Context Sidecar

Zero-click ERP-embedded intelligence — the canonical implementation of Embedded Intelligence and one of the four named differentiators of SuiteCentral 2.0.

What it does

Per 01-executive-summary:

• Embeds inside NetSuite and Business Central (the ERP UI itself, not a separate console)
• Supports cross-system action execution
• Surfaces “evidence and reasoning traces” so the AI’s work is inspectable

The slide script names it “Context-Aware Sidecar” in slide 2 (the differentiation list) and describes the embedded behavior in slide 5 (the Embedded Intelligence pitch).

Per ai-governance-layer-video (01:15-01:32) — the hook video confirms and extends this description:

“Our killer app is the Context-Aware Sidecar. It is non-intrusively embedded directly into the ERP workflow, providing zero-click intelligence. The Sidecar auto-detects the user’s context and surfaces sync status, risk analysis, and related documents without breaking the flow.”

Three specific things the sidecar surfaces automatically, without the user taking any action:

1. Sync status — whether the record the user is looking at is current / out-of-date / pending sync with the other ERP
2. Risk analysis — AI-generated risk assessment for the record or transaction in context
3. Related documents — auto-discovered document linkage to contracts, attachments, or supporting records

The “zero-click” framing and the “auto-detects context” claim are the most differentiating capabilities: the sidecar is not a tool the user has to invoke — it is an ambient intelligence surface that appears next to whatever record the user happens to be viewing in the ERP. This is the concrete architectural expression of the Embedded Intelligence principle.

The NetSuite AP workflow pattern (per narration-scripts)

The Watch-track demo video for the Context Sidecar runs inside a NetSuite Accounts Payable workflow — not a synthetic demo environment. This is important: reviewers see the sidecar working against the exact UI their users will see in production.

The narration walks through the sidecar’s behavior across four record contexts in sequence:

1. Vendor context — When the user focuses on a vendor record, the sidecar auto-refreshes and opens with risk score, compliance alerts, and action-ready recommendations. The narration explicitly frames this as co-locating risk and action: “This shortens time from detection to mitigation because risk and action are co-located.”
2. Customer context — Same pattern as vendor, but scoped to account health and receivables.
3. Invoice context — Same pattern, extending the “multi-record intelligence” story.
4. Purchase order context — Surfaces delay exposure “before downstream schedules are impacted” — proactive operational signal, not reactive reporting.

Two additional behaviors surface in the narration:

• High-risk states expose “stronger controls, including pause-payments mitigation” — the sidecar can actively intervene in high-risk scenarios with one-click controls, not just flag them. The highlight video reinforces this: “High-risk conditions expose one-click controls like pause payments.”
• Debug panel confirms the underlying message flow so technical reviewers can verify exactly what the sidecar did and why.

The narration closes with: “Outcome: zero-click intelligence with explicit provenance, explainability, and human approval.” — mapping directly to the four runtime governance capabilities from read-elevator-pitch Beat 2 (explain, confidence, hallucination detection, human approval).

Data provenance and cross-ERP consistency

Per narration-scripts context-sidecar cues [35] and [56]:

“Data provenance stays explicit, including service-backed responses and fallback behavior.” “The same governed interaction model runs across NetSuite and Business Central.”

Three implications:

1. The sidecar distinguishes service-backed from fallback data sources and makes the distinction visible to the reviewer. This is the auditability requirement made concrete — users can see when the sidecar is showing real-time data vs cached vs mocked.
2. Cross-ERP parity is explicit: the same interaction model runs on NetSuite and Business Central. This is third-source confirmation of the dual-ERP “equal citizens” claim (also in read-talking-points and ai-governance-layer-video).
3. MCP discovery and call paths are policy-scoped (per cue [60]). This ties the sidecar to the broader MCP control-plane architecture documented in suitecentral-2-overview.

AI insights and trend signals

Per narration-scripts cue [50]:

“AI insights summarize trend signals like lead-time movement and open purchase exposure.”

The sidecar also surfaces AI-generated trend analysis — two specific examples from the narration: lead-time movement (supply chain velocity changes) and open purchase exposure (commitment-but-not-yet-fulfilled tracking). These are higher-level than simple record-level risk signals; they synthesize across many records to surface operational patterns.

Two video variants

Per narration-scripts cue [72]:

“Use this full cut for due diligence, and the highlight cut for fast executive review.”

There are two Context Sidecar demo videos in the Watch track, serving different review purposes:

• context-sidecar (the full cut, ~72 seconds of narration) — for technical due diligence. Walks through all four record contexts, the debug panel, data provenance, action dispatch, AI insights, and cross-ERP consistency.
• context-sidecar-highlight (the 90-second executive cut, ~24 seconds of narration) — for fast executive review. Shows the one-click context load, customer/invoice context changes, high-risk one-click controls (pause payments), and cross-ERP parity — ending with “zero-click intelligence with governance and policy controls.”

This two-variant framing is notable: the product explicitly serves both Path A executive reviewers and Path C deep technical reviewers from the same Watch track (see three-review-paths).

Why it matters (to the adoption case)

This is the “killer app” of SuiteCentral 2.0 (per the Preston-Test README.md’s framing). It is the most concrete answer to the question “what does SuiteCentral 2.0 actually do that competitors don’t?” — the architectural difference between sitting inside the ERP vs sitting next to it.

For Squire’s audit-heavy enterprise client base, the “evidence and reasoning traces” claim is also load-bearing: regulated industries cannot deploy black-box AI. The Context Sidecar’s transparency is what makes it deployable into Squire’s clients’ workflows.

Status

• Tier: production-ready
• Test coverage: production-validated (per slide 4 production proof in 01-executive-summary)
• Maturity: confirmed shipped (Tier 2 in February 2026 per the Preston-Test README.md — not yet formally ingested as a source-summary)
• Dashboard URL: none — by design, the sidecar lives inside the ERP, not as a standalone dashboard

What’s NOT yet known (from this source)

This source mentions the Context Sidecar by name but doesn’t go into:

• Implementation details (how the iframe embed works, how MCP plugs in)
• Per-action capability (which “cross-system actions” are live)
• Reasoning trace format
• Security/data-flow architecture

For these, ingest sources like:

• MCP Gateway Architecture
• AI Reasoning Traces Guide
• context-sidecar-video-script.md (in the notebook)

Sources

• 01-executive-summary — claims 5, 8 (named as “Context-Aware Sidecar”; Embedded Intelligence framing)
• ai-governance-layer-video — claim 11 (zero-click intelligence, auto-detect context, three surfaces: sync status / risk analysis / related documents)
• narration-scripts — claims 3-15 (NetSuite AP workflow pattern, vendor/customer/invoice/PO contexts, pause-payments mitigation, debug panel, data provenance, cross-ERP parity, MCP policy-scoping, AI trend signals, full-cut vs highlight-cut framing)