These Are the Best AI Tools in 2026

Overview / Framing

• The speaker argues that “the best AI tool” changes constantly, so you need a process for:
  • testing tools,
  • tracking updates, and
  • combining multiple tools, rather than committing to one model forever.
• They claim to have spent “dozens of hours” testing AI tools across multiple companies and will share:
  • what each tool is good at, and
  • how to combine them effectively.
• The speaker promises a “system” that helps users quickly switch to new AI tools without rebuilding context.

---

Tool #1: ChatGPT “Astra” (and local/agent-like control)

Core idea / feature claims

• Unlike typical chat-only assistants, Astra can “control your computer”:
  • access files,
  • open applications, and
  • perform actions in the background while you continue working.
• It can function like a “ghost employee”, reducing the need for constant copy/paste between tools.
• It emphasizes remote access via a phone app for communicating and controlling tasks.

Examples mentioned

• DoorDash automation
  • Uses order history to place a user’s usual meal order.
  • Can handle expired credit card renewal with user approval.
  • Can also manage scheduling.
• Slack workflow example: receipt processing
  • Collects receipts from the last month.
  • Reads and renames files (e.g., by date/vendor).
  • Ignores certain receipts (e.g., business trips).
  • Sums totals.
  • Sends results to a chief of staff via Slack.

Setup / guide details

• Download/enable ChatGPT, then go to:
  • settings → integrations
  • enable computer/app access
• Tip: click “Show Pet” to create a small icon/voice prompt so it’s accessible even while you’re in Slack.

Prompting framework

• The speaker mentions a workflow prompting structure: MAPS
  • Mission (goal)
  • Ask (clear desired task)
  • Parameters (files/rules/restrictions)
  • Form (deliverable format + notification preferences)

Practical review stance

• Astra is positioned as best for local tasks and computer-executed workflows, minimizing the need to switch between tools and programs.

---

Tool #2: GrokBot (specialist agent orchestration)

Core idea / feature claims

• GrokBot is presented as a specialist “agent platform”.
• It manages a team of hyper-specialized bots that delegate and coordinate with each other.
• It emphasizes orchestration using simple language, including scheduling reminders (e.g., check a Slack post at 8:00 AM and notify if no update).

Security / onboarding angle

• The speaker contrasts GrokBot with more technical “agent platforms” that require expertise (they mention examples like installing platforms such as “Open claw” and “Hermes”).
• GrokBot is claimed to be simpler/safer, avoiding typical complexity such as:
  • cron jobs,
  • Vercel instances,
  • API keys,
  • and automation plumbing.

Team/bot architecture

• Workflow:
  • create a new bot
  • assign a role (examples: HR manager, inbox manager, PR bot, admin assistant bot, revenue bot, etc.)
• Supports group chats where bots collaborate by sharing context.
• The speaker says they currently run ~37 bots at all times.

Demonstrated business automations

• Sales bot
  • Identifies sales bottlenecks/constraints across companies
  • Messages managers
• Financial bot
  • Lives in inbox/Slack/financial tools
  • Consolidates financial tasks
• Inbox management
  • A chief-of-staff bot coordinates with inbox/copywriting/outreach bots to sort emails and draft responses
• “Teach task” / video camera method
  • The user records themselves performing a browser task (“teach task”)
  • The bot learns the workflow
  • Framed as a way to transfer operations across tools

Advanced concept (agent swarm layering)

• Mentions an “agent swarm bot” that can install GrokBot under different accounts.
• Presented as powerful but advanced/risky, with possible cost/complexity concerns.

Source credibility

• The speaker attributes creation/testing to Elon Musk / xAI, and claims xAI employees were early testers.

---

The “main hack”: don’t choose—combine tools

The speaker argues that Astra and GrokBot are different enough that the best approach is to combine them:

• ChatGPT/Astra acts as the “chief of staff” running locally on the user’s computer.
• GrokBot handles orchestrated delegated work using specialized bots.

Hybrid scenario example

• Astra opens GrokBot and provides receipts/reporting instructions.
• GrokBot’s agents generate a visual report and send it to a person in Slack (e.g., a CFO).

---

The “switching problem” and the second brain

Problem analysis

• A major pain point is tool churn:
  • each new AI tool often requires rebuilding context—memories, preferences, client knowledge, and workflows.

Proposed solution

• The speaker argues new tools should work immediately “with the lights on” if user context is centralized.
• Solution: build a “second brain” to store and share knowledge across AI models.
• Mentions general options: Obsidian, Google Drive, Notion, etc.
• Critiques DIY “second brain” setups as costly/complex to maintain:
  • token costs,
  • processing overhead,
  • and incomplete context that reduces decision quality.

Product mentioned: Apex / apex.host

• The speaker claims to have created Apex as a centralized memory/context system.
• Claimed features:
  • Centralizes meetings/chats/comments/decisions/reports (“everything”)
  • “Captures” and transfers context so multiple AI models can reuse it quickly
  • Reduces switching cost/effort and makes adopting new models seamless
  • Emphasizes ownership/control of the “brain” while renting models

---

Guidance / “tutorial-like” takeaways emphasized in the video

• Use Astra for computer-executed local automation and background workflows.
• Use GrokBot for agent orchestration, including:
  • specialized bots,
  • role-based coordination,
  • group chats.
• Use MAPS prompting to improve reliable task execution.
• Use “teach task” / recording methods to transfer real workflows visually.
• Implement a central memory (“second brain” / Apex) so new AI tools can start with user context immediately.

---

Main speakers / sources

• Primary speaker: Unspecified creator/host (self-references include:
  • “I spent dozens of hours testing…”
  • and personal experience/examples)
• Referenced sources/creators:
  • Elon Musk / xAI (credited as creator and early tester of GrokBot)
  • The creator’s own product/company: Apex (apex.host)
  • Third-party ecosystem mentioned: open-source/agent platforms (e.g., “Open claw,” “Hermes”), productivity tools (Slack/Notion/Obsidian/Google Drive), and an example CRM integration mention (e.g., HubSpot).