The Most Useful AI Tools Right Now in 2026

Summary of AI tools & how they’re used (as described in the video)

The speaker highlights “the most useful AI tools” and explains how they fit into a daily productivity workflow—primarily for:

• Research
• Content creation
• Coding
• Marketing
• Transcription & meeting notes
• Voice generation

---

1) Perplexity (Pro plan; not required)

Used as a Google replacement for quick question answering and research-style responses.

The speaker uses it constantly, including:

• Practical advice (e.g., dialing in a camera setting)
• Tech comparisons (e.g., comparing Opus/Claude versions)
• Real-time-ish requests (e.g., listening to a basketball game via radio)

They note that the Pro plan may not be necessary for similar usefulness.

---

2) Comet browser (Perplexity integrated) + assistant with slash-command shortcuts

A browser that “does everything Chrome did,” but with Perplexity built in.

Key usage pattern:

• Use the browser assistant for quick prompts without visiting Perplexity directly.

Key feature: pre-written slash commands (custom prompt shortcuts), such as:

• newsreview: breaks down an article—what’s newsworthy and implications for nontechnical consumers
• comment review: reviews YouTube comments, aiming to ignore trolls and surface actionable feedback

Additional notes:

• Model selection inside Comet (examples named: Gemini, GPT, Claude 4.6, Grock 4.1, “Kimmi K2.5”)
• Agent features exist (browser control), but the speaker prefers manual use because it’s faster

---

3) Claude.ai (Max plan; preferred over ChatGPT)

The main assistant for brainstorming, writing, and coding.

Projects used heavily:

• “YouTube producer”: generates ideas, titles, thumbnails, outlines, etc.
• Journal: freeform thoughts + iterative feedback
• Article cleanup: cleans and organizes articles for weekly AI news roundups
• Help Me Explain News: explains pasted news articles (less used after adopting Comet)

Why Claude over ChatGPT:

• Feels less performative and more likely to say what’s needed.

Mentions using “memory” plus:

• Uploaded CSVs exported from YouTube analytics (titles driving views/subscribers/likes, device platforms, countries, etc.)
• Detailed channel instructions

Result: the “YouTube producer” adapts advice using historical performance data and stored context.

---

4) Optimizely (marketing OS) + “Opal” agent

Sponsored portion focused on AI in marketing workflows.

Described as a full marketing OS with an AI agent (“Opel/Opal” per subtitles).

Differentiator:

• Direct access to Google Analytics, so decisions are based on real performance data.

Integrates with CMS:

• Content generation
• A/B test setup

Agent directory/templates include:

• GEO / generative engine optimization
• Auto-generating QR codes

Supports brand consistency:

• Input “brand tone of voice,” personas, and training instructions; outputs remain consistent.

Includes:

• a free 45-minute AI workshop
• possible giveaway for live attendees (Ray-Ban Meta glasses mentioned)

---

5) WhisperFlow (voice-to-text cleanup via hotkey)

Runs continuously with voice dictation triggered by a keyboard shortcut (Mac: Ctrl + Option).

Value vs basic OS dictation:

• Uses a model to clean disfluencies (“ums,” repeated phrases)
• Produces sanitized/transformed text rather than literal speech-to-text

Used especially when:

• prompting LLMs
• working in Cursor to get code generated

---

6) Cursor (AI coding IDE) + building internal tools

Always open; used to build apps/sites and support ongoing development.

Examples of tools built in Cursor:

• Command center dashboard: tracks audience reach, growth by platform, site visitors, newsletter metrics, revenue, and a mini-CRM
• Thumbnail generator: trained on previously liked thumbnails/headshots to infer a style guide; iterative prompting generates options
• Video titler app: takes an MP4, transcribes it, then suggests titles
• Building a newer “Future Tools” UI preview: news items with hover TL;DR summaries (built in Cursor)

Side perspective:

• AI will increasingly be used for bespoke personal software, not just SaaS—using Cursor/Claude Code/code generation tools to spin up small utilities quickly.

---

7) Nano Banana via Gemini (thumbnail image variation)

Used directly in Gemini for thumbnail generation/iteration.

Workflow described:

1. Generate thumbnail concept
2. Refine the prompt (e.g., “remove extra text,” adjust composition/size)
3. Generate the final image variant

Then the speaker adds final text themselves in Canva.

Mention: some thumbnails were generated by both the thumbnail generator and Nano Banana.

---

8) YouTube Studio “Ask Studio”

Built-in AI Q&A inside YouTube Studio.

It uses the channel’s own analytics context to answer questions like:

• “What videos generated the most subscribers in the last 2 weeks?”

Outputs specific video names driving results (including older videos still performing).

---

9) Feedly (RSS + “AI web alerts”)

Presented as a news workflow.

Core components:

• RSS subscriptions to company blogs (Microsoft, Anthropic, Apple, etc.)
• AI web alerts to find relevant AI news on sources not in the RSS list (examples: The Verge, Fierce Healthcare, payments.com)

Personalization:

• Can be trained to show less of certain types of articles (“teach it what to reduce”)

Access:

• Mentioned across devices (iPhone/iPad/laptop)

---

10) Granola (meeting transcription + action item extraction)

A meeting tool that runs in the background (no extra participant invitation needed).

Captures:

• Notes based on the transcript
• Action items (for self and others)

Time-saving advantage:

• Versus the speaker’s earlier multi-step process (OBS audio → AssemblyAI transcription → Claude for action steps)

Additional feature:

• During the call, typing rough notes/questions; the system fills in missing context from the transcript.

---

11) ElevenLabs (text-to-speech + voice cloning workflow)

Used about weekly (for ads/podcast automation), not strictly daily.

Primary described use:

• Podcast sponsorship voice automation
• Producer gets access to the host’s voice in ElevenLabs, enabling ad reads without contacting the host for recordings

Example scenario:

• Scripts for HubSpot ads → generated voice audio → used automatically

Other features mentioned:

• voice changer, voice isolator, sound effects, music
• image/video generation (broad set)

Technical note:

• Mentions model availability (e.g., “11v3” referenced) and regenerating speech outputs.

---

Overall theme / “analysis”

The speaker argues AI is demonstrably useful for productivity by integrating it into:

• Research (Perplexity/Feedly)
• Content creation (Comet/Claude/YouTube Studio)
• Coding and internal tooling (Cursor)
• Marketing optimization (Optimizely/Opal)
• Transcription and execution support (WhisperFlow, Granola)
• Voice automation (ElevenLabs)

They also mention “Claudebot” experimentation—suggesting it could eventually reduce the number of separate tools needed, though it’s more complex and currently less secure/comfortable to fully show.

---

Main speakers / sources (at end)

• Main speaker: Matt Wolf (video host; multiple self-references, e.g., “subscribe to Matt Wolf”)
• Referenced external creator: Matthew Burman (videos about “Claudebot”)
• Referenced colleague: Alex Vulov (suggested Granola as the simpler alternative)