Solaris · management-layer agents

An AI manager that lives in your working chats

One agent, two roles. In your team room it assigns tasks, reports on the CRM and remembers every commitment. In each client room it is a project manager: it knows the context, brings new people up to speed, answers to the point — and never says a word it shouldn't.

Running inside Solaris right now — one team room and 30+ client rooms
AI Manager · 9:00 AM
Morning briefing
New leads in the last 24 hours: 14, all logged in the CRM
Waiting on a reply to the client: 3 threads over four hours old
Deal #4820 has not moved in four days
You promised the client documents yesterday — still not sent

This exact briefing lands in our own team room every morning.

A familiar picture

The chats grow, and the operations eat your day

Every client is a separate room. Every room brings questions, promises and "when will it be ready?". All of it is held in one person's head — usually yours.

Answers arrive late

The client asked in the morning, the team saw it in the evening. Nothing broke — but the relationship did, a little.

Promises never become tasks

Commitments made in a chat stay in the chat. Nobody transfers them anywhere, so they quietly go missing.

Someone says too much

An employee in a client room quotes a price, a date, or explains how you actually work internally. You find out later.

You only know what you dig out

What is happening with leads and deals is visible only if you open the CRM yourself, every day, on purpose.

One agent, two roles

The same agent lives in all of your rooms but behaves differently in each. With your own people it is a manager with the full business context. With clients it represents the company — helpful, precise and bounded.

How it behaves

Team room and client room

Team room

Manager for your team

· Every morning: a CRM report — new leads for the day, week and month, what is untouched, who to call today, which deals are stuck.
· Turns conversations into tasks in your tracker — on command or on its own initiative.
· Remembers commitments and reminds people about them before they expire.
· Answers the team's questions about projects and process with the full business context.

Client room

Project manager for each client

· Added to a group, it introduces itself and starts working. Each client gets its own project context.
· Replies quickly and strictly on the substance of that project. Brings new participants up to speed.
· Never decides commercial questions itself — it hands them to the team and logs a task.
· Records agreements, so no promise made to a client is lost.

The obvious fear, handled

Your commercial information stays yours

An ordinary chatbot can blurt out anything. Here there are two layers: hard role rules, plus a software filter that intercepts a reply if it contains figures that are not part of the agreed project context.

It will never tell a client

· Prices, rates, discounts, budgets, unit economics
· Dates and deadlines the team has not confirmed
· Your methodology, tooling and internal systems
· Anything about your other clients

What it does instead

· Accepts the question politely and passes it to your team
· Automatically creates a task: "client is asking — reply needed"
· At the slightest doubt it hands over to a human rather than inventing an answer
· If the filter catches something, your team gets the case for manual review

[guard] blocked a reply containing "$4,500" in a client room → replaced with a handover to the team, task created
What it can do

Not a bot with buttons — an employee with context

Underneath are the same models we deploy for enterprise clients. It reads a chat like a person and answers like a person who is actually across the project.

CRM reporting

Every morning: leads for the day, week and month, what is unprocessed, who to call, which deals have stalled.

Tasks from conversation

A commitment is spoken, a task appears in the tracker — whether it came from your team or from a client request.

Answers with context

Each client has its own project file. The agent answers on substance instead of talking in generalities.

Onboarding newcomers

Someone new joins the room and the agent greets them and explains what the project is, who is on it and where things stand.

Commitment memory

It remembers who promised what and by when, and which questions were left unanswered. Every morning the team gets the list of what is overdue and what is due today.

Escalation

Anything genuinely complex goes straight to the right person, along with a task so it cannot be forgotten.

Power mode

Ask in a chat, get a finished file back

In the team room or in a direct message, describe what you need. The agent opens its tools, does the work and sends back the finished artefact — not an explanation, not an instruction, the file itself. Up to 25 minutes on a heavy task; voice messages are transcribed automatically.

Documents · Word · PDF

Contracts and NDAs, acceptance certificates, proposals, offers, briefs, statements of work, internal policies and templates.

Spreadsheets · Excel · CSV

Cost estimates, price lists, budget models, data summaries, tables with formulas and charts, ad-hoc analysis.

Content · posts and email

Posts for any channel, publication series, email campaigns, sales scripts, articles and pitch decks.

Web · landing pages

A landing page or an HTML proposal built and deployed to your site directly, plus slide decks in PPTX.

Images

Original illustrations for the task at hand, post covers, banners, brand visuals, in any aspect ratio you need.

Research

Company and counterparty checks, competitive analysis, niche and audience research — delivered as a written report with conclusions.

How it is implemented

Three steps, and it is working

01

We tune it to your business

We load the context: team, projects, policies, your task tracker and CRM. Together we decide what may be said to clients and what stays internal.

02

You add it to your chats

You add the agent to a group like any other participant. It introduces itself and picks up the context. A new client means a new room and zero setup.

03

It runs the project

Morning reports, tasks from conversations, answers to clients, commitments tracked. You step in only where a human decision is actually required.

Pricing

Less than one junior manager, working in every room at once

We start with a paid pilot so you can judge the agent on your own chats before committing to an implementation.

Pilot
$900
2 weeks · one-off
  • One team room and one client room
  • Context and rules configured
  • Tasks pushed into your tracker
  • Pilot results and a scale-up plan
Most chosen
Implementation
from $5,000
one-off · then $1,000 / month
  • Team room plus your client rooms
  • Daily CRM reporting
  • Confidentiality guard, both layers
  • New rooms connect themselves
  • Support, monitoring and retraining
Enterprise contour
from $12,000
project plus dedicated support
  • Unlimited rooms, hosted on your own server
  • Integrations: your CRM, trackers, calendars
  • Trained on your internal policies
  • SLA and a named support engineer

Support is only proposed once the contour is live and useful. Discovery is free.

Questions

What people ask about the AI manager

How is this different from a normal chatbot?

A chatbot answers from scripts and buttons. This is an agent on the same models we deploy for enterprise clients: it reads the whole conversation, knows the context of that specific client and of your team, and decides for itself where to answer and where to hand over to a person. And it has hard boundaries — commercial information never leaves the internal side.

What if it tells a client something it shouldn't?

There are two layers. The first is the role itself: prices, deadlines, methodology and internal process are forbidden at the instruction level. The second is a software filter — if a reply meant for a client contains figures that are not in the agreed project context, the reply is automatically replaced with a handover to your team, and a task is created for you.

How long does implementation take?

A pilot goes live within days: we gather the business context, connect your tracker and CRM and add the agent to the chats. After that every new client room connects itself — you simply add the agent to the group.

What does it work with?

Chat is the primary environment — Telegram today, other messengers on request. Tasks go to your tracker, reports come from your CRM. In the enterprise contour we connect your systems by agreement: CRM, trackers, calendars, knowledge bases.

Does it replace a project manager?

No — it takes the routine off one. The agent closes the operational layer: fast answers, recorded commitments, reports and follow-up. Decisions, negotiation and the relationship stay with people; the agent makes sure only the things that genuinely need a human reach one.

See it running before you decide

On a short call we show the live contour — the morning briefing, the commitment log, the incident history — and tell you honestly whether your company is ready for it.