Good morning, Anton.
While you were away, CognAItive triaged overnight. Here's what's waiting — start anywhere.
Reactivate Auto R&D
Chat with Ahsas — new idea
Push a meeting recording
Draft a memo for Kemper
New here? Meet CognAItive — 3-min interview
Overnight triage
since 18:42 yesterday
0 need reply
23 emails arrived · 16 handled by rules
→ CognAItive drafted 3 replies
0 open
Outlook tasks · 1 overdue, 2 due today
→ 2 can be auto-closed
0 chats
Need attention · Ahsas, Wayne, +3
→ 1 is blocking Auto R&D
0 waiting
Your approval · memo + DevOps task
→ from Rewards Riches
Needs your attention
ALERT — bonus redemption down 18%incident · ALR-0291
Tableau watchdog fired at 03:04. Ontario, sustained since 02:00. CognAItive can open an incident project and pull the stats.
03:04
Ahsas has a new idea1:1 · unread
“had a thought about radar drift and retention offers” — could be project-worthy. CognAItive is listening.
07:41
Auto R&D re-raised by Kemperdormant 112d
Kemper asked about it in standup. CognAItive can rehydrate the full project history instantly.
07:14
Rewards Riches memo awaiting your sign-offcreate→curate→approve
CognAItive drafted the progress memo for Kemper. Review and approve before it sends.
06:50
Yesterday's Auto R&D sync — recording readymeeting ingest
Push it into the project for a structured Decisions / Actions / Pending breakdown with collective sign-off.
18:02
Presenter's guide
Drive the demo cold. Every flow, how to trigger it, what to say — and a Launch button that runs it live.
The one idea to land
CognAItive is a governance and memory layer for AI at work: the agent is replaceable muscle, but the defensible product is organisational state — memory with provenance, an org chart that routes authority and access, and autonomy earned step-by-step on evidence. Every flow is an angle on that.
⚙ AI-discovered
⚑ human-flagged
✓ human-approved gate
Two personas: Anton (analyst) · Kemper (exec) — top-right avatar
How to use: Click Launch on any card to run that flow live (it sets up the right context for you). Purple buttons switch you to Kemper first. Return here anytime via the guide icon in the rail.
Skills Library
Knowledge you can execute, not just retrieve. Every taught procedure becomes a versioned skill any agent can run — with provenance.
6skills
100%with provenance
14runs this week
3projects reusing skills
Teach CognAItive a skill
Describe once — or let CognAItive propose one from what it observes. Three ways skills are born: taught, correction-learned, behaviour-mined — all confirmed by a human.
Journey table mapping
v3
Non-obvious table name → source of truth for points crediting. Join paths, gotchas, and the columns that lie.
reads: SQL · Journey,bonus_outcomesno write · no PII
View file
History
Run
last run 2h ago
DBA request logging
v5
The full SOP for logging DBA work in Azure DevOps — fields, priority logic, linking back to projects.
writes: Azure DevOps (scoped)no prod DB
View file
History
Run
last run today
Memo distillation — Kemper format
v2
How Kemper likes memos: status → what changed → decisions → asks. Learned from 11 approved memos.
View file
History
Run
last run today
Meeting ingest — 3-part split
v4
Decisions / Actions / Pending breakdown with owner suggestion from org chart + who spoke, conflict check vs memory.
View file
History
Run
last run yesterday
Radar drift scoring
v0 · draft
Drift-velocity calculation over hourly radar windows. Learning from Tiona's analysis in Radar Drift Offers.
View file
Run
draft — not yet approved
Agent Registry
Every agent has an identity, an owner, scoped permissions with expiry, and a kill switch. Industry average: 50% of agents run ungoverned. Here: zero.
6agents registered
6governed · audited
0shadow agents
97.2%eval pass rate
0drift alerts
TRG
Overnight Triage Agent agt-001owner: Anton Reyes
Sorts inbound email, tasks, chats and approvals before you log in. Drafts replies; never sends.
outlook:readoutlook:draft ⏱ 90dmemory:read
healthy · eval 98.1%
active
audit trail ↗
RTV
Deep Retrieval Agent agt-002owner: Anton Reyes
Rehydrates projects across 6 source systems. Permission-aware — respects the three access tiers at retrieval time.
drive:readsql:readtranscripts:readtier-2: flag-only
healthy · eval 97.4%
active
audit trail ↗
MMO
Memo Drafter agt-003owner: Anton Reyes
Distills project memory into exec memos. Cannot send — output always routes to the approval queue.
memory:readdraft-onlyskill: memo-kemper-v2
healthy · eval 96.8%
active
audit trail ↗
DVO
DevOps Skill Agent agt-004owner: Wayne Goss
Logs and tracks Azure DevOps work items. Scoped to its own identity — not an inherited copy of Wayne's access.
devops:create ⏱ 60ddevops:readown identity
healthy · eval 97.9%
active
audit trail ↗
ING
Meeting Ingest Agent agt-005owner: Ahsas Ranjan
Transcribes, diarises and splits recordings into the 3-part structure. Output locked only via collective sign-off.
recordings:readmemory:propose
healthy · eval 95.6%
active
audit trail ↗
DRF
Drift Scorer agt-006 · sandboxedowner: Tiona Welsh
New agent for Radar Drift Offers. Running in sandbox until its eval suite passes 3 consecutive weekly runs.
sql:read (radar_* only) ⏱ 30dno-writesandbox
probation · eval 91.2%
active
audit trail ↗
Playbooks
Autonomy is earned per step, not granted per agent.
A workflow is the human process — how the work actually gets done today. A playbook is the codified version of that workflow — chained skills with policy gates that CognAItive can run. You bring the workflow; CognAItive turns it into a playbook.
Skills → Playbooks → Agents. The trio that future-proofs this: a skill is one procedure, a playbook chains them with gates, and an agent runs playbooks under a scoped identity — the foundation for agents running agents. The org chart also defines each agent\'s persona — who it acts as, what it may touch, whose authority it inherits.
OBSERVEwatches, builds the trace→
DRAFTdoes it, you approve every time→
ACTauto-proceeds unless policy triggers→
AUTOend-to-end, audit trail only
Meeting → Tracked Work → Exec Memo
5 steps · 3 autonomous · 1 act-w/exceptions · 1 gated · runs after every project sync
Ingest meeting
agt-005 · 3-part split
AUTOeval 95.6%
Match vs DevOps
agt-004 · dedupe items
AUTOeval 97.9%
Log new tasks
agt-004 · devops.create
ACTeval 97.9%
Draft exec memo
agt-003 · memo-kemper-v2
AUTOeval 96.8%
Send memo
approval gate · you
DRAFT14/14 approved
Playbook library
Your org has thousands. Find the ones that are yours, your team\'s, or adjacent to your work — or search across all of them.
Workflow Intelligence
CognAItive watched a week of real work and found the patterns worth automating.
5candidates this week
14.5hmanual hrs / wk identified
2quick wins
3cross-dept matches
To action 5
Cross-department 3
Grouped by what CognAItive is proposing. Quick wins pay back immediately; skills and workflows are reusable builds; gaps need a missing integration; emerging patterns are trending toward routine.
Quick wins
2h
/ wk
Monday data-quality checks across 3 reporting portals
You + 2 teammates run the same check sequence every Monday morning.
you +2every Mondaydetected: tool logs + chat
Suggestion: the
sql.query skill already covers 80% of this. One playbook closes the gap.Reusable skills
2h
/ wk
Manual VIP cohort exports for Tiona
The same filtered export, requested in chat 7 times this month — reworded each time.
Tiona7× this monthdetected: chat patterns
Suggestion: a single parameterised skill — “vip-cohort-export”.
Multi-step workflows
1.5h
/ wk
Friday status email assembled from DevOps items
Ahsas pulls the week's work items and summarises them by project, by hand.
Ahsas9 wks runningdetected: integration logs
Suggestion: multi-step workflow with an approval gate —
devops.pull → summarise → draft.Skill gaps
1.5h
/ wk
Grafana alert context lookups
DevOps-adjacent chats reference checking Grafana manually before acting.
Wayne, Felixdetected: chat + missing integration
Suggestion: flagged to the platform team — a Grafana connector would close this.
Emerging
3×
/ 2 wks
“Drift summary” report — asked 3× in 2 weeks
Not yet routine, but trending that way since Radar Drift Offers launched.
Anton, Tionadetected: trend detection
CognAItive matches abstracted pattern descriptions across departments — so two teams solving the same problem can share one build. No raw content crosses team boundaries.
BA×Customer Success
94% matchScheduled data validation with exception flagging
Your Monday portal checks and CS's player-transaction reviews are the same shape — both could run on one validation playbook.
abstracted pattern only — no raw content crossed departments
Marketing×Finance
88% matchMulti-source aggregation into a formatted report
Campaign performance decks (4 sources) and monthly cost reports follow an identical assembly pattern.
abstracted pattern only
Call Centre×DevOps
81% matchContext-enriched escalation routing
Fast-fact sheet → email L2 with context, and Grafana check → route to on-call: the same enrichment-then-route shape.
abstracted pattern only
Org chart
Not a diagram — infrastructure. Authority, content access and agent ownership all route through it.
radar data
bonus outcomes
retention offers
incident response
ontario growth
Knowledge layer
Memory beyond any single project — shared entities, domain owners, and the company soup. Project-scoped memory lives in each project's panel.
Cross-project
Company
Usage & cost
The FinOps layer for your agents — what's spent, by whom, on what, with budgets and alerts. No surprise bills, no ungoverned spend.
This is the platform’s track record — not an efficiency score, but a ledger of initiatives it has actually moved. The honest answer to “what has it done for us?”
Knowledge Risk
The expertise graph, inverted. Not “who knows X” — where is the company one resignation away from losing X?
2 critical single points of failure. Two topics are held in exactly one person's head with no backup — if either leaves, the knowledge is at High resolution risk — hard to recover. Capture it while they're here.
2critical · bus factor 1
3at-risk · bus factor 2
1resilient · bus factor 3+
6topics scored
Knowledge moments — when risk peaks
Someone is leaving
CAPTURE NOW
Wayne Goss · 9 days notice. The highest-value moment for a knowledge system. Run a targeted capture of everything only he holds, so it stays with the team.
Capture the handover
Someone is joining
REHYDRATE
Sam Rivera · new DBA, day one. Onboarding as role-rehydration — instead of weeks reconstructing what their predecessor knew, they inherit it now.
Rehydrate the role
Scored from the same behavioural signals as the expertise graph — decisions authored, facts corrected, skills taught, sources owned. Tap a card for the evidence; tap Capture to turn one head's knowledge into the team's.
Team flows
Manager view. CognAItive surfaces automation candidates — and so can you, from what you see that the AI can't.
All reports
Heavy load
Has unflagged patterns
AI discovery (⚙) plus your own observation (⚑) covers both blind spots — an employee being too manual often doesn\'t know help exists, and won\'t ask.
A
Auto R&D
Skills: DevOps · Drive · SQL
CognAItive creates · retrieves · drafts — you curate · approve · send. Nothing leaves CognAItive without your sign-off.
Project memory
✕
PROJECT MEMORY
ACTION ITEMS
DECISIONS