Agents that do the work. You make the calls.
“What can AI take off our plate — without taking the decision?”
Dot isn’t a chatbot. It’s Dotwork’s AI, built into every experience — from a highlighted sentence to a whole plan, from the page you’re on to the Slack thread and the AI you already use. It works across your whole operating model and runs the skills your teams define: collecting updates, explaining variances, drafting readouts, checking what changed. It does the legwork, drafts the result and routes every decision to the person who owns it.
Dot · skill run
Explain the variance- ✓Read the October forecast and actuals
- ✓Traced the gap to three initiatives
- ✓Checked the strategy each one serves
- ✓Drafted the explanation and a reforecast
Needs review
Move $0.4M from Data Platform to Customer Onboarding for Q4
Routed to the finance owner · keeps both bets inside their guardrails
Illustrative view · sample data
The pressure
Most AI agents stall at the same place: they don’t know why the work exists.
An agent that can read tickets can close tickets. It can’t tell you whether the work is funded, which strategy it serves, what it costs or who has to approve the change. So it either asks a person for everything, or guesses — and nobody can see which.
What it looks like today
- Agents that summarize activity but can’t weigh a tradeoff
- AI drafts that ignore the budget and the strategy behind the work
- Automations that act without a clear owner for the decision
- No way to see what the AI did, for whom, or what it cost
What changes
What changes.
A chatbot in a corner of one tool
AI built into every page, record, cell and conversation — across the whole operating model
Prompts every person rewrites from scratch
Skills: the jobs your teams repeat, written down once and run on request or on a schedule
Automation that makes the call
Drafts and recommendations routed to the person who owns the decision
AI spend nobody can see
Every skill run metered, so you know what each job costs
Not a chatbot
Dot isn’t a chat window. It’s in every experience.
Most enterprise AI waits in a chat box for someone to remember it’s there. Dot is built into Dotwork at every level of interaction — so help shows up where the work is, at the size of the question.
- 01Any text field
A sentence
Highlight text in a description or update and ask Dot about it, in place.
- 02Planning grids
A number
Select a cell — or several — and ask Dot why it moved and what’s behind it.
- 03Every item
A record
Dot drafts the status update from what actually changed, and proposes edits for your approval.
- 04Every view
A page
Open Dot beside any page and it reads what’s on screen — the cards, rows and figures you’re looking at.
- 05Across Dotwork
The model
Questions that follow the relationships from strategy to spend, and skills that run the jobs your teams repeat.
- 06AI Studios
The platform
Model the business, connect systems and build skills by describing them to Dot.
Learn more - 07Slack and Teams
Your team
Dot asks for updates, follows up and carries on the conversation in the thread.
- 08Claude, ChatGPT, Copilot
Any AI
The same model and the same permissions, from the assistant your people already use.
Learn more
One AI at every level — the same model, the same permissions and the same approvals behind each one.
Agents, skills and you
Dot knows the job. Skills know the method. You own the call.
The agent
Dot knows the job.
Dot is one agent that works across your whole operating model. It reads your ontology and your permissions before it starts, follows relationships and history to the answer, and splits big jobs across parallel sub-agents that report back.
- Grounded in your model on every request
- Sees only what the person can see
- Parallel sub-agents for big jobs
The skills
Skills know the method.
A skill is a playbook for a job your teams repeat — the method, the questions it answers, the readouts it produces and the signals it watches. Run it when you need it, or put it on a schedule.
- Written once, run by anyone allowed
- On request or on a schedule
- Build one by describing it to Dot
The call
You own the decision.
Dot drafts first. Changes that matter come to you as an approval card, or go to the inbox of the person who owns the decision — and run with that person’s permissions when they approve.
- Draft, then confirm
- Approvals routed to the owner’s inbox
- Approved work runs as the approver
Grounded in how you run
Before Dot does anything, it reads the operating model.
Other agents start from the activity in front of them. Dot starts from why the work exists — so it can weigh a tradeoff, not just summarize it.
Your ontology
The kinds of records your company runs on and how they relate, read before every request.
Your permissions
What the person asking can see and change, so Dot never reaches past them.
Relationships
From a key result to the initiatives behind it, the budget that funds them and the apps they run on.
History and trends
How every field has moved over time, so Dot can tell drift from noise.
Memory you write
Context your teams attach to a site, a space, a kind of record or a skill — which can never widen anyone’s access.
The page you’re on
Dot starts from the view, room or record in front of you, not a blank prompt.
How a skill runs
Dot does the legwork. The decision waits for you.
- Step 1
Ask or schedule
Run a skill from chat or a button, or let it run on its schedule.
- Step 2
Read the model
Dot pulls the records, relationships and history the job needs.
- Step 3
Work the job
Big jobs are split across sub-agents working in parallel.
- Step 4
Draft the result
A readout, a status update, field changes or a message — as a draft.
- Step 5
You decide
Approve, edit or decline. Changes go through only when someone with the right to make them says so.
What Dot does
The work nobody wants, done before the meeting.
Collect updates by conversation
Dot asks owners for status in chat or Slack, follows up on what’s missing and writes the answers back to the model.
Draft status updates
Start an update and Dot drafts it from what actually changed on the record.
Write readouts
Reports drafted from the model for a review, a steering meeting or the board.
Spot what’s moving
Signals ranked from the trends in your own data, so the outliers come to you.
Summarize what you’re looking at
Summaries of the cards and rows on screen, with the records behind them.
Change the model, with your approval
Update fields, status, comments and relationships — proposed first, applied when approved.
Skills across Dotwork
A skill for every job your teams repeat.
Each solution comes with the jobs Dot takes on in that area. These are a few — and your teams can write their own.
Strategy
- Strategy snapshot
- Assumption mining
- Red-teaming a bet
- Review composition
Money
- Variance explanation
- Cost mapping suggestions
- Contract spreading
- Readouts for close
Portfolio and delivery
- Flow review
- Dependency checks
- Status drafting
- Benefit tracking
Architecture
- Cost linking
- End-of-life review
- Blast radius
- Sprawl finding
Controls and AI
- Policy tracing
- Waiver review
- AI intake
- AI value audit
Communication
- Announcement drafting
- Formal status
- Audience tailoring
- Feedback routing
Build your own
Describe the job. Dot writes the skill.
Tell Dot what you want done — the question, the method, what good looks like. Dot drafts the skill, you refine it in the editor, attach the context it should always know, and decide whether it runs on request or on a schedule.
You
Every Monday, find initiatives with no update in two weeks, ask their owners for status in Slack, and give me a summary of what changed.
Dot · draft skill
- Purpose
- Keep initiative status current
- Method
- Find stale updates → ask owners → summarize
- Produces
- A weekly status readout
- Schedule
- Mondays
Illustrative view
Governed by design
Useful AI you can still account for.
Dot runs inside the same permissions, approvals and budgets as the rest of Dotwork — and you can see what every skill run cost.
Acts as a person, never as a superuser
Dot works with the permissions of the person asking. Scheduled skills run as the person who scheduled them.
Approvals where you want them
Turn on approval for changes, or let a person flag any request for review. Approvals can go to someone else’s inbox.
Every skill run metered
Token and dollar spend tracked by skill, on an AI usage page your admins can see.
Your choice of model
Each site sets its default model from the families Dotwork supports, and whether people may choose their own.
Bounded runs
Every run has limits on steps, time and tool calls, with rate limits on each tool.
Open by design
The same model and permissions sit behind Dot and behind any AI your people connect through MCP.
How it compares
Most agents know the activity. Dot knows the operating model.
Copilots in work tools act on tickets, documents and messages. Agent platforms give you a builder and leave the context to you. Dot starts from how the company runs — strategy, money, portfolio, architecture and controls — and keeps people on the decisions.
| Capability | Copilots in work tools | Agent-building platforms | Dot |
|---|---|---|---|
| Works within the person’s permissions | Partly | ||
| Grounded in strategy, money, portfolio and controls as one model | |||
| Reasons over relationships and history, not just documents | Partly | Partly | |
| Reusable skills, run on request or on a schedule | Partly | ||
| Build a skill by describing it | Partly | Partly | |
| Approvals routed to the person who owns the decision | Partly | Partly | |
| Spend metered for every skill run | Partly | ||
| Same model open to any AI through MCP | Partly | Partly |
The model underneath
Every module makes the next one smarter.
Steer, Fund & Optimize and Govern & Deliver all read and write the same Enterprise Operating Model. The first module you license builds the model. Every module after it adds context instead of another silo — and Dotwork AI reads all of it, stopping where the decision is yours.
See the Enterprise Operating ModelBe Decisive.Build Fast.Connect Everything.
Start ConnectingSolutions
Platform
Resources
© 2025-2026 Dotwork, Inc. All rights reserved.