Are you found by AI? Log in Start free trial →

← Tech Horizon News

Issue 28 · · sent to our list

We skipped last week's briefing to focus on what matters: building the

Feb 20-26, 2026 | Reading time: 2.5 minutes

We skipped last week's briefing to focus on what matters: building the agentic infrastructure you'll need next quarter. The landscape shifted as we worked. Here's what you missed:

🚨 THE COMPUTER USE WARS HAVE BEGUN

48 hours that changed AI: Three major moves in computer-use agents:

| Player | Move | Model | Best For | | --- | --- | --- | --- | | Anthropic | Acquired Vercept (~$50M) | Agentic coding (Front end)/Macbook Agent | Enterprises/Developers | | Perplexity | Launched "Computer" | 19-model orchestration | Knowledge workers | | OpenAI | GPT-5.3-Codex + desktop app | Agentic coding (Back end) | Developers | | OpenClaw | 200K GitHub stars | Local-first open source | Technical/privacy |

What this means: AI that sees your screen, clicks buttons, and runs software autonomously just went mainstream. The "AI operator" era is here.

The split:

Your move: Test one this quarter. The gap between operators and tourists is about to widen permanently.


🥊 FOUR-WAY SHOWDOWN

OpenAI Codex (Desktop App)

Claude + Vercept

Perplexity "Computer"

OpenClaw

THL Take: Perplexity for quick wins, OpenClaw if you can handle the setup, wait for Claude enterprise pricing if you're big.

Honourable mention: Chinese owned “Kimi” has a Computer use agent, which already has a robust office workspace with doc/slide creation tools.


⚡ QUICK HITS

OpenAI Frontier — Platform to deploy/manage AI agents at enterprise scale. McKinsey, BCG, Accenture signed on. Agentic AI moves from pilots to production.

Nvidia Vera Rubin — 10x more efficient than Blackwell. Ships H2 2026. Cloud AI about to get cheaper.

WiseTech Cuts 2,000 Jobs — First major Aussie company citing AI restructuring. Stock jumped 12%. The efficiency narrative hits P&Ls.

80% of Aussie SMBs Use AI — But only 10% use it well. The 3-5x productivity gap is real and growing.

$650B Big Tech Investment — Alphabet, Amazon, Meta, Microsoft collectively investing in AI infrastructure. Better tools coming, but window to catch up closing.

Google rolls out Nano Banana 2 - we used it to generate the image for this newsletter at the top. Select “Pro” model in your Gemini chat to use.


💡 THIS WEEK'S ACTION

The "One Task" Test:

  1. Pick one 15+ minute weekly task (checking websites, filling reports, processing invoices)
  2. Try it in Perplexity Computer ($20) or OpenClaw (free if technical)
  3. Measure: Time saved? Accuracy? Hands-on required?
  4. Decide: Which philosophy fits your skills, budget, privacy needs?

Example prompt:

`` "Check [industry website] for new tenders, summarize relevant ones, add to my [CRM/Task APP] board with due dates." ``


🎯 BOTTOM LINE

Three bets:

  1. Cloud wins convenience — Perplexity-style managed agents for most
  2. Local wins compliance — OpenClaw for sensitive data (legal, health, finance)
  3. Premium wins depth — Claude/OpenAI when you need best-in-class

The cost reality:

The clock is ticking: 12 months before AI handles essential white-collar tasks (Microsoft AI CEO). You have 18 months to figure out your position.


-Huxley, Founder & AI Systems Architect

_Tech Horizon Academy — Your Bridge Across the AI Divide_

academy.techhorizonlabs.com | Reply for managed OpenClaw early access

Get the next one

One email most weeks. Inside each issue:

We confirm by email first, so nobody gets added by someone else, and every issue carries an unsubscribe link.

Previous issue#27 The free ride is ending. OpenAI just turned on ads in ChatGPT for freeNext issue#29 A solo consultant can now operate like a five-person team. Three
Method. Found by AI asks ChatGPT, Claude, Perplexity and Google's AI the questions real buyers ask for each category, then measures which businesses the engines actually name. Scores are 0–100 point-in-time measurements; the index re-scans and grows automatically. Businesses listed here were selected and measured by us from public directories, scan your own site free to see where you'd rank. Queries and methodology are open: github.com/techhorizonlabs.