帮助台与工单全新
用三面板控制台为你的客户提供支持，就像最好的帮助台那样：左侧是带计数器的已保存视图（未分配、我的、逾期）；中间是带搜索、排序和批量操作的队列；右侧是客户上下文栏，配有实时 SLA 计时器，以及一个工单 AI 面板，它会对主题进行分类、解读情绪、指出相似工单并推荐文章。收到的邮件会自动变成工单，基于事件和基于时间的自动化推动队列前进，路由把每个主题发送给合适的团队，宏一键回复，CSAT 在结束时衡量服务质量。事件会把工单归组到一个公开状态页下，你还可以把报表导出为 PDF 和 XLSX。所有这一切都与客户记录绑定，无需其他工具。
- 三面板控制台：带计数器的已保存视图、带搜索、排序和批量操作的队列，以及客户上下文栏。
- 每个工单的实时 SLA 计时器：计时只在工作时间内运行，并在违约前发出预警。
- 工单 AI 面板：在对话旁提供分类、情绪、相似工单和推荐文章。
- 带公开状态页的事件，以及导出为 PDF 和 XLSX 的报表。
- Parent and child cases: one ticket can hang off another, with the full tree inside the ticket itself and a warning before you close a parent that still has an open child.
- Rule-based auto-replies: channel, category, priority, business hours, first customer message only and keywords, evaluated in order, and the first match wins.
- Outcome code at close: a short list you control, scoped to tickets, phone calls or both, with an optional required explanation and a report by outcome.
- Quick text in the composer: type :: and the start of a shortcut to drop in a ready phrase, from the personal, team and company libraries.
- Ticket email templates per event, language, category and queue, with a preview that shows which template won before anything goes out.
- Work tabs in the console: several cases open at the same time, each with its own draft and its own sub-tab.
- Scheduled follow-ups and reminders written in plain words, with the screen echoing the time that was saved, not the words you typed.
- A swarm session on the ticket: who was called, in what role, to answer which question, by when, and what was concluded.
- A supervision wallboard for whoever coordinates: queues, people and channels on a single screen that refreshes itself and separates what is about to breach in the next hour from what already breached.
- AI that acts under a policy: off by default; category and priority each set to off, suggest or automatic, and automatic only writes above the confidence threshold you define; a reply draft never goes out on its own and resolving is never automatic.
在 Sellio 中免费，你只需为机器付费
带 SLA、控制台和 CSAT 的帮助台通常是一项单独的订阅。在 Sellio 中，它随 CRM 一起附带：你只为机器付费。
没有请求被遗忘
SLA 在营业时间内为每个工单计时，并在违约前提醒，因此没有任何工单会无人回复地搁置。
更快的首次响应
宏和 AI 建议的回复在几秒钟内完成首次接触，同时保持品牌语气。
坐席看到完整的客户全貌
交易、合同和过往对话都显示在工单旁边，无需打开另一个工具。
更少工单，更多自助服务
公开的帮助中心自行解决常见问题，减轻团队的队列压力。
在与 CRM 同一个系统中提供支持：坐席无需切换标签页就能看到交易、合同和客户历史。
基于事件和基于时间的自动化，加上自动路由，让队列自行运转。
AI 建议的回复加快首次响应，同时不失你的品牌语气。
公开的帮助中心在常见问题变成工单之前就予以解答。
一个能自我打理的队列
每个工单都带有一个 SLA，只在营业时间内计时，并在违约前发出预警。基于事件和基于时间的自动化会重新分配、排定优先级并升级，而路由把每个主题发送给合适的组。坐席专注于解决问题，而不是整理队列。
一屏尽览全部上下文
在控制台中，客户对话与团队的内部备注以及他们在 CRM 中的完整历史并排显示。宏一键应用回复和操作，在结束时，CSAT 调查衡量服务表现如何。
三面板控制台
左侧是带计数器的已保存视图，让你一眼就知道有多少工单未分配、逾期或属于你。中间是带搜索、排序和批量操作的队列，让你一次重新分配或关闭多个工单。右侧是上下文栏，汇集了客户历史、实时 SLA 计时器和工单 AI 面板，全都无需离开屏幕。
工单 AI 与带公开状态页的事件
AI 面板对主题进行分类、解读情绪、显示相似工单并推荐知识库文章，让坐席回复得更快、更有上下文。当更大的问题发生时，工单会归组为一个带公开状态页的事件，你还可以把报表导出为 PDF 和 XLSX，分享给任何需要的人。
Parent and child cases
A ticket can hang off another one. Point it at the parent and the children show up in a list inside the ticket itself, with the whole tree in view. The guards are real: the system refuses a loop, where A is the parent of B which is the parent of A, refuses a chain that runs too deep, and warns you when you try to close a parent that still has an open child. It fits the incident that spawns several tickets: you work the parent and see everything that depends on it.
Auto-replies became rules
It used to be one fixed text for every ticket that arrived. Now they are rules evaluated in order, and the first match wins: conditions on channel, category, priority, business hours, first customer message only and keywords, each with its own text. There is also the silent rule, which matches and deliberately says nothing, for the subject where any automatic reply gets in the way. The loop guard cannot be turned off: replying to an automated mailbox is two answering machines talking to each other forever. And anyone who never created a rule stays exactly as they were, with the single text they always had.
Outcome codes at close
Resolving a ticket used to mean free text in the Resolution field. That describes what happened, but it does not classify: two agents tell the same outcome ten different ways, and nobody can answer how many cases this month were a product defect and how many were a customer configuration mistake. Category does not solve it, because category classifies the subject at intake, not the outcome at the end. Now the outcome is a short list you control, in Settings. Each code has a label, a scope (ticket, phone call, or both, so the same vocabulary serves both channels instead of two lists saying the same thing) and a switch for requiring an explanation, for codes like "Other", which on its own explains nothing. There is a report by outcome, and the requirement to fill it in can be turned on. Three decisions matter. Turning the requirement on before any code exists blocks nothing, otherwise every ticket in the company would stop closing the day someone ticked the box, with nothing to pick. The requirement is enforced by the server, not by the screen, so closing from the list, from a macro, from the API or from the terminal goes through the same check. And a code already in use cannot be deleted, only deactivated: deleting it would take with it the meaning of every ticket closed under it, and the old reports would quietly start lying.
Quick text behind a typed shortcut
The only ready-made text in the console was the macro, and a macro also applies a status and a priority: dropping in a phrase changed the ticket without anyone asking for it. Quick text writes nothing but the text, which is why it can be light and personal. In the composer, type :: and the start of a shortcut to narrow the list, or :: on its own to see the whole library. There are three libraries, personal (only you see it), team and company, and the one closest to you wins, so you can keep your own version of ::hi without changing the house standard. A shortcut has no spaces and no colon and fits in 40 characters; the search ignores case and accents and looks at the shortcut, then the title, then the folder, never the body; and the order is always the same, so you never pick the wrong phrase from muscle memory. The phrases accept the same variables as macros, filled in at send time. Whoever administers can generate a starter company library, and running it again duplicates nothing and never overwrites what you rewrote.
The ticket email in the reader’s language
Five moments of the ticket can carry a text of your own: opened, updated, resolved, survey and closure. Each template is written in one language and can be restricted to a category, to a queue, or to both. The choice is declared and it runs in this order: the system first decides what to say, visiting the families from the most specific to the most generic (queue and category, then queue, then category, then the generic one), and only then in which language to say it, with the exact language first, then the same base language, then the reserve language. The language that counts is the requester’s, the person who reads it, not the agent’s. A family with none of those three languages is skipped: sending the generic text in the reader’s language beats sending the right text in a language they do not read. The preview shows which template won and in which language before anything goes out. And if you register nothing, nothing changes: with no template the ticket behaves exactly as it does today, so you can adopt this one queue at a time.
Several cases open, no draft lost
The console used to open one case at a time, and picking another replaced the previous one: whoever works in volume keeps three or four cases going and lost the draft at every switch. Now there is a tab bar above the conversation. A single click opens a preview tab, in italics, and the next preview takes its place, so browsing a queue of twenty cases does not open twenty tabs; writing in it or pinning it makes the tab permanent. The same case never opens twice, a blue dot marks a written and unsent reply, and each tab remembers the sub-tab you were on. Up to eight tabs: above that the oldest tab with no draft is dropped, and if every tab has a draft the limit gives way, because losing written work is worse than a long bar. Close saved takes the tabs with no draft and keeps the ones that have one. The tabs live in your own browser, on this device.
When the case does not end today
Two ways out, and neither is a note on the side. A scheduled follow-up is a task about the ticket: it can go to someone else and it shows in the day’s list like any other task; a reminder is always yours. You write when in your own words, "tomorrow 9am", "in 2 hours", "friday", in the ten languages of the product: a day with no time becomes 9 a.m. in the organization’s time zone, a duration keeps the current time of day, a time that has already passed goes to the next day, a day in the past is refused instead of guessed, and nothing goes further than 365 days. The screen echoes the time that was saved, not the words you typed, and says so when it assumed 9 a.m. or pushed the deadline to tomorrow. The swarm session, in turn, is the record of calling people in: who joined, in what role, to answer which question and by when. There is one open session per ticket; reporter and scribe are single seats, expert has no limit, and whoever is called gets a notice carrying the reason and starts following the ticket. Closing requires saying what was concluded, and that outcome goes back to the ticket as an internal note, so the next agent reads it without opening anything else. The swarm records the session, it does not open a Slack channel and does not start a video call: the conversation keeps happening where it already happens.
The supervision wallboard, made to live on a TV
It is the same operation seen by whoever coordinates, not by whoever answers. One screen, refreshing itself every 20 seconds, with big numbers and a traffic light, so it reads from five meters away. At the top, the headline: open, unassigned, waiting for a reply, what already breached, what is about to breach, and how many people are actually available. Below it, the queues ordered by what hurts first, the people with whoever is drowning at the top, the channels, and the trail of the latest assignments. Two readings the product could not give before. The first is the difference between "about to breach in the next hour" and "breached": a board that only shows the fire arrives late, so what can still be saved shows up in amber, and the window is yours, from 30 minutes to 4 hours. The second is the queue stranded by a skill: when a queue requires a skill nobody on the roster has, routing leaves the item unassigned on purpose, and from the outside "nobody has picked it up yet" and "nobody can pick it up" look identical. The board turns that queue red, because it is configuration waiting for people, not a system failure, and only whoever coordinates can fix it. The board is read-only: it points, it does not reassign. It belongs to whoever administers, because it shows every person’s workload. And when the deadline math looked at a sample instead of the whole, it says so on the screen instead of pretending completeness.
AI that acts, inside a policy you write
The AI panel always suggested; now it can write to the ticket, and the dial for how much it writes is yours, in Settings. Everything starts off. Once on, each of the two fields the AI can reach, category and priority, has its own mode: off, suggest or automatic. Only those two: resolve, close and reply are not on the list, and no mode includes them. In automatic, the AI only writes when confidence clears the threshold you set (80% by default), when the current value was filled by the form and not by a person (unless you allow overwriting) and while the daily cap on autonomous actions has not been reached; otherwise it goes back to suggesting, with the reason written next to it: low confidence, human value kept, cap reached. The cap is shared with the Copilot, so how much the AI has already acted today is a single count, and when it hits, the policy screen and the ticket say so instead of stopping silently. The proposed category has to be one of the options in your field; anything that matches none is discarded, never written. A reply draft is always a suggestion: accepting it puts the text in the composer, and sending remains your click. Every decision, act or suggest, is recorded on the ticket and in the audit trail, and you can ask it to classify now by hand, with the same policy in charge.
- 1
连接客服邮箱：收到的消息在队列中变成工单。
- 2
设置 SLA、营业时间、组和路由规则。
- 3
在控制台中处理，配有对话、内部备注、宏和 AI 建议的回复。
- 4
关闭工单，收集 CSAT，并在帮助中心发布常见答案。
晚上 10 点的客户邮件
该消息会立即变成工单，路由把它发送给合适的组，SLA 从下一个工作时间开始计时。
每周重复的一个问题
一个宏一键回复并关闭；如果它非常常见，帮助中心的文章会在工单创建之前就把它解决。
一位急切且不满的客户
一项基于时间的自动化对该案例排定优先级并升级，控制台显示完整历史，让坐席带着上下文回复。
The customer asked you to come back after month-end
You write "friday 9am" on the ticket, the follow-up becomes a task linked to it, and the screen confirms the time that was actually saved.
The case stalls and needs a specialist
You open a swarm session with the reason and the deadline, call the people you need, and what was concluded comes back to the ticket as an internal note.
我需要另一个帮助台工具吗？
不需要。客服运行在与 CRM 同一个系统中，坐席无需切换标签页就能看到交易、合同和客户历史。
SLA 在营业时间之外也计时吗？
不，SLA 计时器只在工作时间内运行，并在违约前发出预警。
AI 会替我回复吗？
它不会取代坐席；AI 面板对主题进行分类、解读情绪、显示相似工单并推荐文章，以加快回复。
三面板控制台是如何运作的？
左侧是带计数器的已保存视图，中间是带搜索和批量操作的队列，右侧是带 SLA 和 AI 的客户上下文。
我可以处理更大的事件并导出报表吗？
可以，相关工单会归组为一个带公开状态页的事件，你还可以把报表导出为 PDF 和 XLSX。
客户邮件如何自动变成工单？
收到的消息会自动变成工单，路由把主题发送给合适的团队，无需手动操作。
基于事件和基于时间的自动化有什么作用？
它们会自行重新分配、排定优先级并升级工单，让队列保持运转，而无需坐席亲自整理。
有客户满意度调查吗？
有，工单一关闭，CSAT 就衡量服务表现如何。
这能减少进来的工单量吗？
能，公开的帮助中心在常见问题变成工单之前就予以解答，减轻队列压力。
What is the difference between quick text and a macro?
A macro also applies a status and a priority; quick text only drops in the phrase, without touching the ticket, and it sits behind a shortcut you type after ::.
Do I have to register email templates for support to work?
No. With no template registered nothing changes, so you can adopt it one queue at a time.
Can an agent work several tickets at the same time?
Yes, the console has work tabs: up to eight cases open, each with its own draft and its own sub-tab, and Close saved never carries a written reply away with it.