Activity & pricing

The goal's activity feed and audit timeline, what goals cost, and publishing the goal's segment.

The activity feed

One call answers "what's happening and what's next": the goal's lifecycle as four phases — Discover → Auto-fit → Deploy → Live — each with a state, plain-language reasoning, and a next-step action:

curl "$API_BASE/v1/goals/$GOAL_ID/activity" \
  -H "Authorization: Bearer $TOKEN" \
  -H "X-Workspace-Id: $WORKSPACE_ID"

from predictai import PredictAI

client = PredictAI(token="pa_live_…", workspace_id="ws_…")

data = client.goals.get_goals_by_goal_id_activity(GOAL_ID)

import { PredictAI } from "@predictai/sdk";

const client = new PredictAI({ token: "pa_live_…", workspaceId: "ws_…" });

const data = await client.goals.getGoalsByGoalIdActivity(GOAL_ID);

{
  "data": {
    "goal_id": "b7e9c2d4-…",
    "status": "serving",
    "phases": [
      {
        "id": "discover",
        "title": "Discover drivers",
        "state": "done",
        "summary": "7 drivers found",
        "reasoning": "3 more are suggestive (shown dimmed in the graph).",
        "metrics": [{ "label": "Candidates", "value": 64 }, { "label": "Drivers", "value": 7 }],
        "cta": { "label": "Re-run", "action": "rerun" },
        "refs": { "run_id": "a1f3…" }
      },
      {
        "id": "fit",
        "title": "Auto-fit best model",
        "state": "done",
        "summary": "Best model · Driver-weighted trees",
        "metrics": [{ "label": "Accuracy", "value": "91.4%" }, { "label": "Skill", "value": "34.2%" }],
        "refs": { "tournament_id": "e7a1…", "model_id": "9f2c41d8-…", "training_id": "t_5b2e…" }
      },
      { "id": "deploy", "title": "Deploy to serving", "state": "done", "summary": "Live", "refs": { "promotion_id": "p_88c1…" } },
      { "id": "live", "title": "Monitor & retrain", "state": "active", "summary": "Full autopilot" }
    ],
    "stats": {
      "drivers": { "validated": 7, "exploratory": 3 },
      "model": { "label": "Driver-weighted trees", "score": 8.41, "accuracy": 91.4, "skill": 34.2, "state": "done" },
      "deploy": { "state": "done", "live": true, "status": "deployed" },
      "serving": { "inference_count": 212, "last_inference_at": "2026-07-15T06:00:12+00:00", "next_inference_at": "2026-07-16T06:00:00+00:00", "live": true }
    },
    "resources": {
      "model_id": "9f2c41d8-…",
      "training_id": "t_5b2e…",
      "segment_id": "seg_44aa…",
      "deployment_id": "p_88c1…",
      "last_inference_id": "3f1a…"
    },
    "config": { "auto_fit": true, "auto_deploy": "review", "autopilot": "full", "retrain": "auto", "live": true }
  }
}

Phase states are pending, active, review (waiting on you), done, failed, or off. resources gives you stable IDs into the underlying artifacts — the champion model, its training, segment, deployment, and the latest forecast. Poll this while work is in flight, or subscribe to the goal's realtime channel instead.

The timeline

The autopilot's accountability record: every automated action with its trigger, references, and charge — plus the current autopilot status:

curl "$API_BASE/v1/goals/$GOAL_ID/timeline?limit=50" \
  -H "Authorization: Bearer $TOKEN" \
  -H "X-Workspace-Id: $WORKSPACE_ID"

from predictai import PredictAI

client = PredictAI(token="pa_live_…", workspace_id="ws_…")

data = client.goals.get_goals_by_goal_id_timeline(GOAL_ID, params={"limit": 50})

import { PredictAI } from "@predictai/sdk";

const client = new PredictAI({ token: "pa_live_…", workspaceId: "ws_…" });

const data = await client.goals.getGoalsByGoalIdTimeline(GOAL_ID, { params: { limit: 50 } });

{
  "data": {
    "goal_id": "b7e9c2d4-…",
    "autopilot": {
      "mode": "full",
      "monthly_spend_cap": 500.0,
      "month_spend": 75.0
    },
    "timeline": [
      {
        "ts": "2026-07-14T02:10:00+00:00",
        "actor": "system",
        "action": "retrain_triggered",
        "reason": "drift detected on ad_spend",
        "refs": { "training_id": "t_77e0…" },
        "charge": 50.0
      },
      {
        "ts": "2026-07-10T09:12:00+00:00",
        "actor": "user",
        "action": "serve_requested",
        "reason": "Forecast now",
        "refs": {},
        "charge": 0.0
      }
    ]
  }
}

Parameter	Meaning
`limit` optional	Rows returned, newest first, capped at 200.

month_spend sums the automated charges this UTC month; when the next automated action would exceed monthly_spend_cap, the autopilot pauses itself and notifies you — monitoring continues, spending stops.

What goals cost

Goals charge per action, upfront, with automatic refunds when an action fails to start:

Action	Charged when	Rate key
Discovery run	Queued — at creation and on every re-run	`discovery.run` (scaled 1×/2×/4× by workspace candidate volume)
Tournament	Queued	`discovery.tournament` (scaled by model selection)
Live day	Per day a goal serves	`discovery.live_day`
Prediction	Per inference on Shared serving	see Billing
Scenario / parse / explain / sensitivity	Per run; saved results are free to view	`discovery.scenario*`
Automated retrain	When the autopilot retrains	standard training meters
Fleet bake-off	Once at fleet goal creation	`discovery.fleet_bakeoff`
Fleet scoring run	Every fleet run, priced in member blocks	`discovery.fleet_run`
Fleet scenario	Per covariate-shock scenario	`discovery.fleet_scenario`

The create response's credits_charged is the first run's charge; GET /v1/goals/price shows every rate for your plan before you commit, and GET /v1/goals/{goal_id} reports the goal's lifetime total_cost. Rates, credits, and the ledger live in Billing.

Publish the goal's segment

Discovery compiles an internal segment — the target plus its validated drivers, correctly lagged. Publish it as a normal, editable segment you can use in your own pipelines:

curl -X POST "$API_BASE/v1/goals/$GOAL_ID/publish-segment" \
  -H "Authorization: Bearer $TOKEN" \
  -H "X-Workspace-Id: $WORKSPACE_ID"

from predictai import PredictAI

client = PredictAI(token="pa_live_…", workspace_id="ws_…")

data = client.goals.post_goals_by_goal_id_publish_segment(GOAL_ID)

import { PredictAI } from "@predictai/sdk";

const client = new PredictAI({ token: "pa_live_…", workspaceId: "ws_…" });

const data = await client.goals.postGoalsByGoalIdPublishSegment(GOAL_ID);

{
  "data": {
    "segment": {
      "uid": "seg_9b3d…",
      "name": "Daily sales · discovered drivers",
      "origin": "discovery_published",
      "…": "…"
    },
    "created": true
  }
}

A 201 means a fresh copy was created. Publishing is idempotent: while the published copy exists, calling again returns it with 200 and created: false. The copy is yours — it counts toward your plan's segment quota (403 — "Segment limit exceeded. Your plan allows 10 segments per workspace.") and editing it never affects the goal, which keeps serving from its own internal segment.

The activity feed

The timeline

What goals cost

Publish the goal's segment

On this page