Skip to content
Flexday AI Docs

Capabilities

Observability and insight

See how Solutions are used and what they cost, with usage dashboards, history, versions, run step trails, Agent analytics and global search.

Written for
  • Everyone

Last reviewed

Flexday AI measures how what you build is used. Every completed model call is metered, changes to resources are recorded with the person who made them, every Flow run keeps a step-by-step trail, and every Agent has analytics of real use. Most resources have an Insights menu, though the views it holds vary by resource.

What you can do

  • See what each Solution costs. Usage and AI cost roll up per Solution, including its Agents and the AI steps in its Flows.
  • Find out who changed what. Lists show who last changed each item and when, and workspace admins can read the full audit trail through the platform's API. A resource's History tab lists its AI operations.
  • Follow versions. A resource's Versions tab lists each deploy or publish, with a short summary of what changed and when.
  • Debug a workflow. A Flow run's step trail shows every step's input, output, status and timing.
  • Understand how an Agent is really used. Analytics show volumes, people, tools, cited sources, guardrail events and feedback, leaving out Playground chats and evaluation runs.
  • Know who downloaded a file. Each File Store keeps an access log of downloads, previews and issued links.
  • Find things. Global search finds Solutions, Templates, people and most resources inside the Solutions you can see.
The Usage & cost page of the Harbourline workspace for the last 30 days: total cost, tokens, model calls and average cost per operation, a chart of cost per day, and the start of the cost by model and cost by service breakdowns
Notice the date ranges and the Export CSV button above the totals, and the breakdowns below them: by model, and by the stage of work that spent the tokens.

How it works

ViewWhereWhat it shows
UsageThe workspace's Usage & cost page; Solutions, Fact Bases, Data Portraits, Doc Bases, LaunchPads, File Stores and AgentsTokens and estimated cost by model, task and resource, with a CSV export
AnalyticsSolutions, Flows, AgentsActivity over time; for Agents, Playground and evaluation conversations left out
HistorySolutions, Templates, Fact Bases, Data Portraits, Doc Bases, File Stores, LaunchPads and AgentsAI operations: builds, inference, profiling and more
VersionsApps, Flows, Agents, Fact Bases and Data PortraitsEach version, newest first; apps and Fact Bases add a draft row when changes are waiting
Runs and sessionsFlows and AgentsStep trails, turns, tools and citations
Access logEach File StoreDownloads, previews and listings by file, caller and date
Build logThe BuilderOne line per action the Builder took

These views follow the same access rules as everything else. In a Restricted workspace (a setting the platform's operators apply; there is no Studio switch for it), members and viewers see usage for the Solutions shared with them; owners and admins see the whole workspace.

The Tools and Knowledge section of the Service desk agent's Analytics: its document search tool with a call count and an average time, the IT knowledge Doc Base as its knowledge source, and the IT articles its answers cited, with how often
Notice that these counts come from conversations outside the Playground: Playground chats and evaluation runs are left out of analytics.

Example

A finance lead wants to know why the monthly AI bill rose. The workspace Usage page shows most of the increase came from one Solution. Its Usage view shows an Agent's model was changed to a larger one. The Agent's Analytics show conversations doubled after a Teams Bot went live, and the workspace's audit trail shows who changed the model and when. The team decides to keep the larger model for that Agent and switch evaluations to a cheaper judge model.

Who uses it

PersonaWhat they get
Functional userCost and adoption of the Solutions they own
BuilderRun trails and build logs to see exactly what happened
IT and securityThe audit trail and file access log for reviews and investigations
Agent ownerReal-use analytics and feedback to improve instructions and knowledge