Skip to content
The Offset Journal

Lots of Ideas on How to Use ChatGPT at Work

Hubspot's guide on how to use ChatGpt at Work with 100 prompts to try. Expert AI Marketing strategies advice.

I’m sure we’re all working hard to fold AI into our workflows, right? This free download (email address only) offers real practical ideas.

Download, then skip to P7 (unless you’re really new to AI) and you’ll find sections covering sales & marketing, customer support, project management along with best practices, advanced prompt engineering and 100 prompt ideas. Great to browse or work through.

Beginner Levels and up

Pop over to Hubspot and grab the download

OK ok – don’t lose your zen. You’re really best downloading the pdf (especially if you want the 100 prompts), but if not, here’s a summary of the main points:

  • AI = machines doing stuff that used to need human brains (like understanding language).
  • ChatGPT uses a combo of pretraining (learning from tons of internet text) and fine-tuning (human trainers refining it).
  • It predicts text based on context—kind of like predictive text on steroids.
  • It’s good, but not perfect: it can still make stuff up. Use with a critical eye.

  • Voice assistants (Siri, Alexa)
  • Netflix-style recommendation engines
  • Face and voice recognition
  • Fraud detection systems

  • Understands natural, conversational language
  • Keeps context over multiple messages
  • Comes in different “flavours” (GPT-4 Preview for deep work, GPT Mini for fast/simple tasks)

  • Knowledge lookups: Quick answers without the Google rabbit hole
  • Task management: Reminders, to-dos, deadlines
  • Writing support: Emails, reports, and proofreading
  • Training help: FAQs, onboarding, process explainers
  • Problem-solving: Brainstorming, decision support
  • Technical support: Code tips, bug fixes, explanations
  • Customer service: Instant FAQ handling, ticket triage
  • Idea generation: Brainstorms, creative prompts
  • Project tracking: Progress updates, task reminders
  • Wellbeing resources: Tips, mental health pointers

  • Automate lead gen & qualify leads faster
  • Personalise at scale (chat, emails, offers)
  • Streamline campaigns & CX without losing the human touch

  • Fast, accurate replies to FAQs
  • AI chatbots = 24/7 frontline for repetitive queries
  • Escalates complex issues to humans
  • Continuously improves with data

  • Auto-assign tasks based on skills + availability
  • Real-time updates and status tracking
  • Simplifies planning and collaboration

  • Extracts insights from data
  • Recommends actions based on past patterns
  • Helps with resource allocation, planning, and problem-solving

  • Schedule meetings, manage calendars
  • Create to-do lists and prioritise tasks
  • Set smart reminders
  • Keep work organised and efficient

  • Find and recommend training materials
  • Share expert insights and trends
  • Suggest skills to develop based on goals
  • Support continuous learning across roles

  • Bias: Models reflect flaws in training data
  • Accuracy: May hallucinate answers—fact check important stuff
  • Privacy: Don’t feed it sensitive info
  • Oversight: Humans still need to stay in the loop
  • Ethics: Stay aligned with your values + policies

  • Be clear on the “why” (e.g. support, content, automation)
  • Fine-tune ChatGPT on company-specific data
  • Set usage guidelines (tone, content rules, etc.)
  • Monitor, moderate, and ask for feedback
  • Don’t blindly trust—validate important outputs
  • Protect privacy and data
  • Have backup plans when AI hits a wall
  • Measure performance regularly

  • Be specific: Clear instructions = better results
  • Add context: Background info helps the AI help you
  • Define the outcome: Tell it the tone, format, or output type you want
  • Iterate: Refine prompts based on responses
  • Break big asks into chunks
  • Let ChatGPT help write prompts if you’re stuck

Better reasoning: Smarter workflows, deeper insights

Web browsing: Real-time info from the internet

Multimodal input: Understands images + text

Canvas mode: Visual collaboration space for flowcharts, plans, mind maps

Personalised AI: Fine-tune tone and expertise for your team

Want more? Check out all Carefully Curated Content here.

Related ccc

How does that linkedin algorithm work - data from 130000 posts

LinkedIn algorithm study: what actually worked across 130,000 posts

Flypost analysed 130,000 LinkedIn posts from 2,500 B2B creators and found that most common LinkedIn advice does not hold up. The useful finding is refreshingly simple: posts with images and first-person openers performed best. Many familiar tricks - hashtags, question hooks, perfect timing and ideal post length - either did…

More info > LinkedIn algorithm study: what actually worked across 130,000 posts

AI visibility and query fan out

AI visibility: query fan-out and why AI visibility needs more than rankings

Query fan-out is the way AI search systems break one user question into several smaller searches before building an answer. That means a page can rank well in Google and still be missed if it does not answer the related sub-questions AI systems are checking in the background. The full…

More info > AI visibility: query fan-out and why AI visibility needs more than rankings