of finding out when a ticket sits unassigned because its owner left in March. Docs → |
| Dependencies block deactivation | Active tickets, SLA policies and reports must be cleared before a user can be switched off. Slower than one click, and that is the point. Docs → |
| Unavailability covers open tickets | Picking a replacement used to only redirect new tickets. You can now hand over existing active ones too. They do not come back automatically when the period ends. |
| Comments save themselves | Navigate away, leave Helpdesk, or have your laptop die mid-sentence. The draft is still there. |
| Nudges in the timeline | Pending-on-user chases are now recorded, so you can see who was nudged and when. |
| Faster reports | Plus watcher email details now visible in Bot View. |
📊 Analytics
Find out what you are not answering
Query Clusters is a new tab inside the Handled and Not handled drill-downs. Rather than scrolling through thousands of individual requests, you get themes, and how much volume each one drives.
| Column | What it tells you |
|---|---|
| Name | An auto-generated title for the theme. No manual tagging. |
| Articles | The knowledge base articles the NLU reached for across the cluster. |
| Tools | The tools and skills it invoked. |
| Volume | How many requests sit behind the theme. |
Start with the Not handled tab. It is a ranked list of everything your AI Colleagues are failing at, ordered by how much it is costing you. That is a roadmap written by your own employees. The Handled tab is the mirror image: which themes are being resolved, and what is doing the resolving.
Clustering works on meaning rather than keywords, so "how do I apply for leave" and "leave application process" land together. Strays go to a Miscellaneous bucket instead of being forced somewhere wrong, and once enough similar ones pile up they graduate into a cluster of their own. Greetings, filler and test messages are filtered out, so your volumes are not flattered. Everything syncs hourly.
Fire off an export and walk away
Click export. Keep the tab open. Do not navigate away. Do not close your laptop. Hope it does not time out at minute nine. Discover the date range was empty only after opening the file.
Pick your date range and up to 10 recipients. The dialog checks the range actually contains records before you submit. Then close the tab and get on with your day.
When it is ready, recipients get a secure download link that stays live for 7 days. Every export sits in one place with its status, searchable and filterable, refreshing itself while something is still cooking. Failed ones retry, finished ones re-run, both pre-filled from the original request.
🔌 Integrations & Connectors
Seven new connectors. The first five are MCP-based, which means the agent discovers every tool the app offers the moment you link it. No hand-building actions one at a time, and anything those apps add later shows up on its own.
Find, read and share files.
Search, read and send mail.
Post messages, search people and channels.
Work with contacts, companies and deals.
Search across Jira, Confluence and connected Atlassian apps.
Privileged-account and access tasks handled through the agent, instead of bouncing someone into a separate console.
HR, payroll and workforce data, so employees can self-serve pay and time questions inside the assistant.
⚙️ Workflows Studio
The AI can build the tables it needs. If a workflow needed a data asset that did not exist, you had to abandon the flow, go build it in the Workflows Admin console, and start again. It can now create DOA tables, Master Tables and Data Models mid-generation, either because you asked ("create a DOA for expense approvals by department and amount") or because it worked out the workflow needs one. Everything it creates is put in front of you before it saves, and duplicate names are blocked. Docs →
A Stop button. You can now stop a generation you already know is going nowhere, instead of watching it finish out of politeness. Re-running means pasting the prompt back in, which is a small price.
Delete the drafts you never published. Unpublished workflow apps and AI-generated chats can now be deleted. They used to just accumulate. Deeply unglamorous, long overdue.
🏛️ Admin Console
Settings grew up. Same URL, new name, tidier layout. Old audit log bookmarks redirect themselves.
| Area | Before | Now |
|---|---|---|
| The section itself | Settings | Admin console |
| Channels | Channels | External channels |
| Virtual assistant | Virtual Assistant | Leena Channels |
| Bot authentication | Its own folder | Gone. Its pages sit directly under Leena Channels as Apps authentication, Web authentication, Other auth settings and SAML apps |
| Widget configuration | Buried in VA settings | Its own Widget config item under Leena Channels |
| Audit Logs | A top-level module | Admin console → Logs, alongside Web login errors, Email logs and WhatsApp logs |
| Data Subject Requests | Elsewhere | Admin console → Logs |
| Role | Gets |
|---|---|
| System Admin | Everything |
| Channel & Authentication Setup Manager | Authentication, all of Leena Channels, External channels, Email settings |
| Audit & compliance manager | The whole Logs section, including Audit Logs and Data Subject Requests |
| Template manager | Templates |
Config Manager has been retired into System Admin, and Audit Logs – Admin / Agent is replaced by Audit & compliance manager. Anyone who already holds those keeps their access. Use the new roles from here on.
- Hidden now means blocked. Opening a restricted page by URL returns Access denied rather than an empty screen, so a link shared between two admins on different roles will only work for one of them.
- Setup is System Admin only. A Channel & Authentication Setup Manager can configure authentication but cannot open Dashboard settings → Setup.
- Logs belongs to Audit & compliance manager. No other delegated role reaches those screens.
- Templates adapts to whoever opens it. Any role with at least one template type gets in, and sees only its own tabs. A Notification Admin lands straight on WhatsApp templates.
🛠️ Platform & Global Settings
Workvivo is now a channel, set up from Global Settings like any other. Setup guide →
Everything else here is the long tail: a month of things that quietly stopped going wrong. Open whichever one you have been complaining about.
Channels
- Slack — Notifications arrive with their attachments in the same thread, and you can see the bot thinking while it works
- WhatsApp — Messages arrive in the right order with formatting intact, and Tags are visible and filterable in logs
- Google Chat — Tables in agent responses render properly
- SMS — Delivery status in analytics now matches what actually reached the handset
- Links that open desktop apps are clickable again
Voice
- Users can rate a voice call
- Internal system messages stopped appearing in transcripts
- Email addresses render correctly in transcripts, and enabling voice on an autonomous agent no longer errors out
AI Colleagues and Studio Assist
- Studio Assist degrades gracefully on long tasks instead of going silent, plus a round of interface improvements
- Quick replies show up reliably. Most were being generated and then dropped before anyone saw them
- Run explanations are based on the complete picture rather than partial results
- Duplicate responses that were blocking people mid-conversation are fixed
- Streaming and multi-chat are on by default for new bots, small talk frequency is respected, and run counts are labelled correctly in charts
Dashboard, notifications and analytics
- No more being logged out after stepping away
- The unified dashboard works on mobile, and user names are editable
- Audit log filters load on first open, and invalid date ranges are rejected
- Name sorting is no longer case-sensitive, and column sorting is consistent
- Scrolling and pagination fixed across log tables, and duplicate SendGrid domains blocked
- Dashboard login fixed on iOS, and dashboard users work across regions
- Notification exports show which channel each one went out on, and unread counts stopped counting things you had already read
Apps, reliability and accessibility
Apps — Login stopped offering bots that cannot sign you in, first load is faster, and an iOS layout issue is fixed.
Reliability — A daily crash caused by oversized report attachments, fixed. A redirect loop that was taking parts of the platform down, fixed. Integrations reconnect on their own when a connection drops mid-session. Scheduled jobs stopped running twice, and small file uploads work again.
Accessibility — Another round of fixes across the dashboard.
👋 That's August
Anything here you want walked through, or a rollout you want help planning: your Leena SPOC is the person to ask. Implementation PM if you are still onboarding, CSM once you are live. Leena AI support is always there too.