0 of 0 rated outputs
Target 80%PILOT PROGRESS
Run a pilot you can prove.
Agree the work, repeat real tests and make the next product decision from visible usefulness and reliability evidence.
65/100
READY FOR THE PILOT
3 foundations still need attention before handover.
Set up the pilotPlan not saved
Set the terms of the pilot
Agree the jobs, success measures and review rhythm before testing begins.
PILOT SCORECARD
Evidence, not activity
Only completed pilot tests count here.
0 of 0 checked outputs
Supporting evidenceof 5 minimum runs
Across the agreed scopeacross 0 defined cases
QA pass + useful ratingLEARNING LOOP
Make the next run better
Use real ratings and intake outcomes to choose the next improvement. The main pilot score stays simple; the evidence opens below.
Waiting for evidenceRun the first tracked job
Open next step →A completed Skill output with a human rating gives Alex its first useful signal.
No ratings yet
Completed pilot outputs
0 became tracked Work
Patterns appear after feedback
LEARNING EVIDENCE · LAST 90 DAYSSee the patterns behind the next stepOpen detail
SKILL PATTERNS
Where Alex is strongest
Only tracked pilot runs count.
Add a pilot test case to start comparing real Skills.
REPEATED FEEDBACK
What people want changed
Needs-work tags are grouped so one comment does not get lost.
No repeated improvement pattern is recorded.
INTAKE OUTCOMES
What material creates work
Manual Intake remains secondary intelligence and keeps its source trail.
No manual intake has been submitted in this window.
ALEX RECOMMENDS
Where attention may pay off
Alex finds opportunities from live workspace evidence. You decide what enters Work.
WorkPriority 5
Resolve overdue work: The UK Parliament is conducting an inquiry into how the UK can maintain secure, affordable and low-carbon electricity during periods of very low wind and solar
This item passed its deadline and remains waiting.
Open the existing Work Item and reset the owner, deadline or next action.QualityPriority 4
Complete the evidence for Planning Reform: Internal Leadership Brief
This recent Skill output still needs Alex QA and a human usefulness rating.
Open the draft, run the missing check and record whether it was genuinely useful.ReadinessPriority 3
Set the strategy for Skills & Apprenticeships
This active Issue has no current draft or approved Issue Strategy.
Open the Issue, generate a first strategy and amend it before approval.SEVEN-DAY RELIABILITY
Is the operating layer dependable?
87%
959 of 1104 completed events were clean.
Parked Microsoft activity is excluded.
Mon
Tue
Wed
Thu
Fri
Sat
Sun
REPEATABLE TEST PROGRAMME
Prove Alex on real Public Affairs jobs
Each run preserves the exact brief and creates reviewable Work.
1
Add the first real job
Start with a recurring brief, meeting preparation or correspondence task the team already performs.
FOUNDATION & QUALITY DETAILInspect the evidence beneath the pilotOpen detail
FOUNDATIONS4/7
Core pilot steps complete
WORKING NOW2/6
Areas ready to use
QA CHECKS0
0 passed in 30 days
HUMAN SIGNAL1
0 rated useful
GETTING READY
What the workspace still needs
Google, Microsoft and Slack are helpful, but optional for the first useful output.
1
CORE FOUNDATIONSecure the workspace
Require a signed-in user and keep every organisation inside its own workspace boundary.
Prototype modeTurn on secure mode →
✓CORE FOUNDATIONExplain the organisation
Give Alex enough commercial and operating context to understand why policy change matters.
4/4 core fieldsReview profile →
✓CORE FOUNDATIONBuild the approved Brain
Alex needs approved priorities, positions and evidence before it can produce defensible work.
5 priorities · 4 positions · 1 sourcesReview Brain →
✓CORE FOUNDATIONAdd the live operating picture
Active Issues and real stakeholder records turn company knowledge into day-to-day Public Affairs work.
6 active Issues · 4 stakeholdersReview workspace →
5CORE FOUNDATIONLet Alex prove the monitoring works
Run the first checks so Alex can keep today’s intelligence current for the team.
I've checked what I can, but a few sources need another look. Any reliable findings remain available.Check the latest sources →
✓CORE FOUNDATIONComplete one real task with Alex
Give Alex a genuine brief and take it through to useful, reviewable work.
1 completed in 30 daysGive Alex another task →
7CORE FOUNDATIONCreate quality evidence
At least one real output should pass Alex QA and be rated useful by a person.
0 QA checks · 1 human ratings · 0 provenTest an output →
+OPTIONALAdd a collaboration channel
Slack or Microsoft can deliver Alex into the team’s workflow, but neither blocks the core pilot.
2 external destinationsManage integrations →
WORKING NOW
Can Alex do the essentials?
These cards show what is ready today and where one more check is needed.
Healthy
Core configuration
ConfiguredOpenAI and Supabase are available to the server.
Attention
Access boundary
PrototypeSecure mode is not enforced.
Attention
Latest monitoring
A few sources need another lookI've checked what I can, but a few sources need another look. Any reliable findings remain available.
Healthy
Tasks completed by Alex
Working reliably1 recent task completed within the expected standard.
Attention
Team updates
2 need attentionOne recent update did not arrive. It is still safely available in Alex.
Waiting
Quality evidence
Baseline neededOne output still needs to pass QA and be rated useful before handover.
M
Microsoft 365 remains parked
View integration →Its state is excluded until a suitable test tenant is available.
QUALITY EVIDENCE
Inspect recent Alex outputs
Alex QA supports judgement. A person still decides whether the work was useful.
RECENT ALEX OUTPUTQUALITY EVIDENCEACTION
Leadership brief · 3 Sept, 12:23Planning Reform: Internal Leadership Brief
Northstar Group plc
QA not checkedHuman rating needed