# Nexon Automation Strategy: ConnectWise + Zofiq Fit

## Executive Summary

Nexon already has a mature automation and AI strategy. Their current MSP operations stack is built around ServiceNow, LogicMonitor, N-central, Genesys, and a LangSmith/Fleet proof of concept for broader agentic workflows. They are not looking for generic AI ideas; they are already investing in AI, hiring for it, and targeting measurable workforce efficiency.

Our strongest play is not to lead with a full rip-and-replace motion. The strongest path is a coexistence motion:

> We keep ServiceNow as the front-end system, use Recursyv to sync selected tickets into ConnectWise PSA, and use Zofiq where its prebuilt PSA agents can reduce triage, classification, dispatch, summary, oversight, and technician-assist effort.

This lets us prove value without asking them to abandon ServiceNow upfront. If the ServiceNow to ConnectWise PSA sync works and Zofiq proves measurable effort reduction, we can use that evidence to open the broader ConnectWise stack conversation.

The fit becomes weaker if we position Zofiq as a replacement for Nexon’s broader LangSmith/Fleet strategy. Based on the Zofiq documentation reviewed, Zofiq does not appear to provide partner-built automatic agents, N-central integration, KB write-back, cross-company PSA analysis in chat, or disclosed model selection.

## Key Scale Indicators

- Ticket volume: approximately 600k-700k tickets per year.
- Monthly ticket volume: approximately 50k-58k tickets per month.
- MSP RMM estate: estimated around 100k endpoints.
- ServiceNow users: approximately 300-400 licensed/active users.
- AI workforce target: 25 roles identified over the next 12 months, with around 15 in core MSP service delivery.
- Zofiq documented agent throughput: up to 10 tickets per minute for ConnectWise agents.

## Their Stack

| Area | Current Tool | Observed Role |
|---|---|---|
| ITSM / CSM | ServiceNow | Core platform for incidents, requests, routing, assignment, auto-close notes, incident summarization, and Now Assist triage. |
| Monitoring | LogicMonitor | Alert suppression, correlation, diagnostic suggestions, root-cause hints, and ServiceNow ticket updates or auto-resolution. |
| RMM | N-central | Current RMM for the MSP fleet. Prior ConnectWise RMM review was dismissed early due to the ServiceNow integration gap. |
| Contact center | Genesys | Part of the service delivery stack, but not deeply discussed for AI workflow replacement. |
| Automation / agents | LangSmith Fleet POC | Planned enterprise agent-building layer for M365, back office, ServiceNow, SharePoint, CRM, and customer reporting workflows. |
| Models | Claude, Copilot, Anthropic via AWS | Frontier-model approach for complex work. Cost and model-quality trade-offs are an active concern. |

## Their Strategy

| Signal | What it means |
|---|---|
| AI program | They already have a modeled business case around profitability, market currency, retention, and workforce replacement over one to two years. |
| Hiring | They are recruiting a team of five for AI. This is not exploratory; they are building internal capability. |
| Workforce target | 25 roles identified over the next 12 months, with about 15 in core MSP service delivery. |
| Automation target | They believe they have already removed about 20% of L1/L2 work and have another 20% targeted. |
| Primary use cases | M365 automations and customer reporting are the two main AI streams already planned. |
| Data approach | They have run analytics across ticket data to identify automation candidates and believe they know what is achievable. |
| Platform posture | They do not want to add platforms unless those platforms solve problems ServiceNow plus Fleet cannot solve. |

## Their AI Layers

| AI Layer | Actual Capabilities | Competitive Read |
|---|---|---|
| ServiceNow Now Assist / AI Agents | Incident summarization, resolution notes, chat/email recommendations, incident assist, categorization, assignment group routing, AI Search, AI Agent Studio workflows, and native ITSM context. | Very strong inside ServiceNow. It is their incumbent workflow layer and already covers triage, assignment, categorization, auto-close notes, and incident summaries. |
| LogicMonitor Edwin AI | Alert deduplication, event correlation, topology-aware enrichment, root-cause analysis, blast-radius analysis, incident summaries, remediation suggestions, and governed automation via existing tools. | Strong for monitoring-originated incidents. It can pre-process noise before tickets hit ServiceNow and update or auto-resolve tickets when telemetry changes. |
| LangSmith / LangChain platform | Agent observability, traces, evals, prompt/version management, human annotation, cost/latency/error monitoring, deployment runtime, human-in-the-loop, multi-agent coordination, and self-hosted options. | Strong as their internal AI engineering control plane. It is not an MSP PSA product by itself; value comes from their team building workflows on top of it. |

## Where Zofiq Helps

| Zofiq Capability | Relevant Detail |
|---|---|
| Copilot Assistant | Embedded ConnectWise pod and standalone app with ticket autofetch, blindspot insights, path to resolution, follow-up prompts, ticket summaries, file/image analysis, web lookup, time entry, KB links, like tickets, and RMM device context. |
| Learning and context | The documentation says agents use historical PSA patterns, current ticket context, SOPs, and configured knowledge sources. Managers can also create AI Assistants and Knowledge Bases by uploading documents. This is contextual learning/configuration, not documented partner-built automatic agent creation. |
| Ticket lifecycle agents | Company, urgency/impact/priority, board/type/subtype/item, contact, agreement, dispatch, and summary-line agents write back to ConnectWise with configured scope and SOP guardrails. |
| Dispatch depth | Can assign owner/team/status, evaluate calendars, in/out board, workload, team membership, historical success, pod mapping documents, and optionally create calendar slots. |
| Control model | Shadow Mode, simulations, workflow logs, reasoning, audit trail visibility, one-run-per-ticket behavior, human-edit conflict halt, and per-agent enablement reduce rollout risk. |
| Ops reporting | Insights Agent gives trend reports; Oversight Agent flags breached or near-breach tickets based on ConnectWise SLA states. |

## ServiceNow To PSA

Recursyv is being explored as the bridge between ServiceNow and ConnectWise PSA. Their role would be to sync selected ServiceNow ticket/data flows into ConnectWise PSA so Nexon can keep ServiceNow as the front-end system while we provide the ConnectWise PSA record Zofiq needs.

This is important because it changes the motion from migration-first to coexistence-first. We can avoid asking Nexon to move off ServiceNow before proving value.

The ServiceNow to ConnectWise PSA mapping needs to define:

- Which ticket types sync.
- Which system owns each field.
- How notes sync.
- How status, priority, assignment, and closure sync.
- What data needs to exist in ConnectWise PSA for Zofiq to classify, route, dispatch, summarize, and report against the ticket.

## Gaps

| Technology | Gap | Why it matters |
|---|---|---|
| ConnectWise PSA / ServiceNow | ServiceNow integration | We must prove tight read/write integration with ServiceNow. Without it, the opportunity stalls. |
| Zofiq | Differentiated workflows | M365 automation and reporting are already covered in their plan; we need use cases beyond generic triage or dispatch. |
| Zofiq / N-central | RMM fit | N-central integration is not documented in the Zofiq KBs reviewed. If Nexon keeps N-central, Zofiq’s documented RMM context features may not apply unless support is confirmed separately. |
| Zofiq | Knowledge workflow | Zofiq can surface documentation and use uploaded KB context, but the reviewed docs do not show KB write-back or automated documentation maintenance. |
| Zofiq / ConnectWise PSA | Cross-company PSA analysis | Copilot appears focused on the current ticket/company context. It should not be positioned as a PSA-wide business analysis tool that can freely compare trends, performance, or operational patterns across all companies in chat. |
| Zofiq | Agent creation | The documentation describes configuring available agents, trigger conditions, outcome controls, and SOPs. It does not describe partners creating their own custom agents that run automatically. |
| Zofiq | Model transparency | The reviewed documentation does not state which LLMs, SLMs, or model providers Zofiq uses for Copilot or Agents. This remains a direct product/security/commercial question. |
| Zofiq | Throughput and burst handling | Zofiq’s documented ConnectWise agent throughput is up to 10 tickets per minute. If a burst exceeds capacity, Zofiq prioritizes newer tickets; older skipped tickets remain untouched and are not automatically retried. |
| Zofiq / ConnectWise PSA | Feedback submission/recovery | If an engineer submits incorrect feedback or a live agent overwrites a PSA field incorrectly, there is no documented native revert button. Recovery should be assumed to require manual correction using PSA history/logs. |
| Zofiq / ConnectWise Platform | Data residency | Australian data residency and processing are explicit requirements. APAC/Australia availability must be evidenced clearly. |
| Zofiq | Quality vs cost | They believe frontier models are worth paying for on complex work and cheaper models belong on simpler work. We need to be clear where Zofiq fits. |
| Zofiq | Token economics | Their token bill is rising. Zofiq’s per-ticket pricing can be a commercial counterpoint if outcomes are comparable. |
| ConnectWise RMM | RMM replacement | N-central displacement is a separate play, but ServiceNow integration is the critical unlock for revisiting it. |

## Recommended Positioning

We should position this as:

> ConnectWise can meet Nexon where they are. Keep ServiceNow as the operational front door, use Recursyv to sync selected ticket flows into ConnectWise PSA, and prove Zofiq on the PSA workflows it is documented to support.

We should avoid positioning Zofiq as:

- A replacement for LangSmith/Fleet.
- A partner-extensible automatic agent-building platform.
- A cross-company PSA analytics tool.
- A KB maintenance/write-back tool.
- An N-central-integrated automation layer unless Product confirms support.
- A general autonomous MCP-style platform.

## Recommended Next Step

The next internal step should be to validate three things before taking a strong message back to Nexon:

1. Whether Recursyv can reliably sync the selected ServiceNow ticket workflows into ConnectWise PSA with the fields Zofiq requires.
2. Whether Zofiq’s documented 10 tickets/minute throughput and burst behavior is acceptable for Nexon’s real ticket arrival patterns.
3. Whether Product can confirm the open Zofiq questions around model provider, data residency, N-central support, feedback/recovery, and any enterprise throughput options.

If those checks come back positively, we have a credible coexistence-led path. If not, the opportunity should stay focused on specific ConnectWise stack areas rather than overextending the Zofiq story.
