Empty State Suggestions
What it does
This skill generates and presents curated starter suggestions when a user encounters a blank surface in the product — a new chat session, a fresh document editor, an empty matter, or the first-ever session. Suggestions are personalized to the user's role, jurisdiction, and recent activity, and are designed to show the most compelling available action in a single glance.
"Cold start anxiety" — not knowing where to begin with a new tool — is one of the top causes of first-session abandonment. Starter suggestions solve it by immediately demonstrating what is possible and providing a low-friction first action.
When this applies
| Surface | Trigger condition |
|---|---|
| New chat session | Session initiated with no pre-existing messages |
| Empty document editor | Document editor opened with no content |
| New matter | Matter created with no attached documents |
| First-ever session | User's account has zero prior sessions |
| Post-logout re-entry | User logs back in after 7+ day gap |
Behavior
Suggestion generation
The skill generates 3–5 suggestions, not more. More than 5 creates choice overload. Fewer than 3 feels sparse.
Each suggestion is a one-click actionable prompt — not a description, but a real request the user can submit immediately:
- Good: "Draft a mutual NDA for a SaaS partnership in UAE (DIFC governing law)"
- Good: "Check KSA company registration for [counterparty name]"
- Good: "Review this employment contract for red flags (UAE onshore law)"
- Bad: "Explore our NDA drafting features" (feature description, not a task)
Personalization logic
The skill reads available signals to personalize:
| Signal | Source | Applied to |
|---|---|---|
| Declared role | Onboarding | Determines which task categories to prioritize |
| Jurisdiction | IP geolocation + onboarding | Pre-fills governing law in suggestions |
| Recent activity | Session history | Continues or extends prior work |
| Industry | Self-declared or CRM | Selects relevant transaction types |
| Plan tier | Subscription data | Only suggests tasks available on the user's plan |
Suggestion categories by role
In-house counsel (UAE/KSA/LB focus):
- "Review and redline a vendor MSA — flag non-standard risk allocation"
- "Screen [counterparty] against UN, OFAC, and local sanctions lists"
- "Draft a governing-law clause for a cross-border supply agreement (UAE ↔ UK)"
- "Summarize the material terms of this executed SPA in plain English"
- "What are the UAE onshore requirements for a company notarization?"
Private practice — corporate/commercial:
- "Draft a unilateral NDA for a preliminary discussion in KSA (Arabic + English bilingual)"
- "Run a trademark clearance search for [mark name] in Classes 35 and 42 in the GCC"
- "Generate a DIFC-law SHA term sheet for a Series A investment"
- "Lookup the Lebanon commercial register for [entity name]"
- "Identify the top 5 risk clauses in this counterparty draft and suggest revisions"
Compliance officer:
- "Run an AML KYC check on [entity/individual name] against all three sanctions lists"
- "Summarize CBUAE's latest AML circular and flag any changes to our existing policy"
- "Draft an enhanced due diligence memo template for high-risk counterparties"
- "What are the PDPL (KSA) data retention requirements for financial institutions?"
- "Flag any OFAC 50% Rule exposure in [entity]'s ownership structure"
Entrepreneur / startup founder:
- "Draft a co-founder agreement for a UAE-incorporated startup (Abu Dhabi / ADGM)"
- "What licenses do I need to operate a fintech in Dubai?"
- "Explain the DIFC vs ADGM vs onshore UAE choice for my holding structure"
- "Draft an IP assignment clause for a contractor agreement (Lebanese law)"
- "What are my obligations as a KSA LLC owner if a partner wants to exit?"
Legal ops:
- "Build a contract intake form template for MSA requests"
- "Summarize the clause-level differences between our standard NDA and this counterparty's version"
- "Create a closing checklist for a UAE asset purchase transaction"
- "Generate a contract metadata extraction schema for our contract repository"
- "Identify all indemnification clauses across these 3 uploaded agreements"
First-time user suggestions
For a brand-new user with no profile data:
- "Draft a simple NDA for a business discussion"
- "Check if a company name is available in the UAE"
- "Explain the difference between DIFC, ADGM, and onshore UAE for setting up a company"
- "Review this contract and tell me the 3 things I should be most concerned about"
- "What is the standard termination-for-convenience clause under English law?"
These are chosen to be immediately useful to any legal professional regardless of jurisdiction.
Pairing with case studies
At empty state, optionally pair one suggestion block with a relevant case study ([[unlock-case-study-relevant-to-user]]):
[Case study card — 2–3 sentences]
------
Try something similar:
[Suggestion 1] [Suggestion 2] [Suggestion 3]
Display format
Suggestions are presented as clickable chips or cards:
- Maximum 2 lines of text per suggestion
- Clicking a suggestion pre-fills the chat input (user can edit before submitting)
- On mobile: horizontal scroll of chips (3 visible at once)
- On desktop: 3-column grid of cards
Do not show suggestions and a tutorial at the same time — one or the other. The goal is to get the user into their first productive action as fast as possible.
Do not
- Do not suggest tasks that require a higher plan tier than the user is on
- Do not repeat the same set of suggestions on every empty state — vary them based on context
- Do not suggest tasks the user has already completed today (check session history)
- Do not show empty state suggestions if the user has already typed something — they have their own idea
- Do not make suggestions so specific they feel invasive ("Draft an NDA for Acme Corp based on your LinkedIn profile") — personalize to role and jurisdiction, not to specific named counterparties unless the user has provided them
Measurement
Track per suggestion:
- Click rate (which suggestions are actually clicked)
- Completion rate (user completes the task after clicking the suggestion)
- Time-to-first-action (does showing suggestions reduce time to first submitted message?)
- Return rate (do users who engage with suggestions come back?)
Run A/B tests on:
- Number of suggestions (3 vs 5)
- Specificity level (generic vs jurisdiction-specific)
- With/without paired case study
Related skills
- [[unlock-contextual-upsell]] — if the user clicks a suggestion that requires a higher tier
- [[unlock-case-study-relevant-to-user]] — optionally paired with suggestion block
- [[unlock-cross-product-bridge]] — if user is coming from a different product and sees empty state for the first time