A complete walkthrough of every part of OrgTP, written for someone using it for the first time. No prior experience with any operating system for business is assumed.
No sections match.
Try a different word, or raise a ticket and a human will help.
OrgTP is where your people and your AI agents run the company together, out of one shared memory.
Two halves, and either one works on its own:
The thing worth understanding early: everything you do in a meeting becomes memory. Decisions, numbers, who owns what, what was tried and did not work. Ollie reads it back to you later, and so does every agent you connect. Most tools forget the meeting the moment it ends. This one is built so the meeting is the input.
| Thing | Plain meaning |
|---|---|
| Rock (also called a Quarterly Priority) | A big goal you commit to finishing in the next 90 days or so. |
| KPI / Scorecard | The handful of numbers that tell you the business is healthy, read weekly. |
| To-Do | A small action item, due within about 7 days, owned by a named person. |
| Issue | Anything that needs working through as a team: identify it, talk it through, decide what happens. |
| Headline | A short piece of customer or employee news, good or bad, shared in the meeting. |
Where to go next: if you are setting the company up, read section 3. If somebody else set you up and you just want to work, read sections 4 and 5.
Go to orgtp.com, click Sign In, and choose Continue with Google or use your email and password.
The first time you create an organization, OrgTP walks you through seven short steps. Every step after the first has a Skip for now link and a Skip, take me in escape straight to the dashboard, so you can do the bare minimum and fill the rest in later. Nothing here is permanent.
| Step | What it asks | Why it matters |
|---|---|---|
| 1. Your profile | Company name, industry, team size, and your own name and role. | Submitting this step is what brings your organization into existence and claims your seat. |
| 2. Your team | Coworkers by name, role, and email. | Each one gets an invite and appears on the chart marked "Invite sent", so the chart is real before anyone logs in. |
| 3. Your goals | One or two big outcomes, with a timeframe and an owner. | These become your Rocks. One owner each, always. |
| 4. Your KPIs | The numbers each seat watches, with an optional target and a cadence. | This is the start of the Scorecard you read every week. |
| 5. Your agents | Your AI tools: a name, what it does, its runtime, and a human owner. | The step no other operating tool has. An agent with a seat and an owner is accountable the way a person is. |
| 6. Your teams | Groups of seats, such as Leadership or Web Team. | Each team gets its own meeting and its own scoreboard. |
| 7. Your first meeting | A recap, plus a preview of your weekly meeting and its agenda. | Start the meeting right there, or go explore the chart first. |
Coming from another tool? Do not retype any of it. Section 17 imports your Rocks, to-dos, issues, headlines and scorecard, and the roster comes with them.
Where: /dashboard, the screen you land on after signing in.
This is your cockpit, not the company's. It greets you by name and stacks what you personally owe and own into panels you can rearrange:
Rearrange it. Tiles can be moved and hidden, and Reset to default puts them back. Your layout is yours and does not affect anybody else.
Controls: switch organization or team at the top, toggle "show only mine", and open any meeting or to-do directly.
Where: /l8 for the list. The meeting itself opens from there.
The Delta Meeting is the weekly leadership meeting, and it is where the whole system pays off: one timed screen the team runs the meeting from, so it starts and ends on time and nothing quietly falls through. Delta, because the meeting is about what changed.
Past and upcoming meetings, each with a status badge. Use New Meeting to create one (title, date and time, type, recurrence, team), then Create + Open. If you are on more than one team, a team filter appears. The ? explains the meeting types: Weekly Leadership, Departmental, Quarterly, Annual, and One-on-One.
Everything lives on one page: title and time, Start and End controls, an add-to-calendar panel, a "Driving toward" strip of longer-term targets, the attendee list, an optional Agent Reports panel summarising what your AI seats did, and a sticky agenda bar with a per-section timer.
Click Start meeting. That freezes a snapshot of your scorecard and Rocks, so the meeting becomes the official record of where things stood that day rather than a page that keeps changing underneath you. A completed meeting is read-only, and can be restarted if you ended it early.
Make it sound like your company. Rename the sections to whatever you already call them in Settings, Language, and save whole agendas as reusable formats in Meeting formats.
Where: the follow-ups page for any meeting, reached from the meeting itself.
OrgTP can attend the meeting, record it, transcribe it, and turn what was said into to-dos, issues and headlines you approve. This is the shortest path from "we talked about it" to "somebody owns it".
The recording (video where there was video, audio always), a speaker-labelled transcript, and Ollie's read on the meeting. From there one click drafts to-dos, issues and headlines from the transcript, each of which you approve before it becomes real. Nothing is created behind your back.
A recorded transcript labels turns as "Speaker 1", "Speaker 2" and so on. The Who was speaking? panel maps each one to a real person and rewrites the transcript. If people introduce themselves at the top of the call, OrgTP fills those names in for you and you just check them before applying.
A meeting trapped inside the tool is half useful. Two buttons sit on every meeting.
Opens a tabbed panel with everything the meeting can leave as:
Tabs only appear for what that meeting actually has, so there are no dead buttons.
Creates a short link that opens the meeting. Paste it into Slack or an email instead of making somebody hunt through a list.
Who can open it: anyone on your team who could already see that meeting, and they will be asked to sign in. Forwarding the link outside your organization gets a not-found page, because the link is a shortcut and not a key. Stop sharing turns it off for everyone holding it.
Where: /me/todos, your personal queue, separate from the dashboard.
A to-do is a commitment with a name and a date on it. The queue is built so nothing sits in it unowned or forgotten.
Assign a to-do to several people and you get one shared to-do, not a copy each. Everybody on it sees it, everybody ticks their own box, and the item closes when the last person is done. The card tells you where things stand: "shared with Brian and Dana", "1 of 3 done", and "your part is done, waiting on Dana" once you have finished yours.
That matters because the alternative silently loses work. Separate copies drift apart, get edited individually, and leave nobody able to say whether the thing actually happened.
Open Edit and use the Assigned to picker. It opens with the current people selected, so adding or removing somebody is one click and the rest of the to-do is untouched. Removing the last person who had not finished completes the item; adding somebody new reopens it. Both follow from the same rule: done when everyone on it is done.
Priority P1 to P4, due dates, recurrence, descriptions, file attachments, an inline timer for tracking real effort, and a project to file it under. Done closes it. To Delta promotes it into the leadership meeting. Delegated items can be verified or reopened by whoever handed them out.
Where: /projects for the list, and each project has its own page.
Some work is bigger than a to-do and smaller than a quarterly Rock. Projects is where that work lives: a named piece of work with an owner, a team, a due date, and the to-dos that make it up.
You can file a to-do into a project from the to-do form directly, and create a brand new project from that same form without leaving the page.
Where the hours go: the timer on each to-do rolls up per project, and Reports, Time turns that into hours and cost by person, project and task over any date range. Useful for billing, and for finding out where the quarter actually went.
Where: /dashboard/kpis, the full company scoreboard.
Every measurable number in the company, grouped by the seat that owns it. The point of a scorecard is that a bad week is visible in five seconds, without anybody having to raise it.
A tab strip switches between Trends, Weekly, Monthly, Quarterly and Annual views. A filter row scopes by team, by owner type (all, human, or agent), and by date range. Yes, agents own KPIs here the same way people do.
KPI tables show the title, owner and goal, then either editable per-period cells (green when the goal was met, red when it was missed) or a latest value with a sparkline. Type the actual straight into the cell to record the week.
In the meeting: the Scorecard section reads these numbers and nothing else. No discussion, no explanations. Anything red becomes an issue and gets its time later, which is what stops a scorecard review turning into a forty-minute detour.
A Rock is one big outcome somebody commits to finishing this quarter. Not a project list, not a wish: a named person, a date about 90 days out, and a thing that is either done or not.
There is no separate Rocks page, on purpose. Rocks show up where you are already working:
Each Rock carries an owner, a due date, an On Track or Off Track toggle, and milestones you can tick off underneath it. Off track is not a failure state, it is the signal that this Rock needs meeting time this week.
Where: /dashboard/ids, a cross-team master list of every open issue, for owners, admins and managers.
An issue is anything that needs working through as a team: a problem, a decision, a tension, an opportunity nobody has picked up. Writing it down is the whole trick. An issue on the list gets solved; an issue in somebody's head gets raised at the worst possible moment.
Filter by team and by status (open, identified, discussed, solved). Each row shows the issue, its status, owner, team and date, and you can hand an issue to a different team.
Most issue work happens inside the meeting, in the Issues section, where you take them one at a time: identify what is actually going on, discuss it, then decide. Solving one prompts for a resolution note and an optional follow-up to-do, so the decision survives the meeting it was made in. This page is the bird's-eye view for anyone running several teams.
Where: /dashboard/team. A visual, editable chart of your humans and your AI agents, and who reports to whom. Filter by all, agents, or humans. Click a tile to edit the seat or see its meetings and accountability. Add people, invite by email, or import a roster by CSV.
Edits save to a draft until you press Publish, and publishing is what turns the chart into an OOS your agents can read (section 15). Draft and published are kept apart so a half-finished reorganisation never leaks into what the agents believe.
Seats, not people. A seat is a position with responsibilities and an owner. People come and go from seats; the accountability stays put. An empty seat is a visible, honest state, and somebody joining can claim theirs.
Where: /dashboard/members. Active members and pending invites. Invite by email with a role, a team, and per-feature access. Edit, resend or revoke. Roles run from Owner and Admin down to Observer and Free, and the dropdown explains each one.
Settings, Users goes further: a matrix of who sees which navigation items, so a new member lands in a workspace scoped to their job rather than the whole product at once.
Where: /dashboard/teams. Sub-teams so meetings, issues and KPIs scope to a group rather than the whole company. Name it, add members, done.
Where: /team/ plus the seat id, reached by clicking any name. Shows what that seat owns: quarterly priorities, responsibilities, seat fit, open to-dos, open and solved issues, and meeting history. It answers "what is this person actually accountable for" in one screen.
Where: /dashboard/usage, for owners, admins and managers. Daily activity and a per-member table over any date range. Useful for spotting the person who never got started, well before it shows up as a missed commitment.
Where: /team/review. Rate your direct reports on seat fit and on your company's core values. One row per person; columns for Understands, Wants and Capacity plus each value; a computed verdict of Solid, Needs conversation, or Wrong seat. Click a cell to cycle it.
Where: /dashboard/ollie, and inside any meeting.
Ollie is the intelligence layer over everything above. Not a chatbot bolted onto the side: Ollie reads your meetings, to-dos, KPIs, Rocks, issues and org chart, which is why the answers are about your company rather than about business in general.
Chat first. Ask what happened in a meeting you missed, who owns something, what keeps slipping, what the numbers have been doing. Because Ollie reads the same memory your agents do, the answer cites your actual record.
The read you did not ask for. Insights surface the pattern across weeks: the Rock that has been off track three meetings running, the KPI trending the wrong way before it goes red, the person quietly carrying too much. An Insight can be generated for a meeting, for a page, or for the organization.
Ollie sits inside the Delta Meeting itself, so you can ask mid-conversation instead of breaking to go look something up. Every meeting is expected to include one Ollie Insight, which is the habit that turns a meeting from a status update into a decision.
Ollie is configured in Settings, Ollie. Metered AI features draw on a wallet balance rather than a per-seat fee, and every new organization starts with credit already in it, so you can try this before deciding anything. Settings, Billing shows where you stand. The Lab is where newer Ollie capabilities appear first.
This is the part that turns how you operate into something reusable, by your agents and, if you choose, across organizations.
An OOS (Organizational Operating System) is a structured file of your company's coordination rules and knowledge: who owns what, how decisions get made, what has failed before, and what was learned. Each rule is a Claim, tagged with how confident you are and how much evidence backs it.
Agents read your OOS, which is the difference between an assistant that has to be briefed every morning and one that already knows how your company works. Publishing your org chart is what produces the first one.
Where: /dashboard/oos-operating-plan. A guided planning workspace covering where you are going, the numbers you are driving toward, and an honest look at strengths and weaknesses. It turns strategy into a 90-day execution list and into OOS claims. Fill in the tabs, generate the plan, and admins can push it into the OOS.
Where: /dashboard/ceo, for owners. The whole company on one screen: long-term direction, scoreboard health, this quarter's execution, the team, and what needs attention. Driven entirely by the Operating Plan, so it is never a second thing to maintain.
Where: /claims and /graph. Coordination knowledge published across the platform, browsable by topic, each claim carrying its confidence, evidence, and the failure mode it exists to prevent. The graph maps how organizations connect through shared patterns, and in Best Practices mode you can pull one into your own org as a claim.
This is the part that compounds. A lesson one organization pays for once becomes something the next one can start with.
Where: /dashboard/source-documents. Upload the raw material you already have, a playbook, your SOPs, a CLAUDE.md, and OrgTP extracts the coordination intelligence into OOS files. You do not have to author any of this from scratch.
Where: /blueprints. Switch Blueprints on in Settings, Labs and it appears in your sidebar.
A Blueprint is a versioned package of how an organization runs, captured from a real working account and applied to another. Configuration travels. Data never does.
What travels: your language and custom section names, meeting formats and agendas, KPI definitions and groups, teams and the shape of the org chart, seat responsibilities, Ollie's configuration, priority templates, and operating plans.
What never travels: people, live KPI values, meeting history, to-dos, keys, and billing. Seats travel as roles, never as the people sitting in them.
Open seats: an organization created from a blueprint lands with a chart full of open seats. Every role is there, nobody is in one. That is the honest state, and it is the point. As each person arrives they claim their seat and the chart fills in.
Applying a blueprint shows a preview first and can be undone. Later versions can be pushed to linked organizations, which is what makes this useful to a coach or a franchisor running the same operating model across many companies.
Switching from another operating tool? Do not retype it.
Where: /import/ninety for Ninety.io, /import/bloom for Bloom Growth. Both work identically.
Re-running an import does not create duplicates, so a partial or nervous first attempt costs you nothing.
Every company already has its own words for this stuff. Fighting that is how software gets abandoned, so OrgTP bends instead.
Core values live with your organization too, and show up in People Review so you are rating people against what you actually said mattered.
Reached from the settings menu. Personal settings affect only you; company settings need admin rights.
| Page | Where | What it is for |
|---|---|---|
| Profile | /settings/profile | Your details for this org: pronouns, title, bio, work-style metrics, extra emails and phones, LinkedIn. |
| Account | /settings/account | Your login identity: email, password, two-factor, connected accounts. |
| Preferences | /settings/preferences | Theme, date format, default team, timezone. |
| Communication | /settings/communication | Everything OTP can send you, with a switch for each. Applies to your address across every organization. |
| My data | /settings/my-data | What OrgTP holds about you, and getting a copy of it. |
| Billing | /settings/billing | What your agent team costs, with a live monthly estimate and your current wallet balance. |
| Integrations | /settings/integrations | Connect outside tools to your agents. |
| API keys | /settings/api | Create and revoke the keys that let agents connect. See section 20. |
| Ollie | /settings/ollie | How the intelligence layer behaves for your organization. |
| Meeting presence | /settings/meeting-presence | What the notetaker says when it joins a call, and whether it announces itself. |
| Redaction | /settings/redaction | Terms to blur out of transcripts. |
| Language | /settings/language | Rename sections and vocabulary to match your company. |
| Users | /settings/users | Who can see which parts of the product. |
| Teams | /settings/teams | Team membership, from the settings side. |
| Labs | /settings/labs | Switch on beta features such as Blueprints. |
| Coaches | /settings/coaches | Which outside coaches can see your workspace, and revoking that. |
| Import | /settings/import | Bring data in from another tool. |
| Configuration | /settings/configuration | Company name, website, description, and whether to list the org publicly. |
| Danger zone | /settings/danger | Destructive organization-level actions, behind confirmations. |
This is how an agent such as Claude Code or ChatGPT actually reads and writes your OrgTP data. It is the half of the product nobody else has, so it is worth ten minutes.
Where: /settings/api. Click Generate New Key, name it, and copy it immediately, because it is shown once. The page hands you a copy-paste MCP setup snippet so the agent connects in one step. Revoke any key at any time.
MCP (Model Context Protocol) is the standard that lets an agent talk to OrgTP's server safely. Read-only tools work without a key; publishing to your OOS requires one.
Where: /dashboard/agents/connect. Per-person connection, one key each, so an agent acts as somebody rather than as a nameless service. That is what keeps the audit trail meaningful when an agent creates a to-do or updates a KPI.
Where: /settings/integrations. Connect the outside tools your agents need to reach. A large catalogue of providers is available, and connecting one there makes it usable from Ollie rather than requiring separate plumbing.
There is a fuller walkthrough at Connect your agent, including a copy-paste prompt that generates your first OOS.
/l8.