The Tools Are Not the Strategy. The Routing Is.

The AI Tools Field Guide: Every Tool, Every Task, No Wasted Subscriptions

Welcome back to the front row.

Every fourth issue of NVM, we go deep.

Not a roundup. Not three bullets and a take. A proper field guide on one topic, written for practitioners who need to actually do something with it, not just nod along. The regular format returns next week. This is the one you save.

Issue #8 is Deep Dive No. 02. The topic: every AI tool a growth marketer is running right now — the free LLMs, the intelligence platforms, the specialized tools — mapped by what each one is actually built for, with the prompts that unlock them. Not a ranking. Not a features list. A routing guide.

You're probably paying for tools you're underusing. This is the issue that fixes that.

DEEP DIVE II: THE AI TOOLS FIELD GUIDE

What Each Tool Is Actually For, and the Prompts That Unlock It

The first thing to get straight: there are three distinct categories of tools in your stack right now, even if you haven't named them.

Category one: Free LLMs — general-purpose AI assistants you can open in a browser and start typing. ChatGPT, Claude, Gemini, Perplexity, Grok, DeepSeek. These are thinking partners, writers, strategists, and researchers. The differences between them are real, consequential, and consistently misunderstood.

Category two: Intelligence platforms — SEO, keyword, traffic, and competitive tools with structured databases behind them. Semrush, Ahrefs, Similarweb, SparkToro, Answer the Public. These do not generate content. They surface data you couldn't find by guessing.

Category three: Specialized AI writing and marketing tools — purpose-built platforms like Jasper, Copy.ai, and Surfer SEO that sit between categories one and two, wrapping an LLM with marketing-specific templates and constraints.

This guide covers all three. Start with the category that matches the gap in your current workflow.

CATEGORY 1: FREE LLMs — WHAT EACH ONE IS ACTUALLY BUILT FOR

The mistake everyone makes first

Most marketers pick one LLM and use it for everything. The problem is not that any single tool is bad. The problem is that each model has a distinct architecture, training emphasis, and interface design that makes it substantially better at specific task types — and meaningfully worse at others. Using ChatGPT for every task because it's familiar is like only ever using one pan in a kitchen.

Here's the honest breakdown.

Claude — Best for: Long-form writing, brand voice, deep instruction-following

Claude's distinguishing feature for marketers is context retention. The 200,000-token context window means you can paste your full brand guidelines, three competitor landing pages, 50 customer reviews, and a content brief — and Claude will synthesize all of it in one response without losing the thread. No other free-tier tool comes close to this on complex, instruction-heavy tasks.

Where it genuinely outperforms everything else:

• Writing long-form content that holds brand voice across 2,000+ words without drifting
• Following detailed formatting instructions without simplifying them mid-task
• Editing existing drafts while preserving the author's style rather than overwriting it
• Structured document work: strategy decks, campaign briefs, SOPs, stakeholder memos

Where to use something else: Claude does not have real-time web access at the free tier. Do not use it for breaking news research or live competitor monitoring. It also doesn't generate images.

Three prompts to start with:

Email sequence prompt:

You are writing for [BRAND NAME]. Voice: [describe in 2–3 sentences]. Audience: [describe with specifics — job title, pain, sophistication level].

Write a 3-email onboarding sequence for a new [free trial / subscriber / customer]. Email 1: value-first, no pitch. Email 2: one case or proof point in 3 sentences. Email 3: clear CTA with the one thing they should do next.

Length: under 200 words per email. No exclamation points. End each email with one sentence that makes them want to open the next.

Competitive analysis prompt:

Here are three competitor landing pages: [paste text of each]. Here is our current landing page: [paste text].

Analyze: (1) What claim is each competitor leading with? (2) What proof do they use? (3) What objections are they pre-empting? (4) Where is the gap — what is no one claiming that our audience cares about?

Give me a gap analysis, not a summary.

Brand voice audit prompt:

Here is our brand voice guide: [paste]. Here are five pieces of content we've published: [paste].

Score each piece 1–10 on voice consistency. For any score below 7, identify the specific sentence or phrase that broke voice and rewrite only that part.

Best practice: Open a Claude Project and upload your brand guide once. Every conversation inside that project inherits it. This alone eliminates 80% of repetitive setup prompts.

ChatGPT — Best for: Brainstorming, creative variation, multi-format output, agent tasks

ChatGPT remains the broadest general-purpose tool in the stack. Its strength is not depth on any single task — it's breadth across many. Creative brainstorming, ad copy variations, image generation, voice, and the Canvas function for iterative content development all sit in one interface. If you need 30 headline variations in three minutes, ChatGPT produces them faster and with more tonal range than any alternative.

The Canvas function is underused. It turns content generation into an iterative editing session — you generate, select, and refine without starting a new prompt each time. For campaign creative, this is significant.

Where it genuinely outperforms everything else:

• Generating a high volume of creative variations fast (ad copy, subject lines, hooks)
• Image generation (DALL-E 3 integration) directly inside the workflow
• Custom GPTs for specific, repeatable marketing tasks
• Deep Research mode for structured competitive reports (available on Plus/Pro)

Where to use something else: ChatGPT tends to over-polish. For brand voice work or instruction-heavy drafts requiring tight stylistic control, it smooths out distinctiveness. Claude handles nuance in style more faithfully. For real-time research with citations, Perplexity is faster and more transparent.

Three prompts to start with:

Ad copy variation prompt:

Write 20 Facebook ad headlines for [product/service]. Target audience: [describe]. Core benefit: [one sentence]. Constraint: no questions, no exclamation points, all under 40 characters.

Group the 20 into 4 buckets: fear-based, curiosity-based, outcome-based, and social proof-based. Label each.

Hook generation prompt:

Here is a LinkedIn post I'm drafting: [paste full post].

Give me 8 alternative opening lines for this post. Each one must stop scrolling in under 5 words. No "I," no questions, no clichés. Label each by the emotion it activates: curiosity, fear, authority, surprise, relief.

Content calendar brainstorm prompt:

Our brand covers [topic area]. Our audience is [describe]. We publish [cadence].

Generate a 30-day content calendar. For each post: working title, format (carousel, text post, short video, long-form article), the angle that makes it share-worthy rather than generic, and the CTA.

Mix: 40% educational, 30% practitioner perspective, 20% data/proof, 10% product-adjacent.

Best practice: Use Custom GPTs to encode your brand's recurring tasks — a GPT trained on your brand voice and product details saves 10 minutes per brief. Build one, use it for 90 days, then refine it based on what breaks.

Gemini — Best for: Google ecosystem integration, multimodal research, real-time web

Gemini's edge is integration, not raw generation quality. It is wired directly into Google's index, Google Workspace, and Google Ads data. For marketers who live in Docs, Sheets, Gmail, and Search Console, Gemini's practical value is not its writing quality — it's its ability to work across those systems without copy-pasting between tabs.

Gemini Deep Research (available on AI Pro/Ultra) is now a legitimate research tool. It pulls from a wider source range than most alternatives, making it particularly useful for early-stage market analysis and cross-industry trend research.

Where it genuinely outperforms everything else:

• Working directly inside Google Docs, Sheets, and Gmail workflows
• Cross-referencing Google Search data with content tasks
• Early-stage market research with broad source coverage
• Multimodal tasks: analyzing images, PDFs, and data files alongside text

Where to use something else: Gemini's free-tier writing quality is adequate but not distinctive. For content requiring stylistic precision or brand voice consistency, Claude produces cleaner first drafts. For real-time social data, Grok is more current.

Three prompts to start with:

Search intent analysis prompt:

I'm building content for the keyword [target keyword]. Look at what Google is currently surfacing for this query.

Tell me: (1) What search intent is Google interpreting — informational, transactional, navigational? (2) What content format is dominating the first page? (3) What question is every top result answering? (4) What is no one covering that someone searching this phrase would still want to know?

Campaign brief from data prompt:

Here is our Google Ads performance data from the last 90 days: [paste data or upload sheet].

Identify: (1) Which ad groups have the highest CTR but lowest conversion rate — what's the disconnect? (2) Which keywords are spending budget with zero conversions? (3) Where is the CPL improving quarter-over-quarter?

Give me three specific actions to take this week based on this data.

Best practice: Integrate Gemini with Google Workspace. Open a Gemini chat directly from a Google Doc to draft, edit, and rewrite without switching tabs. This sounds minor. In practice it eliminates a category of friction that compounds across a week.

Perplexity — Best for: Real-time research with citations, competitive monitoring, fact-checking

Perplexity is not a ChatGPT alternative. It is a different category of tool — an answer engine rather than a content generator. Its design principle is citation-first, meaning it doesn't generate claims and then find sources. It finds sources and builds answers from them. For a marketer who needs to verify something quickly and trust what comes back, this architecture matters.

Where Perplexity earns its place: any task where you'd normally open five browser tabs and synthesize them yourself. Competitor pricing changes. Industry news. Platform updates. Audience questions being asked in a category. Perplexity does this faster than manual research, and it shows its work.

Where it genuinely outperforms everything else:

• Breaking industry news with source attribution
• Competitive research on specific claims (pricing, features, messaging)
• Question research: what is your audience actually asking right now?
• Fact-checking a draft — paste a claim and ask if it's sourced

Where to use something else: Perplexity is not a content creation tool. Do not use it to draft landing pages, email sequences, or long-form articles. The outputs are structured for research consumption, not publication. Use it upstream of Claude or ChatGPT.

Three prompts to start with:

Competitive intelligence prompt:

I'm researching [competitor name] in the [industry] space.

What are they currently saying about: (1) their pricing model, (2) their primary audience, (3) their top claimed differentiator?

What has changed in their messaging or product in the last 90 days? Cite sources.

Audience question research prompt:

What are the top questions people in [industry/role] are asking about [topic] right now?

Format as: question | why they're asking it | what most current answers get wrong. Cite sources.

Industry brief prompt:

Give me a brief on what has happened in [industry/topic] in the last 7 days. Focus on: (1) platform changes, (2) significant data releases, (3) major brand moves. Cut anything that is commentary without news. Cite sources.

Best practice: Run Perplexity first, then Claude second. Use Perplexity to get sourced orientation on a topic — key facts, current landscape, competitor positions. Then take that material into Claude for actual drafting, strategy synthesis, or content creation. The two-step approach consistently produces better results than either tool alone.

Grok — Best for: Real-time social intelligence, X/Twitter trend monitoring, competitive social listening

Grok's value to marketers is narrow but genuinely difficult to replicate elsewhere: it has native access to real-time X/Twitter data. Every other tool is working from indexed web content — typically 24–72 hours old. Grok is current to the hour. For brands and campaigns where X is an active channel, or where your category has a loud practitioner conversation happening on X, this is a meaningful edge.

Use Grok as a social radar — not a content creator. What is your competitor's audience reacting to right now? What narrative is forming around a product launch? What questions are practitioners asking this week that nobody has answered yet? These are Grok's jobs.

Where it genuinely outperforms everything else:

• Real-time social sentiment on a brand, product, or topic
• Tracking what a competitor's X audience is responding to this week
• Spotting weak signals before they become loud trends
• Monitoring the immediate reaction to a campaign or news event

Where to use something else: Grok's content generation quality is adequate but not exceptional. Its irreverent, occasionally contrarian tone is a feature for some use cases and a liability for others. Do not use it as your primary content drafting tool unless that tone matches your brand.

Three prompts to start with:

Social sentiment prompt:

What is the current sentiment on X toward [brand/product/topic]?

Give me: (1) the overall tone — positive, negative, mixed, (2) the top 3 complaints or concerns appearing most frequently, (3) the top 3 things people are saying positively, (4) any emerging narrative forming around this in the last 7 days.

Trend detection prompt:

What are practitioners in [industry] discussing most actively on X this week?

I want weak signals — topics that are getting traction but haven't hit mainstream marketing publications yet. Flag anything that feels like a 30-day-early story.

Competitor social monitoring prompt:

What has [competitor] posted on X in the last 14 days, and how is their audience responding?

Identify: their top-performing post by engagement, the narrative they're pushing, and any audience frustrations surfacing in the replies.

Best practice: Use Grok for the Monday research sweep. Before you finalize any campaign, content calendar, or competitive analysis, run a 10-minute Grok check on your category. What's the X conversation right now? What are you about to post that might land wrong, or right, because of a current moment? $30/month (SuperGrok) is a competitive intelligence budget, not an AI subscription.

DeepSeek AI — Best for: Cost-conscious teams, technical tasks, API workflows

DeepSeek is the open-source challenger that punched above its weight class in early 2025. For marketers, the relevant advantage is price and availability — its API is among the cheapest for teams building AI-assisted workflows. If you're running automated processes (content pipelines, data processing, bulk analysis) where cost per task matters, DeepSeek is worth evaluating.

For individual daily use, the gap between DeepSeek and Claude/ChatGPT is real in writing quality. Use it for structured, technical, or template-driven tasks rather than creative or brand-voice-sensitive work.

THE ROUTING GUIDE: WHICH TOOL FOR WHICH TASK

Task	First Choice	Second Choice	Never Use
Long-form content drafting	Claude	ChatGPT	Perplexity
Ad copy variations (bulk)	ChatGPT	Claude	Grok
Breaking industry news	Perplexity	Grok	Claude (no web)
Competitor research (current)	Perplexity	Grok	ChatGPT
Social sentiment, X monitoring	Grok	Perplexity	Gemini
Brand voice / instruction work	Claude	ChatGPT	Grok
Image generation	ChatGPT	Gemini	Claude
Google Workspace integration	Gemini	—	—
Deep research reports	Perplexity	ChatGPT	Grok
Cost-sensitive bulk API tasks	DeepSeek	Gemini	—

The default stack for a growth marketer with one paid subscription: Free tier Perplexity (research) + Free tier Claude (writing) + $20/month ChatGPT Plus (creative variation, images, Deep Research). That covers 90% of daily tasks.

CATEGORY 2: INTELLIGENCE PLATFORMS — DATA FIRST, THEN CONTENT

These tools are not AI assistants. They do not generate ideas or write copy. They surface structured data — keyword opportunity, backlink intelligence, traffic patterns, competitive positioning — that makes every LLM output more accurate and more defensible.

The mistake is using intelligence platforms as standalone tools. The right workflow runs them before or alongside LLM sessions, feeding real data into your prompts rather than asking the AI to invent what it doesn't know.

Semrush — Best for: Keyword research, PPC competitive intelligence, all-in-one SEO work

Semrush is the closest thing to an all-in-one digital marketing intelligence platform. Keyword research, backlink analysis, site auditing, rank tracking, paid search competitive intelligence, and — as of 2025 — AI visibility tracking through the Semrush One suite. For marketers who need one tool to cover organic and paid, Semrush is the starting point.

Its AI visibility toolkit now shows how visible your brand is across LLM responses — an increasingly relevant metric as AI-driven search takes share from traditional SERPs.

Three specific use cases:

PPC competitive intelligence: Pull any competitor's domain into Advertising Research. You'll see their active ad copies, their top keywords by spend, and how their messaging has shifted over time. Before writing any new campaign, spend 15 minutes here. You'll know what your competitor is testing, what they've abandoned, and what positioning angles are unclaimed.

Workflow prompt for Claude (after Semrush research):

Here is my competitor's top-performing ad copy from the last 90 days: [paste from Semrush export].

Identify: (1) the core claim each ad is making, (2) what emotion or fear it's activating, (3) what proof element it's using.

Then write 5 ad copies for my brand that address the same audience but take a different positioning angle. Don't copy the structure — counter it.

Keyword gap analysis: Run the Keyword Gap tool between your domain and two competitors. The "missing" column shows keywords they rank for that you don't. Export the top 50 by search volume. This is your next quarter's content calendar, sourced from data rather than intuition.

Content audit + topic authority: Use the Topic Research and Keyword Magic tools to cluster related keywords before briefing a content piece. Give Claude the cluster, not just the single target keyword. Output is consistently better when the AI knows the full topic territory you're working in.

Ahrefs — Best for: Backlink analysis, link-building, technical SEO at scale, agencies managing multiple sites

Ahrefs is the specialist tool for SEO agencies and teams where link-building and technical audit depth are core workflows. Its backlink database and crawling capabilities are the industry standard for this reason. Site Audit scans 170+ issue types with a level of technical detail that Semrush's equivalent doesn't fully match.

The practical rule: if your primary SEO focus is content and keyword strategy, Semrush. If your primary SEO focus is backlinks, site health, and technical issues at scale across multiple domains, Ahrefs. Both score 3.9/5 in aggregate user reviews — the differentiator is genuinely about use case, not overall quality.

Three specific use cases:

Content gap by competitor: Use Content Gap to find topics your top competitors have ranking content for that you don't. Filter by keywords with commercial or informational intent. Export and sort by traffic value — you want the topics competitors are getting real traffic from, not just ranking for.

Workflow prompt for Claude:

Here are 20 content topics my competitors rank for that I don't: [paste Ahrefs export].

For each, identify: (1) the likely search intent, (2) what format is probably ranking (list, guide, comparison, tool), (3) whether I could realistically produce something better than what's ranking.

Flag the top 5 I should tackle first based on potential traffic value vs. difficulty to produce.

Broken backlink recovery: Export your broken backlinks report. You have pages with inbound links pointing to URLs that no longer exist. Each one is traffic leaking from a source that already trusts you. Redirect them. This is an hour of work that compounds for years.

Link prospecting: Use "Best by Links" on a competitor's domain to find their most-linked-to content. If their definitive guide to [topic] has 400 backlinks, that content type has proven demand. Build a better version, then prospect the sites linking to theirs.

Similarweb — Best for: Market intelligence, traffic source analysis, competitive channel strategy

Similarweb answers different questions than Ahrefs or Semrush. It doesn't help you optimize for search. It helps you understand how the whole competitive landscape distributes its traffic — which channels competitors are investing in, where their growth is coming from, and what audience behavior patterns look like across a category.

The practical differentiation: Semrush tells you what keywords competitors rank for. Similarweb tells you whether competitors are getting their growth from paid social, not from search at all — which changes your strategic response entirely.

Three specific use cases:

Channel strategy validation: Before committing to a channel, check Similarweb's Traffic Sources breakdown for 3–5 competitors. If the entire category is getting 60% of traffic from paid search and you're planning to compete on organic SEO alone, you're making a strategic bet most of your competitors have already tried and abandoned.

Workflow prompt for Perplexity (then Claude):

[In Perplexity] What is the current channel distribution for traffic in the [industry] market? Are there any recent reports on how [competitors] are acquiring traffic?

[In Claude, after Perplexity research + Similarweb data] Here is the traffic source breakdown for our top 3 competitors: [paste data]. Here is our current channel mix: [describe].

Where is the strategic gap? What channel is underinvested by the category that could be a first-mover opportunity for us?

Audience overlap analysis: Similarweb's "Audience Interests" and "Also Visited" features show where your competitor's audience goes before and after visiting their site. This is media planning intelligence — where to place content, which publishers to pitch, which channels reach your audience.

Competitor growth signals: Monitor month-over-month traffic changes for key competitors. Sudden traffic spikes — especially from referral or direct channels — often signal a campaign, partnership, or PR moment worth investigating. Grok can help you identify what happened on their social channels the same week.

AnswerThePublic (NP Digital) — Best for: Question research, content ideation, audience language

Answer the Public mines autocomplete and related search data to show you every question your audience is typing around a topic. The output is a visualization of the question landscape for any keyword — organized by who, what, when, where, why, which, how, can, are, will, and comparison queries.

For content marketers and SEO teams, this is one of the fastest ways to generate a genuinely audience-driven content brief rather than an intuition-driven one. The questions come from actual search behavior, not from a marketer guessing what the audience cares about.

Practical workflow:

• Enter your core topic keyword
• Export the questions (filter for the most specific ones — the long-tail questions are often underserved)
• Take the top 10–15 questions into Claude with this prompt:

Here are 15 questions real people are asking about [topic]: [paste list from Answer the Public].

Group them by search intent: informational, comparison/evaluation, problem-diagnosis, solution-seeking.

For each group, identify: (1) the emotional state of someone asking this question, (2) what they're hoping to find, (3) what most current answers get wrong or incomplete.

Give me a content brief for a piece that answers the most underserved cluster.

The free tier gives you 3 searches per day — enough for weekly content planning if you're targeted about what you research.

SparkToro — Best for: Audience intelligence, channel selection, influencer discovery, media planning

SparkToro answers a question that Semrush and Ahrefs don't touch: where does your specific audience actually spend their attention? Not where they search. Where they hang out — which podcasts, YouTube channels, social accounts, newsletters, and websites they actively consume.

The January 2026 update added natural language audience descriptions — you can now describe your audience in plain English ("B2B SaaS growth leads who run Google Ads in MENA") and get behavioral intelligence back without building complex filter combinations. This changed the tool's accessibility significantly.

Practical workflow:

Describe your audience as specifically as possible. The more narrow the description, the more useful the output. Then take SparkToro's top sources (top podcasts, top YouTube channels, top websites) and build a media plan around them rather than defaulting to "let's post on LinkedIn."

Channel validation prompt for Claude:

Here are the top 10 places my target audience spends time online, according to audience research data: [paste SparkToro output].

I currently distribute content on: [list your channels].

Where is the mismatch? Which channels are we investing in that our audience doesn't heavily use? Which channels our audience heavily uses are we ignoring?

Give me a 30-day experiment to test the highest-opportunity channel we're currently not using.

SparkToro is best used quarterly — not daily. Audience habits shift slowly. Run it when you're planning a new campaign, entering a new market, or questioning why your current channels are underperforming.

CATEGORY 3: SPECIALIZED AI MARKETING TOOLS — WHEN GENERIC LLMs AREN'T ENOUGH

A third category exists between the general-purpose LLMs and the intelligence platforms: purpose-built AI marketing tools. They wrap a language model with specific templates, constraints, and integrations designed for marketing workflows.

The honest assessment: Most of what Jasper, Copy.ai, and their equivalents do, Claude or ChatGPT can also do with a well-engineered prompt. The real value of these tools is in two scenarios: teams that need guardrails (consistent templates, brand voice constraints, approval workflows) and teams that need integration with existing marketing stacks.

Tools worth knowing:

Surfer SEO — Writes and optimizes content in real-time against top-ranking pages. The on-page content score updates as you write. Useful for teams producing high-volume SEO content who need integrated optimization rather than switching between tools. The gap vs. free tools: the real-time SERP integration. You can get the same output from Ahrefs + Claude, but Surfer does it in one interface.

Jasper — Brand-voice-trained content generation at scale. The value proposition is the Brand Voice feature — train it on your existing content and it generates new content in that voice with less prompt engineering overhead. Best for content teams producing 20+ pieces per month where consistency at volume is the bottleneck.

Copy.ai Workflows — Agentic content pipelines for marketing teams. You define the workflow (research → outline → draft → SEO check → approval) and Copy.ai runs it with minimal manual intervention. For agencies running repeatable content deliverables at scale, this reduces project management overhead more than it improves quality.

The decision rule: If you're a solo marketer or small team, the free LLMs plus intelligence platforms cover your needs. If you're a content director managing a team producing 50+ monthly assets, the workflow and consistency features of specialized tools start to pay back their subscription cost.

PUTTING THE STACK TOGETHER: THE PRACTICAL FIELD GUIDE

Stop thinking about AI tools as a replacement for work. Start thinking about them as a three-stage workflow.

Stage 1 — Intelligence gathering (before you create anything)

• Perplexity → current landscape, competitor moves, industry news
• Semrush/Ahrefs → keyword opportunity, competitive content gaps
• Similarweb → channel strategy, traffic source intelligence
• SparkToro → where your audience actually lives
• Answer the Public → what they're actually asking

Stage 2 — Synthesis and strategy (turn data into direction)

• Claude → analyze everything from Stage 1, identify the gap, build the brief
• Grok → check social sentiment, real-time X narrative, weak signals
• ChatGPT Deep Research → structured competitive report if you need a comprehensive document

Stage 3 — Creation and variation (execute from the brief)

• Claude → long-form drafts, brand voice-sensitive content, instruction-heavy copy
• ChatGPT → ad copy variations, hooks, creative brainstorming, image assets
• Gemini → Google Workspace integration, Docs and Sheets workflows

The marketers who get outsized ROI from AI tools are not the ones with the most subscriptions. They're the ones who run Stage 1 before they ever open a content editor. The briefing quality determines the output quality. The tools just determine how fast you get there.

PROMPT OF THE WEEK

Building on this issue: the Monday research stack in one prompt chain.

Start every week with this Perplexity prompt:

Give me a brief on what has changed in [your industry/channel] in the last 7 days. Format: (1) the one most significant development and why it matters, (2) any platform or algorithm changes, (3) what practitioners on X and LinkedIn are debating this week. Cut commentary without news. Cite sources.

What this gives you: A sourced, current briefing in under 3 minutes that replaces 20 minutes of tab-scanning — and gives you the raw material for Stage 2 synthesis in Claude.

THIS WEEK

Google's AI Visibility Toolkit in Semrush is now available on standard plans. Previously an enterprise feature, Semrush's AI search visibility tracking — which shows whether your brand appears in LLM-generated answers across ChatGPT, Gemini, and Perplexity — is rolling out to Pro and Guru plans. If you're running SEO for a brand and haven't checked your AI visibility score, this is the week to start. The metric is early, but the category is not.

Perplexity Sonar is now 10x faster than Gemini 2.0 Flash via Cerebras infrastructure. Speed matters for research workflows. If Perplexity felt slow enough to push you back to manual searching, that friction is gone. For real-time research tasks, this changes the practical daily workflow.

SparkToro's natural language audience description feature is now live. You can now describe your target audience in plain English and get behavioral intelligence back without building filter combinations. For MENA marketers targeting specific industry segments, this is the most accessible SparkToro has ever been. Start with: "B2B marketing decision-makers at agencies in the Middle East who use Google Ads." See what comes back.

DeepSeek's API is now the cheapest path to production-grade LLM output for marketing teams running automated pipelines. At a fraction of the cost of GPT or Claude API calls, it's becoming the default choice for bulk content processing, templated copy generation, and data classification tasks where volume matters more than stylistic nuance. If you're building any AI-assisted workflow that runs at scale, price your stack before you build it.

TOOL OF THE WEEK

Perplexity AI — Free tier | perplexity.ai

Not a new tool. But possibly one you're underusing. The free tier gives you real-time, cited web research on any topic, plus access to multiple AI models depending on your task. The use case most marketers miss: competitive intelligence sweeps. Before writing any new campaign, run your competitor's brand name through Perplexity and ask what has changed in their messaging, pricing, and product in the last 60 days. You get sourced answers in 30 seconds. You're no longer going into a campaign blind.

How to use it right now: Set a recurring Monday morning task — run three Perplexity queries before you open any content tool. Your industry news. Your top competitor's recent moves. Your audience's current questions. Ten minutes. Three queries. Everything that follows in your week is better briefed.

Why it matters this week specifically: With AI search taking share from traditional Google results, the same citations-first logic that makes Perplexity useful for your research is shaping how AI answers will reference brands. Understanding how Perplexity sources and cites becomes relevant not just as a tool user, but as a brand trying to appear in those answers.

MY TAKE

I've watched how most marketers use this stack, and the pattern is consistent: they open ChatGPT first, ask it something that requires current data it doesn't have, get a confident but unverified answer, and ship it.

The routing problem is real. And it costs more than time — it costs accuracy, and sometimes trust.

What actually works is treating the AI layer the way you'd treat a team. Perplexity is your researcher — brief them first. Claude is your writer — give them everything the researcher found. ChatGPT is your creative director — push them for volume and variation. Grok is your social monitor — check in before you publish anything into a live conversation.

The separation matters. Each tool has a job. When you give Claude a research task or ask Perplexity to write a landing page, you're asking someone to do work that isn't theirs. The output reflects it.

One more thing. The smartest move in your stack right now costs nothing: build a Claude Project. Upload your brand guide. Every piece of content you produce from inside that project will be better calibrated than anything from a cold prompt. It takes forty minutes once.

That's not a tip. That's the whole infrastructure.

That's Issue #8 of NVM. If this gave you one thing to act on this week, share it with someone who needs it. Reply to this if you have a question — I read every one. See you next week. — Mostafa

NVM · Front row seats to AI and digital marketing

• Subscribe → nvm.marketing/newsletter
• Join the community → nvm.marketing/community
• Write for NVM → nvm.marketing/blog