Today

Decisions evaluated

Outcomes reported

Learned cases

Follow rate

—

Positive after following

—

Estimated value protected

across 0 decisions

followed 0

avg. saved $0

Decision & outcome activity

Decisions Outcomes Followed

Recent decisions

View all

Agent	Action	Recommendation	Confidence	Agent followed?	Outcome	Time
No decisions yet Use the Decision Sandbox or send your first API call to get started.

Agent

Action

Recommendation

Confidence

Agent followed?

Outcome

Time

No decisions yet

Use the Decision Sandbox or send your first API call to get started.

Observed outcomes

Action	Real outcome	Impact	Linked decision	Time
No outcomes yet Report what happened after execution so future decisions improve.

Action

Real outcome

Impact

Linked decision

Time

No outcomes yet

Report what happened after execution so future decisions improve.

Decisions

What BIGHUB recommended, what the agent did, and what we learned.

	Action	Agent	Domain	Recommendation	Confidence	Agent followed?	Outcome	Learning signal	Mode	Time	Action
No decisions yet Decisions appear here when your agents evaluate actions through BIGHUB. Try the Decision Sandbox to send your first test action.

Action

Agent

Domain

Recommendation

Confidence

Agent followed?

Outcome

Learning signal

Mode

Time

Action

No decisions yet

Decisions appear here when your agents evaluate actions through BIGHUB. Try the Decision Sandbox to send your first test action.

No results

Outcomes

What actually happened after actions ran.

Outcomes this month

—

Positive rate

—

Pending outcomes

—

Action Decision Outcome Impact Cost / value Time

Action	Decision	Outcome	Impact	Cost / value	Time
No outcomes yet Outcomes close the learning loop. Use `client.outcomes.report(...)` or click "Report outcome" on any decision.

No outcomes yet

Outcomes close the learning loop. Use client.outcomes.report(...) or click "Report outcome" on any decision.

No results

Learning

How BIGHUB improves decisions from real outcomes.

Follow rate

—

Agents following recommendations

Positive after following

—

Outcome success when followed

Learned cases

—

Decisions with closed outcomes

Calibration accuracy

—

Predicted risk vs actual outcome

Recommendation quality

Was BIGHUB's advice good? Measured by what happened after.

Followed + positive

—

Recommendation confirmed

Followed + negative

—

Recommendation under recalibration

Ignored + positive

—

Agent divergence outperformed

Ignored + negative

—

Recommendation likely should have been followed

These metrics appear as outcomes are reported. They measure whether following BIGHUB's advice leads to better results.

Follow rate by domain & agent

Which domains and agents follow BIGHUB's recommendations, and what outcomes follow

Not enough outcome data yet. Comparative view appears once outcomes are reported.

Notable examples

Real cases where BIGHUB's advice made a difference — or where it didn't

No examples yet. These appear once outcomes are reported and linked to recommendations.

Recommendation quality over time

Weekly follow rate and positive outcome rate — does BIGHUB improve?

Follow rate Positive after following

Not enough weekly data yet. Trend appears after 2+ weeks of outcomes.

Decision quality over time

How recommendation quality improves as outcomes accumulate

Not enough data yet. Quality trends appear after outcomes are reported.

Precedent freshness

Are your precedents still relevant?

Freshness score

—

Precedents < 7 days —

Precedents > 30 days —

Detected drift No drift

Outcome prediction

How well BIGHUB predicts what will happen

Prediction accuracy —

Correct predictions —

Wrong predictions —

Avg. predicted risk vs actual —

Predictions improve as more outcomes are reported. BIGHUB uses precedents to forecast the probable result before execution.

Learned patterns

Pattern	Domain	Confidence	Cases
No patterns detected yet. Patterns emerge as outcomes accumulate.

Calibration

Domain	Predicted risk	Actual outcome	Drift
Not enough data for calibration. Report outcomes to start tracking prediction accuracy.

Decision Sandbox

Test how BIGHUB evaluates an action, what context matters, and how learning affects the final decision.

Configure action

Domain

Value ($) required

Action type

Constraint match: —

Target (optional)

Decision context

Actor type

Customer tier

Result

Run an evaluation to see the decision breakdown.

Constraints

Operating limits that inform BIGHUB's recommendations. Enforced only when your runtime adapter supports it.

	Name	Domain	Constraint	Review threshold	Usage	Status	Actions
No constraints yet. Create your first operating constraint to start shaping recommendations.

Showing 0 to 0 of 0 constraints

Create constraint

Decision domain

Constraint name *

Domain *

Description

Operating limits

Hard limits that apply while BIGHUB evaluates and learns

Action budget (/day) *

Hard value limit (/action) *

Human Oversight

When a human must review before execution

Review above ($)

Learning setup

BIGHUB learns automatically from outcomes

Track outcomes — always on Use similar cases when available — always on Start calibration once outcomes accumulate — always on

Emergency Stop

Operational · 0 constraints paused

Advisory mode: Emergency stop only takes effect when your runtime adapter (e.g. bighub-openai) is configured to respect it. In advisory mode, BIGHUB recommends — your agent makes the final call.

Scope	Reason	Time	Status

Reviews

Actions that need review before execution.

0 pending · 0 approved today · 0 denied today

Action	Domain	Value	Actor	Requested	Actions
No actions currently need human review. Reviews appear when constraints or learned signals recommend extra caution.

No results

API Keys

Manage API keys for programmatic access

Total Keys

No keys yet

Requests (Today)

Across all keys

Last Used

—

ago

Label	Key Prefix	Scopes	Rate Limit	Last Used	Status	Actions
No API keys yet. Click "Create API Key" to generate your first key.

Team

Manage team members and permissions

Total Members

You are the only member

Owners

Admins

Members

Member	Role	Status	Last Active	Actions
You are the only member. Click "Invite Member" to add team members.

Notifications

Configure how BIGHUB alerts you about important events

Email Notifications

Receive alerts via email

Email Recipients

Comma-separated list of email addresses

Slack Notifications

Send alerts to a Slack channel

Webhook URL

Create an incoming webhook in your Slack workspace settings

Discord Notifications

Send alerts to a Discord channel

Webhook URL

Create a webhook in your Discord server settings

Event Types

Choose which events trigger notifications

Do Not Proceed recommendation Warning Review Required Warning Emergency Stop Activated Critical Approaching Limit (80%+) Warning Decision Allowed Info

Integrations

Connect agents fast, evaluate decisions, then report outcomes to improve judgment over time.

Official Packages Choose one to load example code

Implementation Code Python SDK

# Select an integration above

Install: pip install bighub

SDK Reference

Base URL: https://api.bighub.io
Evaluate Action: client.actions.evaluate(...)
Report Outcome: client.outcomes.report(...)
Dry Run (Test): client.actions.dry_run(...)

Primary response fields

Evaluation

recommendation
recommendation_confidence
risk_score
decision_intelligence
request_id

Legacy (compat)

allowed, reason

Outcome report

request_id
status

Use client.actions.evaluate(...) (or MCP bighub_actions_evaluate) then client.outcomes.report(...) to close the loop.

Decision & outcome loop (Python SDK)

Loading example…

Resources

bighub.io GitHub PyPI — bighub PyPI — bighub-openai npm — @bighub/bighub-mcp

Need help? Run the setup wizard again

Today

Decision loop

Free BETA quota usage

Decision & outcome activity

Recent decisions

Observed outcomes

What BIGHUB is learning

Counterfactual impact

Operating constraints

Decisions

Outcomes

Learning

Recommendation quality

Follow rate by domain & agent

Notable examples

Recommendation quality over time

Decision quality over time

Precedent freshness

Outcome prediction

Decision replay

Feedback loop speed

Learned patterns

Calibration

Configure action

Result

All Notifications

Constraints

Create constraint

Decision domain

Operating limits

Human Oversight

Learning setup

Edit constraint

Delete constraint

Change execution domain

Emergency Stop

Activate Emergency Stop

Resume Execution

Reviews

Deny Request

Report outcome

Configure review workflow

Account

User identity

Security controls

API Keys

Create API Key

API Key Created

Team

Invite Team Member

Remove Team Member

Notifications

Email Notifications

Slack Notifications

Discord Notifications

Event Types

Integrations

SDK Reference

Primary response fields

Resources

Decision details

Filter

Notifications