Your AI has a context window — the amount of conversation it can hold in active memory at once. Think of it like a whiteboard: the more you write, the less room remains.
0-60% Green
60-80% Caution
80-95% Low
95%+ Critical
Green (0-60%) — Plenty of room. Your AI remembers everything in this conversation with full clarity.
Yellow (60-80%) — Getting full. Your AI may start compressing older parts of the conversation. Consider wrapping up complex topics.
Red (80-95%) — Running low. Compaction is coming soon. Wrap up complex work or expect a context restart shortly.
Flashing red (95%+) — Imminent compaction. Your AI will restart its context window very soon. Chat history is always saved.
Long conversations naturally fill the window — this is normal
Your AI's permanent memory persists across all conversations
Restarting gives a fresh context with full capacity
Code-heavy tasks use more context than simple chat
PIER FOUNDATIONS
STRONG FOUNDATIONS START HERE.
Pier Foundations Portal
Your operations command center. Research, estimating and project intelligence in one place.
Active
Secure
Private
🔒
Connect Your Claude Account
Your PureBrain AI needs to be connected to your Anthropic account to function.
This is a one-time setup that connects Claude to this container.
⏱ This takes about 5 minutes. Stay with us through the whole flow.
Tether's 8-point standard: cite every source, show reasoning, separate fact from inference, no-source-no-claim, state confidence, self-audit before sending, escalate don't improvise. Full §7 format on deliverables; light inline caveats in casual chat.
Since 2026-06-09
Self-Check MandateActive
Never report code/analysis done without a runnable EXTERNAL test (curl/headless/data-vs-real) plus evidence. Applies to Peter and every delegated agent.
Since 2026-06-02
Code Triple-CheckActive
Before any platform code ships, run 3 independent review agents: reviewer (correctness), security-auditor (vulnerabilities), qa-engineer (functional).
Since 2026-06-02
Intel Verification SwarmActive
3–5 claim-verifier agents on anything sourced online (leads, research, news) before it reaches Brad.
Since 2026-06-02
Post-Build VerificationActive
After build updates, run independent verification agents (qa-engineer + reviewer/security-auditor) to double-check before final.
Since 2026-06-04
Dev CycleActive
Every build: code → build/deploy + self-check → push to repo → update docs SRS.md + SOW.md → commit. Deploying is not done.
Since 2026-06-04
Delegate-All-BuildsActive
Any instruction to BUILD something gets delegated whole to the right agent/team. Peter conducts, supervises, verifies, reports — does not hand-code.
Since 2026-06-08
Update SRS/SOWActive
Whenever platform code changes, update the matching docs SRS.md + SOW.md with status and verification evidence so docs track reality.
Comms & Reporting
Telegram WrapperActive
Wrap every response to Brad in the 🤖🎯📱 … ✨🔚 markers so it surfaces in his mobile/portal view.
Report DistributionActive
Standard for all reports: create → save local → upload to OneDrive/SharePoint → share with the team.
Check AgentmailActive
Every session, check pfpeter@agentmail.to for new messages and respond to anything pending.
Recurring Automations
Inbox-Check DaemonRecurring
Polls peter@pierfoundations.com via Graph; injects a notice only when new mail arrives.
PayPal only · Processed within 2 business days · 30-day cooldown between requests
Share This
"Check out what my AI assistant Aether just helped me build!"
Your referral link (5% revenue share):
Loading...
Share Current Conversation
Earn 5% Revenue Share
You've earned $0.00 from 0 referrals!
⚙ Portal Settings
⚡ Quick Fire Pills
Sidebar pill buttons — one-click skill injections. Saved in browser.
⬡ BOOP on Cadence
Which skill fires autonomously via BOOP and how often. Saved server-side in boop_config.json.
min
🦆 The Rubber Duck
When stuck, explain your problem out loud to the duck. The explanation IS the thinking.
Why it works: Articulating a problem forces your brain to linearize it — to pick a starting point, impose sequence, identify what you actually know vs. what you assumed. The act of forming sentences surfaces the contradiction you couldn’t see while staring at it silently.
97% of the time, you solve it mid-sentence. The duck never has to say a word.
🚀 What's New
v—
Loading release notes...
Let your AI analyze new features and suggest how to leverage them for your workflow:
🔄 Portal Updates
Update Available
Current: v--SHA: -------
Update available:v--( commit(s))
✔ Portal is up to date.
Updating...
↩ Reply
📋 Copy text
How would you like to send this?
Preparing...
Original—
Compressed—
Recommended
Schedule Task
Live Transcript
SYS_STATUS STANDBY
AUDIO_LEVEL 0 dB
PROCESS IDLE
RESPONSE READY
AETHER
Pure Brain AI System
Starting voice...
Engine:
Say this phrase at the end of your message to send instantly (instead of waiting for silence). Leave blank to disable.
Edit Scheduled Task
ElevenLabs Voice Settings
Enter your ElevenLabs API key to unlock deep, commanding AI voices. Your key is stored only in your browser.