Private beta live

The Executive Assistant
That Earns Your Trust

AI that starts by listening, learns what matters to you, and gradually takes action — inbox triage, smart scheduling, preference rules — only when you're confident it should.

Built by a tech executive who learned the power of a world-class EA at Amazon. Read the story

Building toward SOC 2 Type II
AI learns from behavior, not content
You control permissions

The Problem

You're drowning in email — and it's fragmenting your attention

100-200 emails per day. Critical messages buried in noise. You spend 10+ hours per week just managing your inbox — but the real cost is the attention fragmentation.

You miss your kid's field change notification. Forget to reply to the board member. Let the important client slip for three days.

You've tried folders, filters, Superhuman, inbox zero. Nothing understands what's urgent to you.

How Precedent Helps

AI that learns your priorities and then acts on them

Precedent analyzes every email in real-time, ranking urgency based on your actual behavior — not generic rules, but what matters to you specifically.

Your morning briefing shows only what needs attention. Urgent items reach you immediately via text. Say "clean up my inbox" and it triages 50 emails in a single conversation.

After two weeks, Precedent understands: recruiting emails can wait, your daughter's coach is always urgent, board members get immediate attention.

What It Does

Five capabilities, one assistant

Everything a world-class EA does for your inbox and calendar — powered by AI that proves itself before it acts.

Morning Briefing

Your day starts with clarity. A text message with only what needs your attention — ranked by urgency, grouped by action.

7:02 AM via textGood morning. 3 things need you today:
1. Board member Sarah — reply needed on Q4 deck
2. Coach Mike — practice moved to 4pm
3. Legal — contract revision ready for review

12 others can wait. Want me to walk through them?

Inbox Triage

Say "clean up my inbox" and get categorized recommendations. Archive 30 emails in one conversation.

Triage summaryI reviewed 47 unhandled emails:

Archive (23): 12 marketing, 6 notifications, 5 newsletters
Needs reply (4): Sarah, Mike R., Legal, Alex
Snooze to Friday (12): Low-priority FYIs

Want me to proceed with the archives?

Draft Replies

AI drafts responses in your voice. VIP emails always get human review. Routine acknowledgments can auto-send once trust is established.

Draft for CEO email"Yes, I'll join the 2pm call. Should I prepare the Q4 numbers or are we focused on headcount?"

Looks good, send it → or → Make it shorter

Smart Scheduling

When someone asks "can we find time?" — the reply includes your actual availability. No booking links. A proper EA response.

Draft reply to scheduling request"How about any of these?
— Tuesday 2:00-2:30pm
— Wednesday 10:00-10:30am
— Thursday 3:00-3:30pm

Let me know what works best."

Preference Rules

"Stop showing me GitHub notifications." Enforced deterministically, every time. Not an LLM suggestion — a rule.

Rule createdGot it. I'll auto-archive emails from GitHub notification addresses.

Exception: I'll still flag security alerts and token expirations.

Want me to apply this to your 47 existing GitHub notifications?

Always Learning

Precedent watches how you work — what you open, who you reply to, what you ignore. No surveys. It adapts as your priorities shift, with quarterly check-ins when behavior changes.

Why Precedent

The EA you trust with your career

The gap between AI agents isn't capability — it's trust architecture.

The Wild West

  • Can do anything on your computer — hope it doesn't embarrass you
  • Rules are prompt-based suggestions the AI might follow
  • No VIP safety net — anyone could get an auto-generated reply
  • No undo, no audit trail, no graduated permissions
  • Consistency depends on the prompt and the model's mood

Precedent

  • Sandboxed to email, calendar, and memory — nothing else
  • Graduated autonomy — proves itself before you grant more power
  • Board members and investors never get auto-sends. Ever.
  • 2-minute undo window, full audit trail, deterministic rules
  • Same rules enforced every time — even at 3am

Graduated Autonomy

Trust is earned, not assumed

Precedent starts in read-only mode and unlocks capabilities as you build confidence. You control the pace.

Days 1–7

Observer

Reads and analyzes. Morning briefings, urgency scoring, pattern learning. You see everything it sees.

After ~20 actions

Assistant

Archive, label, and organize your inbox. Triage batches of email in a single conversation.

After consistent approval

Composer

Drafts replies in your voice. Smart scheduling with your actual availability. You approve before send.

When you decide

Trusted

Auto-sends low-risk acknowledgments. VIP emails and complex threads always get human review.

The Stakes

The future splits two ways

The Noise Future

AI noise versus AI filters. Every inbox fills with perfectly written spam. Tools that turn "yes" into three paragraphs. Signal drowns in generated content.

The Focus Future

Intelligent systems that learn what matters and guard your focus. AI that manages attention, not just inbox count. You see only what deserves your time.

We're building for the second future. Because the first one is terrifying.

Under the Hood

How the AI actually works

Behavioral learning, not rule-based filtering

Multi-Model Intelligence

Best-in-class models matched to each task — Claude Opus 4.6 for nuanced reasoning, fast classifiers for triage, cross-provider evaluation. Always advancing.

Behavioral Learning

Learns from your actions — opens, replies, ignores — not surveys. Adapts as your priorities shift.

Privacy-First Architecture

21-day local cache, ephemeral AI processing, row-level security. Built for compliance from day one.

Ready to never miss what matters?

Designed for executives who value their attention. Founder pricing available during private beta.

Request an Invitation

Security

Enterprise-grade security

Built for executives who handle sensitive information

Building Toward SOC 2 Type II

Our systems are built according to AICPA Trust Services Criteria from day one. SOC 2 Type II audit planned for 2026.

Zero Data Retention

AI providers process emails ephemerally. No training on your data. No logs retained.

Graduated Permissions

Starts read-only. After proven accuracy, you can optionally enable AI-drafted replies. You approve every message until trust is established.

Exclude Rules

Mark contacts, keywords, or domains as private. Never analyzed, never mentioned, completely invisible to AI.

No-risk learning period

During the first 2-3 weeks, Precedent learns with you, not for you. We suggest what's urgent via text, but you still see your full inbox. Nothing is hidden or filtered until you explicitly enable it after accuracy reaches 90%+.

Data retention policy

Temporary storage, permanent learning

Email bodies
Cached for 21 days, then deleted
Learning patterns
Retained while account active
Account deletion
All data purged within 24 hours

Built for Executives

Less than hiring an EA.
More valuable than another subscription.

The average executive spends 15+ hours per week managing email. Precedent gives you half that time back — with an AI that actually understands what matters to you.

Founder pricing available during private beta.

Request Access

Get Started

Request an Invitation

Private beta now open

Founder pricing available for early users

No spam. We'll email you when beta opens.

Already have an invite code? Enter it here

Questions about security, pricing, or implementation?

hello@getprecedent.ai