The AI Tool Stack I Use Daily as a CTO (And What I Dropped)

Last updated September 10, 2026.

My daily stack runs 8 tools, with two of them splitting the meeting-capture slot because no single tool covers both bot-in-the-room and no-bot workflows. Nothing else overlaps. Nothing sits unused. The monthly cost comes to $308.

Here’s the exact stack, why each tool earned its position, and the casualties that didn’t survive the cut.

The Daily Stack

1. Claude (Max, $200/mo)

Job: Deep work partner, writing, analysis, architecture, strategy

Every significant piece of writing starts with Claude. Client proposals, strategy documents, technical architecture docs, this article. Claude drafts, I refine. The large context window handles entire project contexts that other tools lose track of. I feed Claude a project brief, existing documentation, and constraints in one conversation, then work through deliverables without repeating context.

Why it won: Output quality on professional writing exceeds every alternative I’ve tested. Detailed analysis in my Claude vs ChatGPT comparison.

Why Max rather than Pro. Pro rate-limits long sessions, and my work runs in long sessions. Max removes the ceiling that turned a productive afternoon into a wait, and it carries Claude Code at the usage level real engineering demands. At $200 a month it costs ten times Pro and earns every dollar of the difference.

2. Claude Code (included with Max)

Job: Engineering partner, coding, refactoring, debugging, architecture

Claude Code operates directly in my terminal with full access to my codebase. Multi-file refactors, debugging complex issues, building features across the full stack. Claude Code handles engineering work that browser-based AI tools can’t touch because they lack project context.

Why it won: It understands an entire codebase and executes changes across it, which covers the engineering work I need most. Full review in my coding assistants article.

3. ChatGPT Plus ($20/mo)

Job: Research, data analysis, image generation, quick tasks

ChatGPT handles everything Claude doesn’t: web search for current data, Code Interpreter for quick data analysis, DALL-E for image generation, and plugin integrations for specialized tasks. When I need a chart from a CSV in 30 seconds or current market data for a client presentation, ChatGPT delivers.

Why it survived alongside Claude: Each tool dominates different tasks. Using both eliminates blind spots. Detailed breakdown in my Claude vs ChatGPT comparison.

4. GitHub Copilot (Pro, $15/mo)

Job: Inline code completion, the background assistant

Copilot runs silently in my editor, completing lines and functions as I type. It handles boilerplate, repetitive patterns, and predictable code. I barely notice it working, which means it works perfectly.

Why it coexists with Claude Code: Different scopes. Copilot handles tactical, line-by-line completion. Claude Code handles strategic, multi-file engineering. They complement rather than compete.

5. Perplexity Pro ($20/mo)

Job: Research engine, market research, competitive analysis, fact-checking

Perplexity replaced my Google searches for business research. Ask a complex question, get a structured answer with citations I can verify. Pro Search digs deeper, asking clarifying questions before delivering comprehensive research briefs. When writing AI tool reviews, Perplexity surfaces current pricing, features, and competitive positioning faster than manual research.

Why it won over ChatGPT for research: Perplexity cites every source. ChatGPT summarizes without always showing where the information came from. For research that informs business decisions or published content, citations matter.

6. Make ($16/mo)

Job: Workflow automation, connecting tools, automating processes

Make automates the repetitive workflows across my business: lead qualification, email sequences, data syncing, report generation, notification routing. The visual builder handles complex branching logic that simpler tools can’t express.

Why it won over Zapier: More operations per dollar, better visual builder for complex logic, and direct LLM integration for AI-powered automation steps. Full comparison in my automation tools article.

7. Fathom ($19/mo) + Granola ($18/mo)

Job: Meeting capture, with a bot in the room when that helps and without one when it doesn’t

I run both. Fathom handles meetings where verbatim transcript, CRM logging, or centralized team archives matter (customer discovery, sales calls, executive meetings I need to reference exactly). Granola handles meetings where a bot in the participant list would change the conversation (1:1s, sensitive customer escalations, in-person meetings that Fathom cannot capture at all). As a fractional CTO, I rely on the pairing for note synthesis, CRM follow-up, and searchable meeting archives.

Why the pair beats either alone: Fathom’s verbatim capture handles compliance workflows and precise-quote pulls. Granola’s no-bot model handles sensitive conversations and in-person meetings Fathom cannot touch. No single tool covers both surfaces without compromise. Full comparison in my Granola vs Fathom head-to-head and meeting assistants review.

The Monthly Cost

Tool	Cost
Claude Max	$200
ChatGPT Plus	$20
GitHub Copilot Pro	$15
Perplexity Pro	$20
Make Pro	$16
Fathom Team	$19
Granola Business	$18
Claude Code	included with Max
Total	$308

$308/month for a productivity stack that replaces what would otherwise require a research analyst, a copywriter, a designer, an admin assistant, and several hours of manual work daily. The ROI materializes in the first week.

Claude Max accounts for two thirds of that total, and I would cut every other line before I cut it. The cheap subscriptions save minutes. Max changes what I can take on, which is a different category of return.

What I Dropped (And Why)

Gamma Plus ($10/mo). Dropping at renewal, September 2026. I recommended it without qualification. Gamma then changed its create flow so the default path rewrites content you paste and disregards formatting rules you set in the prompt. The workaround path, pasting text in Preserve mode, turns a 30-second job into about 30 minutes of careful work. I replaced it with an HTML template and a short script that renders the same output in one command. Full reasoning in my updated Gamma review.

How to Build Your Own Stack

Start with one tool. Claude or ChatGPT, either one transforms your daily workflow immediately. Use it for 30 days before adding anything else.

Add tools to solve specific pain points. Don’t subscribe to a tool because it looks cool. Subscribe because you identified a specific task that consumes time, tested the tool on that task, and measured the time savings.

Audit quarterly. Every 90 days, ask: did I use this tool this week? If not, cancel it. Unused subscriptions drain budget without delivering value.

Overlap kills value. Two tools that do the same job mean one tool wasting money. When you add a new tool, identify what it replaces and cancel the overlap.

The perfect AI stack doesn’t exist. The right stack, the one that maps precisely to YOUR daily tasks with zero overlap and zero waste, delivers transformative productivity gains at a cost that pays for itself in hours saved during the first week.

This reflects my actual daily tool usage as of July 2026. I pay for every tool listed, no sponsored placements or free accounts influence the selection. Some links may earn a commission, see the about page for details.