AI Features in TekReady

AI Features in TekReady

TekReady ITSM includes three AI-assisted workflows that take the manual grunt-work out of ITSM configuration. All three are opt-in, preview-only (nothing is auto-applied without your explicit confirmation), and aware of your engagement’s own option lists — the AI works with the SLAs, Impacts and Urgencies you’ve defined for this customer, not generic templates.

Prerequisites

The AI features only appear when an administrator has:

1. Enabled the AI master switch in Settings → AI.

2. Configured a provider (OpenAI, Anthropic, or a local Ollama-compatible model) with a valid API key.

When the switch is off, the AI buttons simply don’t render — the non-AI flows still work as normal.

Where the AI is available

The priority-suggestion and item-generation features live on the Incident Categories and Request Categories pages — the two longest, most repetitive sections of an ITSM build; both pages expose the same three category features below. AI is now also available on the Workflows section, where AI workflow assistance helps you build out the lifecycle steps.

Feature 1 — ✨ Suggest priorities (per row)

Sits on every row in the actions column. Useful when you’ve added or imported a row and want the AI to pick the right SLA / Impact / Urgency for it.

How it works. Click the ✨ icon next to a row. A preview modal opens showing the row’s current values alongside the AI’s suggested values, plus a one-line rationale. Each suggested value is a dropdown — you can tweak any of them before clicking Apply selected. Nothing is saved until you click Apply.

Feature 2 — ✨ Suggest priorities (bulk)

Top-of-table button. Use this when a fresh batch of rows has just landed (for example after loading the defaults or importing from a customer spreadsheet) and many of them have empty SLA / Impact / Urgency fields.

How it works. Click ✨ Suggest priorities. The AI inspects every row that’s missing at least one of the three fields and returns a suggestion for each. The preview modal lists them in a table; you can review, edit the dropdowns, untick rows you want to skip, then apply.

Anchoring. The AI is given each option’s rank (Impact and Urgency carry a numeric rank) and each SLA’s resolution-time hours, so it reasons about severity in your terms. For example, an “Cannot sign in” row gets high impact + high urgency + shortest SLA, while “Cannot share screen” gets moderate values because a workaround exists.

Feature 3 — ✨ Suggest items (topic seed)

Top-of-table button, sits next to + Add Row. Use this to brainstorm the catalogue rows you need for a new topic without typing them all from scratch.

How it works. Click ✨ Suggest items and type a topic — a product, system or capability. Examples:

• Intune

• Microsoft 365 email

• VPN access

• Adobe Creative Cloud

• Joiner / mover / leaver

The AI returns about ten candidate rows, each with:

• Category / Type / Item — the three-level classification. The AI prefers to re-use one of your engagement’s existing top-level categories (so “Intune” might come back under Software or Access) rather than invent a new one.

• Code — a short uppercase slug for the ticket code (INC_INTUNE_ENROL_FAIL, REQ_INTUNE_ENROL).

• SLA / Impact / Urgency — already filled, anchored on your engagement’s option lists.

• Why — a one-line rationale.

Every cell in the preview table is editable — text inputs for the four identification fields, dropdowns for the three priority fields. Untick rows you don’t want, edit anything else, then click Create selected.

The prompt is type-aware: on the Incident Categories page, the AI generates symptom / fault rows (“cannot enrol device”, “policy not applying”). On the Request Categories page, it generates planned-ask rows (“enrol new device”, “reset Intune PIN”).

Feature 4 — ✨ AI workflow assistance

Lives on the Workflows section. Use it when you’re building out the lifecycle steps for a request type and want AI help sketching or extending the sequence instead of adding every step by hand.

How it works. Like the other features, it’s opt-in and preview-only: the AI proposes workflow steps, aware of your engagement’s own option lists, and nothing is written until you review and confirm. You can edit or discard anything it suggests before applying.

Safety, deduplication and audit

• Preview, never auto-apply. The AI never writes to the catalogue on its own. Every change requires you to click Apply or Create.

• Idempotent. Re-running the topic suggest with the same topic skips any row that already exists (matched case-insensitively on Category + Type + Item) so you can experiment without creating duplicates.

• Audit trail. Every AI-driven change records the same audit entry as a manual edit, tagged with the action name ai-apply or create. You can see what changed, when, and which logged-in user accepted the suggestion.

• Editor + Draft gate. AI features respect the same access rules as manual edits — you need Editor permission on the engagement and the section must be in Draft.

Cost and performance

The AI uses your configured provider’s pay-as-you-go pricing. A typical run of the topic-suggest feature is one API call returning ~10 rows; the bulk priority suggest splits larger catalogues into batches of 20 rows each so a single click never ties up the page for too long. Most calls return in 5–15 seconds depending on the provider.