Prove · Capability

Your team's AI capability

What one person learns should reach the next person without a document. This measures that: whose work others use, and how fast new work spreads.

Get Started

Three tabs of the Team Dashboard, shown side by side. In the product they are separate screens.

Training tab

Team mastery heatmap

TalkActShareBuildSched Sarah K. Marcus T. Priya M. Jordan L.

Each member by capability: who has what, and where the gaps are.

Usage tab

Engagement tiers

Power · 70+

Regular · 40-69

SKMT

Occasional · 1-39

Dormant · 0

none

Tools tab

Member coverage

Member Score

Priya M. 88

Sarah K. 64

Marcus T. 51

Jordan L. 22

Sorted by score by default. Owners and admins only.

What Capability gives you

Teammate Uses Per Author Use

One person builds something. How many times do teammates use it? That ratio is the whole argument for a shared cloud, and it is a number rather than a claim.

Time For New Work To Spread

How long between somebody making something and other people using it. If that number falls, the team is compounding.

Who Is Contributing, Who Is Using

Members whose work teammates use, and members using teammates' work. Two different problems and you want to see both.

Who Is Actually Making Things

Active builders, apps created and apps deployed, plus the workflows, schedules, agents, endpoints and entities your team registered. The gap between created and deployed is usually the interesting one.

Who Can See This

Workspace owners and admins see per-member capability data, because somebody has to know who needs help. Members do not have this page: the server refuses the request, not just the interface.

Talk, Act, Make

Three stages, not a score. Talking with AI, acting with AI, making with AI. You can see which one somebody is actually at rather than guessing from how much they use it.

What's Next For You

The capabilities you can reach right now, because everything they depend on is already yours. Not a list of everything you could theoretically learn, which is the same as no list at all.

Access from anywhere

AI Agents

Ask your AI agent:

Connect your AI agent

Desktop & Web

Runwork Desktop

Your AI Tools 13 detected

Codex DEFAULT

Claude Code

Cursor

+10 more

Your Team's Skills 39 shared

press-release-lead-monitor

dtp-demand-tracker

Your Integrations 18 connected

HubSpot Sheets lemlist +15

The Knowledge tab shows contributors and shared apps. My Journey shows your own stages, what is next for you, and anything fading.

Get the Desktop app Open in browser

CLI

Why Capability matters

Counts who is making things, and how far what they make travels
Two ladders: what a person can do, and how far it has travelled
Shows what you are about to lose, before you lose it
Active builders as a number, not an impression
Per-member scores, visible to owners and admins only

How Capability works

You do things with your AI

Talking with AI, acting with AI, making with AI. Three stages rather than a score, and you can see which one you are actually at instead of guessing from how much you use it.

The next thing becomes reachable

Capabilities have prerequisites. What is next for you is what you can reach right now because everything it depends on is already yours. Not a list of everything you could theoretically learn, which is the same as no list at all.

Or it fades

Something you have not done in a while decays, and the page tells each person what they are about to lose before they lose it. Uncomfortable, and also how memory works.

Somebody else picks it up

That is the number that matters. How many times teammates use a piece of work for each time its author does, and how long it took to get there. If that time falls, the team is compounding.

And somebody has to make it first

It counts creation, not only spread. Active builders, apps created, apps deployed, and the workflows, schedules, agents, endpoints and entities your team has registered. Those are the same question at two distances: who makes something, and how far it gets. A team where the first number is one person is fragile however well that person's work travels, and a team where five people build things nobody else picks up is busy rather than compounding.

Frequently Asked Questions

Who can see this?

Workspace Owners and Admins. Other members see their own workspace but not the team-wide analytics.

How do I know the team is actually getting better and not just busier?

Two numbers. How many times teammates use somebody's work for each time the author uses it, and how long it takes something new to reach other people. Busy shows up in message counts. Getting better shows up in those two.

What is the difference between my journey and the team view?

Two ladders that are not the same axis. Yours is talking, acting, making, and it is about what you can do. The team's is setup, usage, building, knowledge, and it is about how far a practice has spread. A team can be strong on one and weak on the other.

Does this rank individual people?

For owners and admins, yes. They see a per-member table with a 0 to 100 score, sorted by that score by default, because finding who is stuck is the point of the page. Nobody else can: the server refuses the request from a regular member, and the reason is written into the code. Everyone else sees their own journey. So the ranking is real and it is scoped to the people who have to act on it, rather than a scoreboard the whole team can see.

Does this show who is building and who is only using?

Yes. Active builders counts the people who made something, and it sits beside apps created, apps deployed, and the workflows, schedules, agents, endpoints and entities registered. A team where one person builds and everyone else consumes reads very differently from one where five do, even when the totals match.

How are Adoption Scores calculated?

Each dimension (Setup, Usage, Building, Knowledge) is scored 0-100 based on specific signals. Setup tracks tool installation and sync status. Usage tracks active days and messages. Building tracks apps, workflows, and schedules created. Knowledge tracks skills shared and components contributed.

Does it work with all AI tools?

Yes. The dashboard tracks adoption across every agent Runwork configures, including ChatGPT, Codex, Claude Code, Claude Desktop, Cursor, Windsurf, Gemini CLI, Antigravity, and Cline.

Can I configure what tools my team uses?

Yes. The Onboarding tab lets you set approved tools, mark some as required, and write team-wide instructions that get pushed to every member's AI tools.

Is usage tracking privacy-friendly?

Telemetry is a workspace setting rather than a personal one. Your workspace admin picks the level and it applies to everyone, so an individual member cannot turn it off on their own. The dashboard then works with whatever that leaves available.

Use Cases

Seeing whether one person's work reaches anyone else Finding who to ask about a particular kind of work Spotting a practice that spread and one that did not Onboarding someone into what the team already knows Checking whether adoption is deepening or just widening

Related Features

Adoption & Training Chats & Insights Desktop App AI Agents Skills Integrations Share & Resume

See How Teams Use Capability

Share AI Skills AI Ecosystem

Ready to try Capability?

One shared cloud under the AI tools your team already uses.

Get Started Explore More Features