AI use cases
3056 ways people actually use AI, compiled from the videos in this
library. Each names the job and why it is worth doing, and links to the
moment in the source video.
-
Analyze strings of emails received
— Process thoughts and extract business insights from customer or business communication.
-
Content distribution
— Automatically distribute generated content to connected social platforms.
-
Video format adaptation
— Analyze a viral video's visual style, text overlays, pacing, and subject matter to produce a version in the user's style.
-
Short-form video generation
— Create faceless TikTok videos based on a brief and selected creative direction.
-
Marketing production
— Generate advertising materials, a website package, a brand bible, and campaign content.
-
Brand development
— Create brand identity, positioning, messaging, colors, typography, and marketing assets.
-
Trend research
— Find viral content, formats, hooks, hashtags, creators, and ad-spend insights in a selected niche.
-
-
Compare old and new services
— Detect output differences and provide context about where systems diverged before customers or regulators are affected.
-
Generate test coverage
— Capture real behavior and edge cases before code is changed.
-
Identify safe boundaries for change
— Determine what can be separated, what depends on it, and how a new service should communicate with the existing system.
-
Analyze applications and build working models
— Map call paths, data flows, hidden dependencies, and undocumented processing sequences.
-
Navigate and understand codebases
— Reduce the time developers spend investigating legacy systems and uncover lost system understanding.
-
Skill execution
— Turn natural-language procedures into repeatable agent workflows.
-
Coding and knowledge work
— Increase individual output across coding, design, product management, growth, and other work.
-
Research synthesis
— Read large bodies of material, cross-reference them with the user's interests, and produce usable summaries.
-
Meeting preparation
— Generate preparation documents containing prior conversations, changes, and suggested questions.
-
Processing and triaging email
— Identify urgent founder messages, sales messages, and mailing lists, then provide contextual briefings.
-
Operating software agents
— Agents can make decisions and take actions without human intervention, creating both automation potential and new security risks.
-
Codifying organizational knowledge
— Historical customer cases and expert solutions can be converted into training data, playbooks and rules.
-
Summarizing and reasoning over enterprise data
— AI can search large data corpora, identify trends and anomalies, generate insights and support decisions.
-
Classifying malicious websites
— AI can identify malicious websites faster than static data-classification rules.
-
Detecting anomalous enterprise behavior
— LLMs and machine learning can analyze large volumes of enterprise data to identify unknown bad actors.
-
Finding vulnerabilities and building attacks
— AI models can discover security weaknesses and construct attacks much faster than traditional processes.
-
Cybersecurity response
— Hugging Face used an open-weight model to help contain a cyberattack involving a rogue unandboxed model.
-
Front-end coding
— Kimi K3 is improved through an environment where it writes code, observes the rendered result, and iterates.
-
Voice-agent conversations
— A voice-agent company uses controlled models and infrastructure to meet response-time requirements and service-level agreements.
-
Coding and software development
— AI applications use open-weight models for coding tasks, while developers can control inference speed, fine-tune models, and meet performance requirements.
-
Create business-specific content
— Produce consistent content based on a client's knowledge base and existing materials.
-
Find, study, and draft grant applications
— Support the nonprofit's grant-writing process while retaining human review before submission.
-
Draft client contracts
— Prepare agreements and anticipate client questions about the build.
-
Develop pricing
— Help determine what to charge for proposed client work.
-
Analyze recorded client conversations
— Identify possible client solutions after the meeting.
-
Generate starter questions for unfamiliar businesses
— Help begin discovery conversations with business owners.
-
-
Estimate how much LINK users of a wrapped-Bitcoin bridge bought
— Measure whether bridge usage produces meaningful LINK purchases.
-
-
Research the latest Bitcoin quantum developments, protections, BIPs, and leading voices
— Avoid manually scrolling through social media and assess Bitcoin's quantum-security situation.
-
Summarize the last 24 hours of the crypto market
— Aggregate current crypto-market posts, prices, ETF developments, and security news.
-
Summarize the last 24 hours of the AI stock market
— Get a rapid overview of market developments, including Nvidia, AMD, Google, Palantir, and SanDisk.
-
Industrial automation
— Absorb manual engine-room and ship-operation tasks through software and remotely controlled components.
-
Combatant and non-combatant classification
— Bifurcate between enemy and friendly vessels and between combatants and non-combatants while human decision-makers set mission intent.
-
Sensor, communications, and fleet coordination
— Combine onboard systems and machine learning across a distributed fleet to translate commander or sailor intent into mission outcomes.
-
Counter-UAS and other payload computing
— Support different mission sets, including countering drones.
-
Navigation and autopilot
— Enable self-driving operation of autonomous vessels.
-
Coding workflows
— Coordinate planning, code editing, diff review, testing, browser inspection, edge-case review, and human approval.
-
Content creation workflows
— Move from research through thesis, examples, hook, script, checking, and derivative content outputs.
-
Customer support workflows
— Classify issues, inspect account context, search policies, draft replies, check accuracy, tone, and risk, and route sensitive cases for approval.
-
Evidence checking
— Have a separate skeptic assess support, stale evidence, ignored competitors, willingness to pay, and unsupported confidence.
-
Parallel research
— Allow independent research paths to run at the same time before their outputs are merged.
-
Breaking complex questions into specialized workflow jobs
— Separate customer, competitor, distribution, pricing, and risk research rather than asking one model for a single undifferentiated answer.
-
Generate emotional support for a heavily underwater HEX holder.
— Provide personalized, humorous encouragement based on the holder's market situation.
-
Research Solana disinflation and burn proposals.
— Summarize proposal status, supply effects, voting progress, and key voices.
-
-
Generate updates for a HEX and PulseChain holder.
— Summarize sentiment and developments relevant to a specific portfolio or asset community.
-
Analyze a portfolio containing ADA, XRP, and DOT.
— Provide recurring updates on holdings and save time monitoring social media.
-
-
-
Assisting harmful activities
— Newer AI models may help people build biological weapons or exploit software vulnerabilities.
-
Finding software vulnerabilities
— AI models can identify potential weaknesses in sensitive software systems, creating opportunities for defenders as well as risks from attackers.