AI Operations · Executive Summary
State of Curacel's AI stack across all departments — agents, automations, coverage, and open risks.
Live Agents
10
In production across 7 departments
↑ 4 added in Q1 2026
In Development
5
Agents currently being built
1 overdue · 1 status unclear
n8n Automations
28
Dept-owned workflows
6 inactive · 14 no backup owner
Dept AI Coverage
7/8
Departments with live AI touch
Finance has 0 live agents
Finance Expense Approval workflow is inactive. The only finance automation (n8n #21) is down with no backup owner. Weekly reimbursements are currently processed manually.
14 workflows have a single point of failure. If the owner leaves or is unavailable, the workflow breaks. No handover exists for these processes.
Automated QA & Testing agent is overdue since January 31. No documented blocker. No revised timeline. Engineering has no automated QA coverage.
Q1 2026 was the strongest quarter for AI Ops delivery. 4 new agents shipped, 5 new CS workflows added, and the Meeting Notes Agent was stabilised for reliable channel routing.
Agent Pipeline at a Glance
Live in production
10
In development
5
Ideas / Research
6
Total agents across all stages
21
n8n Workflow Health
Active workflows
22
Inactive / stopped
6
No backup owner
14
People Ops owns 14 of 28 workflows — highest concentration, highest inactive rate (5 of 14).
Estimated Business Impact (Current Stack)
Based on audit estimates — not yet formally tracked
~320
Hours saved / month
across live agents
across live agents
40+
KPIs auto-populated
every sprint
every sprint
10+
Teams using AI-assisted
processes daily
processes daily
~60%
Recruitment steps
now AI-assisted
now AI-assisted
Agent Health
Status, SLA compliance, and quality assessment for all 10 live agents. 5 in-development agents also tracked.
How to read this page
"Health" reflects uptime reliability and output quality based on the audit. "SLA" indicates whether the agent is meeting its expected delivery standards. Formal SLA tracking is not yet in place — establishing it is a Day-1 priority.
Live Agents (10)
In Production
| Agent | Department | Primary Function | Builder | Health | SLA Status |
|---|---|---|---|---|---|
AuntyPelz AI HR Slack Chatbot |
People Ops | Self-service HR queries via OpenAI embeddings. Onboarding DMs for new joiners. | SA Samuel | ✅ Healthy | Meeting |
Fireflies → HubSpot CRM Intelligence |
Commercial | Call transcripts → MEDDIC scoring → HubSpot deal updates. 3-layer deduplication. | SA Samuel | ✅ Healthy | Meeting |
KPI Autopopulation Sprint KPI Auto-fill |
All Teams | Pulls 40+ KPIs from Jira, Nova, Google Sheets across 10+ teams each sprint. | SA Samuel | ⚡ Imminent | Due 14 May |
Interview Assessment AI Hiring Evaluation |
People Ops | CV upload → AI scoring → text/audio responses → 3-tier evaluation for hiring panels. | SA Samuel + Rotimi | ✅ Healthy | Meeting |
Sales Prospecting Auto Outreach Pipeline |
Commercial | 8-step pipeline: ICP → Apollo → HubSpot dedup → GPT-4 messaging → Gmail campaigns. | SA Samuel | ✅ Healthy | Meeting |
Meeting Notes Agent Slack Routing |
All Teams | Fireflies recordings → AI summaries → routed to correct Slack channels. | DE Demilade | ✅ Stable | Meeting |
Claims Vetting Reminders SLA Monitoring |
Customer Success | Logs into insurer portals → scrapes pending claims → TAT urgency alerts via Slack. | SA Samuel | ✅ Healthy | Meeting |
Compliance Agent PDPL Monitoring |
Legal / Ops | Continuously monitors Curacel operations for PDPL compliance issues. | EA Emmanuel A | ✅ Healthy | Meeting |
AI Competency Assessment Skills Evaluation |
People Ops | Structured AI skills tests → Chart.js scorecards → org-wide admin view. | RT Rotimi | ✅ Healthy | Meeting |
AI-Interviewer Candidate Interviews |
People Ops | Conducts structured candidate interviews autonomously. In active use with People Ops. | FI Fikayo | ✅ In Use | Meeting |
In Development (5)
Building
| Agent | Department | Builder | Status | Risk |
|---|---|---|---|---|
360 Feedback Generator |
People Ops | DE Demilade | In Dev | No delivery date set |
Automated QA & Testing |
Engineering | SA Samuel | Overdue | Was due Jan 31 — no update |
Claims Browsing Agent |
Customer Success | FI Fikayo | Status Unclear | May not have been coded — needs verification |
OneHub (SSO Portal) |
All Teams | CU Curacel Team | Core Shipped | Not yet deployed to production |
LLM Financial Reporting Bot |
Finance | EA Emmanuel A | Repo Exists | Not yet started in earnest — high value |
SLA gap
No formal SLA framework exists for any live agent. There is no defined uptime target, no quality review cadence, and no escalation path when an agent fails or produces poor output. This is the most significant governance gap in the current stack.
n8n Workflow Repository
28 department-owned automations. These are built and maintained by departments — AI Ops provides oversight and governance.
Active
22
Running workflows
Inactive
6
Stopped or deprecated
No Backup Owner
14
Single point of failure
Newest Batch
Q1 '26
Customer Success (5 workflows)
Finance Expense Approval (#21) — INACTIVE, no backup owner. Victoria is the sole owner. Weekly reimbursements are now manual. This is the highest-priority workflow fix.
All 28 Workflows
Sorted by department
People Ops — 14 workflows (5 inactive)
| Workflow | Owner | Backup | Date | Status |
|---|---|---|---|---|
| Workshop Flier Automation | Oreoluwa | ✓ | Oct 2025 | Active |
| Slack Notification Alerts | Oreoluwa | ✓ | Oct 2025 | Active |
| Birthday Celebration | Oreoluwa | ✓ | Nov 2025 | Active |
| Double Click Notification | Oreoluwa | ✓ | Jan 2026 | Active |
| Monthly 1:1 Reminders | Precious | None | Jan 2026 | Active |
| Talent Acquisition | Pelumi | None | Oct 2025 | Active |
| Exec Ops (recruitment) | Pelumi | None | Oct 2025 | Active |
| Fullstack Engineer (recruitment) | Pelumi | None | Dec 2025 | Active |
| Senior PM (active) | Pelumi | None | Dec 2025 | Active |
| Account Executive | Pelumi | None | Nov 2025 | Active |
| Xlabs | Pelumi | None | Oct 2025 | Inactive |
| Exec 2.0 | Pelumi | None | Oct 2025 | Inactive |
| Senior PM (inactive) | Pelumi | None | Dec 2025 | Inactive |
| Pulse Check | Precious | None | Dec 2025 | Inactive |
Customer Success — 5 workflows (all Q1 2026) | ||||
| Enrollee Data Formatting | Muslimah | ✓ | Jan 2026 | Active |
| First Time to Value Tracker | Tracy/Joseph | None | Mar 2026 | Active |
| Client Birthday Messages | Joseph | None | Mar 2026 | Active |
| NPS Detractor Messages | Joseph | None | Nov 2025 | Active |
| Client Onboarding Tracker | Tracy | None | Mar 2026 | Active |
Engineering — 3 workflows | ||||
| NudgeBot (Jira stale tickets) | Kolade | ✓ | Dec 2025 | Active |
| Weekly Engineering Highlights | Rotimi | None | — | Active |
| PeopleOps (Eng-built) | Pelumi | None | Nov 2025 | Inactive |
Sales / Marketing — 2 workflows | ||||
| Socials to Slack | Nimi | ✓ | Nov 2025 | Active |
| Inbound Leads Email | Nimi | ✓ | Oct 2025 | Active |
Data Ops / Product / Finance | ||||
| Defmis Workflow (claims quality) | Sophie | None | Jan 2026 | Active |
| Metabase → Google Sheet | Jesutoni | None | Feb 2026 | Active |
| Expense Approval 🔴 | Victoria | None | Dec 2025 | INACTIVE |
AI Coverage by Department
Where AI is embedded, where it is missing, and where the next investments should go.
💼 CommercialWell Covered
Live agents3
Sales Prospecting Auto✅ Live
Fireflies → HubSpot✅ Live (MEDDIC scoring)
AI-Interviewer✅ Live
n8n automations2 (Socials, Inbound Leads)
Next priorityAE/SDR Assistant (idea)
🎯 Customer SuccessPartial
Live agents1
Claims Vetting Reminders✅ Live
Claims Browsing Agent⚠️ Status unclear
n8n automations5 (newest batch — Q1 2026)
Backup coverage4 of 5 workflows have no backup
👥 People OpsMost Active
Live agents3
AuntyPelz AI✅ HR chatbot
AI Competency Assessment✅ Org-wide
360 Feedback Generator🔨 In development
n8n automations14 total — 5 inactive
🔧 EngineeringPartial
Live agents2
Interview Assessment✅ Live
AI Competency Assessment✅ Live
Automated QA & TestingOverdue since Jan 31
n8n automations3 (NudgeBot, Highlights, 1 inactive)
💰 FinanceNot Covered
Live agents0
Expense Approval (n8n #21)🔴 INACTIVE
LLM Financial Reporting BotRepo exists — not started
KPI Autopopulation✅ Includes Finance KPIs
Priority actionReactivate Expense Approval immediately
📊 Data OpsPartial
Live agents1
Claims Vetting Reminders✅ Cross-team benefit
Defmis Workflow (n8n)✅ Claims data quality
Metabase → Sheets (n8n)✅ Daily sync
Backup riskDefmis has no backup owner
📱 ProductLight
Live agentsIndirect — 1
KPI Autopopulation✅ Includes Product KPIs via Nova
Metabase sync (n8n)✅ Daily
Next priorityFeedback synthesis agent (idea)
⚙️ OperationsCovered
Live agents1
Meeting Notes Agent✅ Slack routing (all teams)
Curvis Command Centre💡 Idea — high value
OKR Goal Setting Tool💡 Cross-dept idea
Usage & Adoption
How actively departments are using AI tools — and where adoption is lagging.
Note on data
Formal usage metrics are not yet tracked centrally. These figures are based on audit findings, agent architecture, and team-reported usage. Establishing a usage tracking baseline is a recommended near-term action.
Estimated Weekly Active Users
~85
Across all live agents and n8n automations
Automated Touchpoints / Month
~1,200
Slack messages, CRM updates, KPI writes, alerts
Depts with Active Daily AI Use
5 / 8
Finance, Product, and partial CS are lagging
Adoption by Department
Estimated — based on audit
Commercial
82%
People Ops
78%
Engineering
60%
Customer Success
55%
Data Ops
50%
Product
35%
Finance
10%
Operations
65%
Agents by Usage Frequency
| Agent | Frequency | Users |
|---|---|---|
| KPI Autopopulation | Every sprint (bi-weekly) | 10+ teams |
| Meeting Notes Agent | Every meeting with transcript | All teams |
| AuntyPelz AI | Daily (on-demand) | All staff |
| Claims Vetting Reminders | Daily (scheduled) | CS + Claims |
| Fireflies → HubSpot | Daily (post-call) | Commercial |
| Sales Prospecting Auto | Weekly campaigns | Commercial |
| Compliance Agent | Continuous (background) | Legal/Ops |
| Interview Assessment | Per hiring cycle | People Ops |
| AI-Interviewer | Per hiring cycle | People Ops |
| AI Competency Assessment | One-time + periodic | All staff |
Where Adoption is Falling Behind
Finance — Zero live agents. Only workflow is inactive. No AI-assisted process exists.
Product — Light usage. KPI autopopulation reaches Product KPIs but no dedicated agent exists for product decisions or feedback.
Customer Success — 5 workflows added in Q1 2026 but backup ownership is not in place. Risk of breakage if anyone leaves.
Data Ops — Good coverage but no analytics-specific agent. Metabase is used manually alongside the daily sheet sync.
Risks & Gaps
Prioritised issues that require leadership attention or operational decisions.
🔴 Critical — Needs immediate action
Finance Expense Approval workflow is down
n8n #21 has been inactive with no backup owner. Victoria is the sole owner. Weekly reimbursements are manual. This is the highest-urgency fix — it is simple to reactivate and the business impact is immediate.
No SLA framework exists for any live agent
There is no defined uptime standard, no output quality review process, and no escalation path when an agent fails or degrades. The 10 live agents are running without formal performance accountability.
Automated QA & Testing agent is overdue with no update
Was due January 31. Now overdue by over 3 months with no documented blocker and no revised timeline. Engineering has no automated QA coverage as a result.
🟡 High — Address within 30 days
14 workflows have no backup owner
If any single owner goes on leave or leaves the company, 14 automated processes break with no recovery plan. Includes all of Pelumi's recruitment flows, Precious's 1:1 reminders, Sophie's claims quality workflow, and all 4 of Joseph and Tracy's CS workflows.
Claims Browsing Agent — actual status unknown
Listed as "in development" with Fikayo as builder, but the audit note reads "may not have been coded." This agent appears on the roadmap and in stakeholder communications but may not exist. Needs immediate verification before it is referenced further.
Samuel Bamgbola is a single point of failure for AI Ops delivery
Samuel owns 6 of the 10 live agents. If he becomes unavailable, a significant portion of AI Ops production capacity is lost. The team needs a second builder at his level, or more structured knowledge transfer.
5 People Ops workflows are inactive with no decision made
Xlabs, Exec 2.0, Senior PM (old), PeopleOps (Eng-built), and Pulse Check are all inactive. No record of why they were stopped or whether they should be revived. These create clutter and ambiguity in the workflow inventory.
🔵 Strategic — Worth leadership visibility
Finance has no AI agent — the largest gap by department
Finance is the only department with zero live agents and a broken automation. The LLM Financial Reporting Bot has a repository but has not been developed. This is a high-value gap given the reporting burden on the finance team.
No centralised AI Ops command centre (Curvis)
There is no single place to view the health, output, and status of all agents. Monitoring is ad hoc. As the agent count grows, this becomes increasingly untenable. Curvis — the proposed AI Ops Command Centre — addresses this directly.
AI adoption is not being measured
There is no usage tracking, no quality feedback loop, and no way to demonstrate ROI to leadership. Without measurement, it is impossible to know whether agents are being used as intended or delivering value.
Projections
Where AI Ops is heading — near-term delivery outlook and estimated impact over the next 90 days.
Basis for these projections
Figures are based on current pipeline status, builder capacity, and agent complexity estimates from the audit. They assume the AI Ops team is resourced and a Head of AI Ops is in place to drive delivery.
Agents in Production · 90 Days
14–16
From 10 today — if pipeline unblocked
Estimated Hours Saved · Monthly
~480
Up from ~320 today with 4 new agents
Departments Fully Covered
7–8 / 8
Finance gap closed if LLM bot ships
30-Day Outlook
Near Term| Action | Expected Outcome |
|---|---|
| Reactivate Finance Expense Approval | Manual approvals eliminated, finance team unblocked |
| KPI Autopopulation deployed (14 May) | Sprint reporting fully automated across 10+ teams |
| Backup owners assigned to 14 workflows | Single points of failure removed |
| Claims Browsing Agent status confirmed | Roadmap accuracy restored — remove or restart |
| SLA framework drafted | Agents have defined performance standards for first time |
60–90 Day Outlook
Planned| Agent / Initiative | Impact |
|---|---|
| LLM Financial Reporting Bot | Finance gets first AI Ops agent — closes the coverage gap |
| AE/SDR Assistant | Commercial team's most-requested tool — sales productivity lift |
| 360 Feedback Generator (Demilade) | People Ops gains AI-assisted performance management |
| OneHub deployed to production | Unified SSO for all internal tools — access control simplified |
| Curvis v0.1 prototype | Leadership gets first view of centralised AI Ops monitoring |
If Nothing Changes — Downside Scenario
Finance remains without AI support. Manual processes persist. Overhead grows as the team scales.
14 backup-less workflows continue to be single points of failure. One person leaving creates an outage.
Automated QA agent remains undelivered. Engineering carries manual testing overhead into H2 2026.
Without an SLA framework, agent quality degradation goes undetected until it causes a visible failure.
Without centralised monitoring (Curvis), AI Ops oversight remains ad hoc as the stack scales to 15+ agents.