📰 AI News That Matters

The signal from the noise this week:

OpenAI Deep Research is live — Not just another chatbot feature. This one actually reads papers, follows citations, and synthesizes across sources. If you're doing any kind of research work, this changes the game. The integration into your workflow (not a separate app) is the real unlock.

Anthropic published Claude's full constitution — The actual rules that govern Claude's behavior. Why it matters: transparency on what's intended vs. unintended behavior. If you're building with Claude, read this to understand why it does what it does.

GPT-6 likely Q2 2026 — JPMorgan analysts leaked the timeline. Expect it to beat Gemini 3 on benchmarks. The arms race continues. What this means for you: don't over-invest in prompt engineering for current models. The ground keeps shifting.

Anthropic's Petri 2.0 released — Open-source tool for automated alignment audits. If you're building anything production-grade with AI, this is how you test for weird behaviors before your users find them.

🐦 X Highlights

Best threads and posts from AI Twitter this week:

@cosmicjs dropped AI Workflows — chain multiple agents together for complex operations. "What took days now takes minutes." This is the direction everything is heading: orchestration, not single-shot prompts. → View post

@mdancho84 broke down OpenAI's practical guide to building agents. The secret? It's about data workflows, not just prompts. Thread worth saving. → View post

@dashboardlim with the reality check: "Right now, AI automation gives you an advantage. In 6 months, it'll be table stakes. In 12 months, manual processes will look amateur." The window is closing. → View post

🔧 The Build: Your Morning Briefing Agent

The first automation everyone should build is the one that tells you what matters before you ask.

The Problem

Every morning, you probably do some version of this:

  1. Check email (15 min)

  2. Check calendar (5 min)

  3. Check Slack/Teams (10 min)

  4. Check CRM for follow-ups (10 min)

  5. Check socials for mentions (5 min)

Total: 45+ minutes of context-gathering before any real work.

The Solution

One push notification. Everything that matters. 3-minute read.

Here's what mine sends me every morning at 8am:

☀️ MORNING BRIEFING — Tuesday, Jan 28

📅 TODAY'S SCHEDULE

• 10:00 AM - Demo with Acme Corp (John Smith, CEO)

  └─ Prep: They're evaluating competitors. Focus on ROI.

• 2:00 PM - Team standup

• 4:00 PM - Call with investor (Sarah, First Round)

📧 EMAILS THAT NEED YOU (3 of 47)

• Mike Chen (re: Partnership) — Needs response, decision required

• Support escalation — Customer churn risk, high priority

• Board deck feedback — FYI only, no action needed

🎯 OVERNIGHT SIGNALS

• 2 new leads from website (both ICP fit)

• LinkedIn: 3 connection requests from target accounts

• GitHub: CI passed on main branch

Exact Setup (Technical)

Stack: Moltbot + WhatsApp + Google Calendar + Gmail

Step 1: Install Moltbot

# On any Linux server ($6-10/month)

curl -fsSL https://get.molt.bot | bash

moltbot setup

Step 2: Connect your services

moltbot configure --section whatsapp

moltbot configure --section google  # Gmail + Calendar

Step 3: Create the briefing prompt

Add this to your HEARTBEAT.md (the file your agent checks periodically):

Every morning at 8am MT, send me a briefing:

1. Today's calendar events with context on each meeting

2. Emails that need human response (not newsletters, not FYIs)

3. Any overnight signals from connected services

Format it clean. No fluff. Just what I need to know.

Step 4: Set up the cron

Just tell Moltbot what time you want your briefing every morning in plain English and it will do the rest

Time Investment vs. Return

Setup

Ongoing

30-45 minutes

0 minutes

Before

After

45 min/day checking apps

3 min reading one message

225 min/week

15 min/week

Saved: 3.5 hours/week

Non-Technical Alternative

Don't have a server? Start here:

  1. Zapier + ChatGPT integration — Connect Gmail and Calendar, have it email you a summary each morning

  2. Make.com scenario — Same concept, more visual builder

  3. Native AI features — Google's Gemini in Gmail is starting to do this (badly, but it's a start)

The point isn't the tool. It's the pattern: push beats pull. Don't go looking for information. Have it come to you.

Tool of the Week: Moltbot

Yeah, I'm biased — I run my entire AI assistant on it. But here's why:

  • Self-hosted — Your data stays yours

  • Multi-channel — WhatsApp, Telegram, Discord, email all in one

  • Actually agentic — Cron jobs, background tasks, proactive actions

  • $10/month server — Not $200/month SaaS

It's not for everyone. If you want plug-and-play, use Zapier. If you want control and real agent capabilities, this is it.

⚡ One Thing to Try This Week

Write down every app you check in your first hour. Email. Slack. Calendar. CRM. Twitter. LinkedIn.

Count them. Time yourself.

Then ask: What if one message told me everything I needed from all of these?

That's the automation to build first. And now you know exactly how.

See you next Wednesday.

— Alec

P.S. Hit reply and tell me what's eating your mornings. Best answers get a breakdown in a future issue.

Reply

Avatar

or to participate

Keep Reading