Lexer's support desk. Describe a client's issue, get a researched resolution, and raise it to engineering when it needs to go further.
Sign in with your Lexer account to continue.
The agents, lookups and model calls behind this reply
Lumen answers support questions from learn.lexer.io, and researches the ones that need it across historical tickets, our FAQ, Slack and the Asana support board — then hands back a resolution with clear next steps.
The page is now a casebook on the left and the case you're working on the right. Cached answers and the Escalated board live in the casebook; the conversation reads as messages back and forth — yours on the right, Lumen's on the left — with a researched answer marked Resolution as a ruled sheet on Lumen's side. Nothing has moved server-side — every button does what it did.
Just describe your issue and keep talking. There's no separate intake step any more — Lumen asks a clarifying question only when it actually needs one, and everything you've said stays in the conversation.
How-to questions are answered straight away — no research run, no waiting. A reported fault triggers the full search, which takes a minute or two and shows its progress.
Each conversation researches one issue. For a different problem, start a New conversation — which is also what replaced Clear history, since the conversation is now what Lumen remembers.
Create Support Ticket no longer needs a completed search. If it's clear from the conversation that the issue needs engineering, escalate straight away — the ticket carries the conversation itself, plus a hub and a priority worked out from it.
Priority on this path tops out at P2: without research findings there's no evidence of multi-client impact, and a human can raise it on the board.
Use the paperclip to attach screenshots, PDFs or spreadsheets (.xlsx, .csv) on any message. Up to 5 files per message, 4 MB total, and 20 pages per PDF. Anything you attach is carried through to the search and included if the issue is escalated.
Large spreadsheets are read as text and truncated — Lumen sees roughly the first 30,000 characters. It's told when a file was cut off, but you aren't shown a notice, so treat answers about a big export as based on the top of it.
Files stay in context for follow-up questions — images, PDFs and spreadsheets alike. Lumen keeps the 3 most recent files from earlier messages within the 4 MB budget, and if it has to leave one out it says which, so you know to re-attach it.
New: attachments used to be images only. PDFs and spreadsheets now work everywhere images did — on any message in the conversation, and on a Post comment to the Asana ticket. If a file can't be read (a password-protected PDF, a corrupt workbook) Lumen rejects it and says why rather than quietly answering without it.
Once an issue has been escalated, the chat bar gains a second button: Post comment. It sends your message and any attachments straight to the Asana ticket instead of asking Lumen — use it to explain what a screenshot shows, or to add anything the engineer picking it up will need.
Each comment appears in the thread as a dashed bubble with a delivery status — Posted to LMN-###, or a failure with a Retry link — so you always know whether it actually landed.
Changed: attaching an image to a normal chat message no longer posts it to the ticket. Send shows it to Lumen and keeps it for escalation; once the ticket exists, Post comment is the only route to Asana — so there's never any doubt about what the ticket received.
Check replies opens the ticket's comments in a window — the engineer's exact wording, with who wrote each one and when. The whole thread is shown oldest first, so it reads top to bottom, and the window opens scrolled to the newest comment. Press it as often as you like: it shows the thread as it stands rather than adding anything to the conversation.
The Escalated board now shows each ticket's priority and due date as they stand in Asana, plus a New badge on your own tickets when there's activity you haven't seen. Opening the ticket clears it.
The same signal now appears in the Casebook: an archived answer whose ticket has a reply you haven't read carries a ● New reply dot.
Clicking either indicator — the dot in the Casebook or the New badge on the board — opens that same window. Being shown the comments is what marks them read, so you never clear the signal without seeing what changed.
Both are Asana's to set — Lumen reads them and never writes them, so what you see is what the board says.
The row of hub buttons above the Casebook is now two dropdowns: one for the hub, and a new one for the client. Both filter the Escalated board as well, and your selection survives the trip between the two views — useful when you're picking up where someone left off on one account.
The client comes from the Asana field chosen when the ticket was raised, so only escalated answers have one: choosing a client hides everything that never became a ticket. On the board, a ticket whose cached answer has since been cleared shows up under Uncategorised.
Client response used to write into a panel of its own and could only be used once. It now lands in the conversation as a normal reply — so it survives a reload, and you can ask Lumen to adjust the wording, or press the button again for a fresh rewrite after the conversation has moved on.
Every ticket Lumen raises now opens with an investigation brief — the issue in one line, what's already been checked, what's still open, and what a fix should look like — and is labelled Investigate or Fix.
Steps you can run yourself stay in the answer; anything needing data or backend access is set aside under Notes for engineer, word for word. Nothing is asked of you first — the brief is built from the conversation when you escalate.
Escalating now posts the SLA targets for the ticket's priority straight into the conversation — a first response date and a resolution date, so you don't have to go and look them up.
Resolution targets count business days (weekends skipped; public holidays are not, so a holiday counts as a working day), and P0/P1 first responses are clock time — an hour promised at 4:40pm means 5:40pm. Times are Sydney time with the zone named. They're an estimate, not a commitment.
An informational escalation has no priority, so it gets no dates rather than an invented one.
Just ask — "where's this up to?" resolves the ticket this conversation raised, and you can ask for all your tickets to get them listed by LMN id with their title, when you raised them, and where they've got to. You never need to know the LMN number.
That works from a reopened archived answer too, so a conversation that never ran a search can still be asked about its own ticket.
Tickets that have been removed from Asana are left out of the list but counted and named, so a missing ticket never looks like one you never raised.
A researched answer ends with a Sources section — one line per source actually used, with the real link: the Asana ticket (and its state, so a lead under investigation can't read as a fix), the Slack thread, the learn.lexer.io article. Sources with no link say so rather than showing a made-up URL.
Lumen's own shorter answers do the same, citing anything it looked up on that turn. A Client response keeps only the learn.lexer.io links, as Related reading — internal sources are stripped.
A dot beside Sync KB now reports the state of the Asana index — ● synced 2h ago, with the ticket count and exact time on hover. Green is fresh, amber means it's getting old, and red means the board has never been indexed — in which case resolved tickets aren't being searched at all, and pressing Sync KB fixes it.
Cache response is now a one-off. Once an answer is cached the button becomes Cached ✓ — auto-saving, and everything said or attached from there on is saved onto that same entry as the conversation develops — so what you reopen later is the whole conversation, with its files, not just the answer as it stood when you pressed the button.
There's no second press because there's nothing left to do: the old re-cache overwrote the archived answer and dropped the client response stored with it.
Reopening an archived answer — yours or a colleague's — shows Cached ✓ — auto-saving too: your follow-ups and attachments are saved back onto that shared entry, so whoever opens it next sees them.