Nexus Teams —
your AI team, starts from one sentence.
An IM-native AI staff platform. State the task once — the team lead splits it, dispatches specialists in parallel, reviews the quality, and returns finished work to your chat.
- Connected to Slack, Telegram and 6 more IM channels
- 30+ specialist AI staff to choose from
- Every action logged, auditable and stoppable
Get the Meridian renewal pack ready for Thursday — proposal, updated pricing, and a compliance read on their redlines.
One instruction in. Finished work out.
The difference between a single assistant and a team is the lead. It decides what the work actually is, who should do it, and whether the result is good enough to send back.
Say it once
Brief the team in your own words, in the chat tool you already use. No forms, no ticket templates, no workflow builder.
The lead splits the work
Your team lead breaks the request into subtasks, matches each one to the right specialist, and dispatches them in parallel.
Finished work comes back
Output is quality-reviewed by the lead before it reaches you, filed as an artifact, and delivered straight back into your thread.
From one sentence, to work you can ship as-is.
“We’re launching a whitening toothpaste bundle on Amazon US. All I have is a working product name. Build the listing for me.”
“On it. We’ll run keyword and competitor research first, then write five differentiated copy versions, with a compliance review alongside.”
Listing optimization pack, versions A to E
- Brand-story, benefit-led and value-bundle angles across 5 versions, ready for A/B testing
- All health claims rewritten into compliant wording, plus a keyword placement checklist
It tells you when something is going wrong.
The failure mode of an autonomous system is not that it stops — it is that it keeps going while quietly producing the wrong thing. The lead watches for that and raises it while there is still time to intervene.
- Flags a task that is drifting before it burns the budget
- Escalates anything outside the approved playbook to a human
- Surfaces the blocking decision instead of guessing at it
- Reports what it could not verify rather than filling the gap
Contract review
Clause 11.3 falls outside the approved library
Proposal rebuild
Delivered and filed to Team Shared
Pricing refresh
Margin holds at 61% on the current ladder
Pack assembly
Waiting on the clause decision above
The traps you have hit before, already routed around.
Most agent demos fall over on the second task. These three behaviours are what separate a demo from something you can put in front of a client deadline.
The lead orchestrates
Specialists run in parallel against one brief instead of queuing behind each other. A five-person pack does not take five times as long as a one-person pack.
Quality is checked before you see it
Output is graded against the bar set in the brief. Anything short of it is sent back for rework automatically, so what reaches you is finished rather than first-draft.
Work resumes from checkpoint
A failed step retries from its last good state rather than restarting the task. Long-running work survives transient errors without losing the hours already spent.
Let it run, money and authority stay in your hands.
Not security on a slide — security you can actually delegate behind. Every action is traceable, auditable and stoppable at any time.
Per-task budget
Every task carries a cap and stops the moment it is reached.
Per-staff quota
Daily and monthly usage tracked and enforced per staff member.
Total circuit breaker
Hit the threshold and the whole workspace pauses automatically.
Sensitive-action approval
Messages, payments and public posts all wait for your decision first.
Right in the chat you already live in.
No new tool to adopt and no tab to keep open. The team joins the channel your work already happens in and behaves like a colleague there.
Slack
Team channelsMention the lead in a channel to start work; slash commands and Workflow are supported, with progress and output posted back.
- Assign by @mention in channels
- Slash commands for quick calls
- Workflow automation support
- Files return to channels and Canvas
Microsoft Teams
Microsoft collaborationAssign work in channels or DMs, wired into Teams approvals, Planner tasks, and SharePoint documents.
- Assign in channels or DMs
- Connects to Teams approvals
- Tasks sync to Planner
- Files land in SharePoint
Assign work inside conversations and push progress through Business, useful for customer-facing teams.
- Assign inside conversations
- Business platform connected
- Progress updates in real time
- Reach external customers
Discord
Community collaborationAssign work inside server channels with slash commands, built for community ops and creator teams.
- Assign in server channels
- Slash-command launch
- Role permissions stay controlled
- Threaded collaboration support
Telegram
Instant messagingSend a message in private chat or groups to assign work; bots push progress in real time.
- Assign in DMs or groups
- Bot progress updates
- Inline action buttons
- Files ready where you chat
WeCom
WeComAssign work in groups or DMs; approvals run through WeCom's native flow, files land in WeCom Docs and Drive.
- Assign in groups or DMs
- Wired into WeCom approvals
- Files to WeCom Docs and Drive
- Works across internal contacts
DingTalk
DingTalkPushes messages via the work-notification bot, native to DingTalk approvals and docs, and connects to smart forms and process automation.
- Work-notification bot messaging
- Native approval-flow integration
- Smart forms and process automation
- Docs and drive delivery
Lark
LarkResults write straight into Base tables, edit docs alongside colleagues, group bots handle delivery, and Anycross workflows connect too.
- Results write into Base tables
- Collaborate in docs with colleagues
- Group bots handle delivery
- Anycross workflow support
12,000+ teams already hand the work to their AI team.
“In the two weeks before Black Friday, we added zero headcount — it carried every localized product page and promo. The lead sequenced it cleanly; we just reviewed output.”
“The scarcest thing in the Black Friday sprint is listings and ad creative — it covered several storefronts for me. I watch the big picture, the team handles the rest.”
“Multi-platform content that used to need 3 people in peak season now runs with 1 person + an AI team, at under a quarter of the cost.”
Through Black Friday, the AI team carried the whole content line.
An Amazon home brand replaced a roster of freelance writers with a single reviewer and an AI team, and turned multilingual content around overnight rather than over a fortnight.
30+ specialist AI teammates, across 7 role families.
From team leads to engineering, content, sales, AI and domain specialists, each one comes with built-in SOPs and skills. Pick the roles you need and start today.
Whatever the work is, there is a playbook for it.
5 use cases, 50+ industry plans and a dozen ready-made team templates. Start from one and change what does not fit — nothing here needs configuring from an empty page.
Indie developersIndie developers
One person, a whole team: docs, site copy, user research, launch promotion, all covered.
View plan
Content marketingContent marketing
Topics, writing, visuals, multi-platform distribution, data recap — end to end.
View plan
Automated sales loopAutomated sales loop
Find leads, draft scripts, follow-up reminders — AI runs the flow, you close the key deals.
View plan
Ops & growthOps & growth
Campaign planning, community ops, user segmentation, daily reports, on a rhythm.
View plan
Business researchBusiness research
Market breakdowns, competitor analysis, interview notes — a presentable report in half a day.
View plan
See all use casesSee all use cases
More use cases and detailed playbooks in the solutions center.
Go to centerYou might be wondering.
How is this different from ChatGPT, Claude or Gemini?
Those answer a prompt. This runs a piece of work: the lead decomposes the brief, dispatches specialists in parallel, grades what comes back against the bar you set, and sends anything short of it for rework. You get a deliverable rather than a transcript.
How is cost controlled? Could it overspend?
Four independent limits: a cap per task, a quota per specialist, a daily ceiling and a total circuit breaker for the workspace. Each stops work when it is reached rather than warning after the fact.
Is data safe? Could it change prices or speak externally without permission?
Sensitive actions — outbound messages, payments and anything published externally — are held for approval rather than executed. Every action is logged, attributable and reversible, and the log is exportable for audit.
Which platforms and store backends are supported?
Amazon Seller Central and DTC storefronts including Shopify, plus eight IM platforms — Slack, Telegram, Lark, DingTalk and WeCom live today, with Microsoft Teams, WhatsApp and Discord in progress.
Do I need to code or understand AI?
No. State your goal in chat the way you would with a colleague. There is nothing to install on an endpoint and no prompt syntax to learn.
How do I get started?
Create a team and give it one real piece of work. A card is required before live usage, and you can start from a playbook rather than an empty brief.
Describe the work. Meet the team that does it.
Build a team from your industry and objective in under a minute, or start from a single specialist and grow it.