Sandstone Logo

Why AI-Driven Legal Intake And Triage Matter More Than Ever

Unstructured requests are eating legal’s time and slowing the business. Here’s how AI-driven intake and triage convert chaos into clarity—and turn legal into a proactive, scalable partner.

Jarryd Strydom

September 26, 2025
Up to 70% of in-house legal requests arrive unstructured—via email, Slack, or a quick “can you look at this?”—and leaders regularly estimate that 25–40% of counsel time disappears into triage, chasing context, and rework. That’s not a workflow—it’s a tax on speed and trust.
Done right, AI-driven intake and triage reverse the drain. They turn every request into structured data, apply playbooks in real time, and route work with crafted precision. The result is a legal function that moves in step with the business: faster answers, fewer bottlenecks, and a foundation of knowledge that compounds with every decision.

The Real Cost Of Unstructured Intake

Unstructured intake hides work and blurs accountability. When requests surface in inboxes and chats, three problems follow:
  • Missing context: business owners forget to include contract templates, counterparties, or timelines, leading to back-and-forth and rework.
  • Invisible queues: leaders can’t see volume, cycle time, or workload balance, so capacity planning is guesswork.
  • Playbook drift: advice varies by person and memory, not by policy, making outcomes inconsistent.
These costs add up in ways that are hard to see but easy to feel—slower deals, strained relationships, and legal seen as a blocker. Intake isn’t a portal problem; it’s a knowledge problem. You need a living layer that gathers the right facts, applies positions consistently, and pushes work forward without extra taps on legal’s shoulder.

How AI Agents Turn Intake Into Insight

AI agents anchored to your playbooks can perform the heavy lift at the edge of intake:
  • Triage: classify the request (NDA, DPA, marketing review), extract key terms from attachments, and flag missing information.
  • Risk screening: compare the request to approved positions (e.g., liability caps, data transfer rules) and score risk.
  • Routing: assign to the right workflow—privacy, commercial, employment—based on metadata and thresholds.
  • Acceleration: propose next steps (approve, redline with playbook clauses, request vendor security docs) so legal starts from action, not from scratch.
Because every step writes back to a shared knowledge layer, the system gets stronger. Questions answered today become authored guidance tomorrow. Exceptions become positions. What used to vanish in email now compounds into institutional memory.
What “Layered Knowledge” Looks Like In Practice
Think of a procurement DPA or security review:
  1. Request comes in via Slack or a link. The agent prompts for vendor name, data categories, geography, and timeline. If an MSA or policy is attached, it extracts the essentials.
  2. The agent checks your privacy playbook and data map. It flags cross-border transfers, unusual data types, or missing subprocessor details. Low-risk profiles are auto-routed to fast-track; medium risk gets standard review; high risk escalates with context.
  3. If redlines are needed, the agent drafts them from your approved clause library (e.g., SCCs, audit rights), with a clean explanation summary for the business.
  4. The system logs decisions, positions invoked, and outcomes. Next time, similar requests are faster because the knowledge is already carved to fit your contours.
That’s strength through layers: each intake enriches data; each decision sharpens guidance; each workflow becomes more precise and natural to run.
KPIs That Prove It’s Working
If AI-driven intake is paying off, you’ll see it quickly:
  • Median time-to-first-touch under 4 business hours for standard matters.
  • 30–50% reduction in back-and-forth messages to gather context.
  • Auto-approval/fast-track rates rising to 20–40% for low-risk requests (e.g., mutual NDAs, vendor DDQs).
  • Playbook adherence improving, measured by variance from preferred positions.
  • Request visibility: 100% of matters captured with structured fields, not buried in email.
These aren’t vanity metrics; they’re lead indicators of throughput, predictability, and trust between legal and the business.

A Practical Next Step: Pilot One High-Volume Flow

Start small and prove value in a week:
  • Pick a single workflow: mutual NDAs, marketing claims review, or vendor DPAs.
  • Define the minimum facts: 5–7 fields the agent must collect (counterparty, jurisdiction, template, timeline, risk flags).
  • Codify your positions: acceptable caps, data transfer rules, clause fallbacks, and when to escalate.
  • Connect the edges: accept requests from Slack/email, route to Jira or your matter system, and notify the requester with clear SLAs.
  • Measure baseline: time-to-first-touch, cycle time, and rework. Iterate weekly.
Actionable takeaway: frame the pilot as “from request to decision” in minutes, not “new portal.” Anchor on measurable speed and consistency.

Why This Matters Now

Budgets are tight, workloads are up, and stakeholders expect legal to work at the tempo of the business. Intake is where that promise is kept or broken. When AI agents convert unstructured requests into structured actions—and when each action feeds a living knowledge layer—legal shifts from reactive support to a proactive force for speed, alignment, and trust.
That’s the Sandstone approach: layered data, modular workflows, and natural integration with how teams already work. By transforming playbooks and positions into a living, AI-powered operating system, you build a legal foundation that strengthens with every intake, triage, and decision. Scalable, streamlined operations aren’t a nice-to-have—they’re the bedrock of growth.

About Jarryd Strydom

Jarryd Strydom is a contributor to the Sandstone blog.