Skip to content
Activity Tracking

Every prompt, session and commit, tracked for the whole team

CloudByte AI captures prompts, observations, sessions, tokens, tickets and commits from every developer machine, then adds them up per project and team so you see exactly how AI work gets done.

Start freeBook Demo
01Prompts

Every prompt, from every developer, in one feed

Each prompt is captured as it is sent, with who sent it, the project and repository, the model that answered and the response that came back. Filter the feed by developer, project or team to see how AI is really used, day to day.

  • Prompt and response captured together, with time and duration
  • Developer, project, repository and branch on every entry
  • Filter by developer, project, team or date
  • Totals per project and per team, not just per person
02Observations

Observations sorted into bugfix, feature, decision and more

While the AI works, CloudByte AI records what actually happened as observations and sorts each one by type. You see at a glance where AI time goes: fixing bugs, building features, refactoring code or making decisions.

  • Six types: Bugfix, Feature, Refactor, Change, Discovery and Decision
  • Counts per type for each developer, project and team
  • Every observation linked to the session and prompt behind it
  • Decisions and discoveries flow straight into the Knowledge Base

Activity · Payments team

  • BugfixDuplicate refunds when webhook retries overlappriya · payments-api · 10:12
  • DecisionEvery refund carries an idempotency keypriya · payments-api · 10:31
  • FeaturePartial refunds added to the checkout flowalex · checkout-web · 11:05
  • RefactorPayment client split into retry and transport layerssam · payments-api · 11:40
  • DiscoveryThe ledger job reads from a replica with 30s lagmaya · ledger · 13:18
  • ChangeWebhook timeout raised from 5s to 10sraj · payments-api · 14:02
  • BugfixCurrency rounding off by one cent on partial refundsalex · checkout-web · 14:47
  • FeatureRefund status shown on the order pagemaya · checkout-web · 15:20
03Sessions

Follow a whole session, prompt by prompt

A session groups everything that happened in one sitting: the prompts sent, the responses that came back and the observations generated along the way. Open any session to read it from start to finish and see what kind of work it really was.

  • Prompts, responses and observations in order, on one timeline
  • An AI summary of every session when it ends
  • Duration, tool calls and prompt count for each session
  • Compare sessions from developers on the same project
04Tokens and models

Input, output and model details for every session

Every session records how many input and output tokens it used, which model answered each prompt, how much came from cache and what it cost. Spend is tied to real work instead of a single monthly invoice.

  • Input, output and cache tokens per prompt and per session
  • The model behind every response, including switches mid-session
  • Cost per session, developer, project and team
  • Duration, tool calls and prompt count next to the tokens
05Tickets

Know which ticket each developer is working on

CloudByte AI connects sessions to tickets from Jira and Linear, so you can see what each developer is working on right now and how much AI work went into every ticket.

  • Tickets matched from branch names, commit messages and the session itself
  • Sessions, prompts and tokens added up per ticket
  • What each developer is working on, at a glance
  • Tickets with lots of AI time and no commit yet stand out
06Commits

Commits linked to their ticket, prompts and observations

Every commit is linked to the ticket it belongs to and to the prompts and observations that produced it. Trace any change from the ticket, through the conversation that wrote it, to the code that shipped.

  • Ticket, session, prompts, observations and commit in one chain
  • An AI summary of every commit
  • Works with GitHub, GitLab and Bitbucket
  • Commit rate per developer, project and team
Team totals

Tracked per developer, added up per project and team

Every prompt, observation, session, ticket and commit belongs to a developer and a project. CloudByte AI adds them up at each level, so leads see their team and each developer sees their own work.

Developer

priya

Sessions
14
Prompts
126
Observations
38
Tokens
1.2M

Project

payments-api

Developers
5
Sessions
61
Tickets
17
Commits
43

Team

Payments team

Projects
3
Developers
9
Sessions
148
Spend
$412

  • Captured automatically

    The agent records activity as developers work. Nobody fills in a timesheet.

  • Synced every 3 minutes

    Prompts, sessions and observations reach the dashboard within minutes.

  • No keylogging

    No keystrokes, no file contents and no environment variables, ever.

  • Inside your organization

    Activity stays within your organization, and people only see the projects they can access.

FAQ

Activity Tracking questions, answered

What is captured, how it is linked together and who can see it.

Ask us anythingBook a demo

Prompts and their responses, observations sorted by type, sessions, input and output tokens with the model used, the tickets developers are working on, and the commits linked to them. Everything is tracked per developer and added up per project and team.

Notes CloudByte AI records about what happened while the AI worked, sorted into six types: Bugfix, Feature, Refactor, Change, Discovery and Decision. Each one is linked to the session and prompt behind it.

Everything that happened in one sitting with the AI: the prompts sent, the responses that came back and the observations generated along the way, shown in order on one timeline with an AI summary at the end.

Input, output and cache tokens for every prompt and session, the model that answered each prompt, the cost, the duration, the number of tool calls and the prompt count.

From branch names, commit messages and the session itself, using your Jira or Linear tickets. You see which ticket each developer is working on and how much AI work went into it.

Each commit is linked to the sessions that produced it, and through them to their prompts, their observations and the ticket. Commits come from GitHub, GitLab or Bitbucket and get an AI summary.

It is built for coaching, not surveillance. There is no keylogging and no capture of file contents or environment variables, and people only see the projects they have access to.

Prompts, observations and sessions are tracked on every plan, including Free. Token spend analytics, AI commit summaries and GitHub need Starter or above. History is kept for 7 days on Free, 90 days on Starter and without limit on Team.

See every AI session your team runs.

Install the agent once per laptop and prompts, sessions and commits start showing up within minutes.

Start freeBook Demo
  • Free for up to 5 developers
  • 10-minute setup
  • Your Anthropic key, your spend (Enterprise)