Claude

 

Anthropic's AI assistant, often used for long documents, writing and analysis.

Claude is an AI assistant made by Anthropic, an AI company founded in 2021 with a strong focus on AI safety. Like ChatGPT, it is mainly used as a standalone assistant through a website, desktop app or mobile app.

What it is

Claude works in the same way as other AI assistants. We give it a request in plain language, often along with documents or data, and it responds. It is available as a free version, paid personal plans and business versions for organisations.

What it is commonly used for

•         Long documents: reading and summarising lengthy reports, contracts or policies, and answering questions about them.

•         Writing: drafting and refining documents where tone and clarity matter.

•         Analysis: working through complex information and laying out options and trade offs.

•         Technical work: helping with coding and data tasks.

Working with our own material

Claude is often used by giving it source material, such as a report, a set of notes or a spreadsheet, and asking it to work from that. This connects directly with an earlier post in this series: AI gives better answers when we provide the information, rather than asking it to rely on general knowledge.

Things to keep in mind

•         It does not know our organisation. Context still has to come from us.

•         It can still be wrong. Careful writing does not mean every fact is correct. Review is still required.

•         Use the approved version. As with every tool, follow our organisation's policy on which tools and versions to use.

In a nutshell

Claude is a standalone AI assistant commonly used for long documents, writing and analysis. As with the others, results depend on the brief we give it and the checking we do.

Try this: If you have access to Claude, give it a long document and ask it for the three most important points and any open questions.

Comments

Post a Comment

Popular posts from this blog

Search Is Not the Same as Asking AI: Why How We Ask Matters

Getting average answers from AI? The problem is usually not the tool. It is how we ask. Most people tell me they have tried AI, but the answers they got were not that useful. In many cases, the issue is not the tool or the technology. It is how we ask the question. The good news is that a few small changes can make a big difference to the quality of the answers we get. We are used to search engines For years, search engines have trained us to work in a certain way. We type a few keywords, the search engine returns a long list of results, and we pick the most useful one ourselves. That works well for search, so it is natural to use AI the same way. But AI works differently. AI gives one answer, so it needs context Instead of a list of links, AI gives us a single, direct answer. To get that answer right, it needs to understand exactly what we are asking for. If we give it a keyword or a short, incomplete phrase, it has to guess. Who is this for? What matters most? How long should it be? ...
Read more

ChatGPT

  The tool that introduced most people to AI assistants. When people talk about AI, many of them mean ChatGPT. Launched by OpenAI in late 2022, it was the first AI assistant to reach a mass audience, and for many people it is still the first tool they try. What it is ChatGPT is a general purpose AI assistant. We use it through a website or a mobile app, typing or speaking our requests in plain language. It is not built into a specific set of work tools, which makes it flexible across many types of tasks. What it is commonly used for •          Writing and editing: drafting emails, reports and posts, or improving existing text. •          Thinking through problems: brainstorming options, structuring an argument or preparing for a difficult conversation. •          Working with files: summarising documents, analysing spreadsheets or reviewing images...
Read more

What AI Is Good At, and What It Is Not

  Using AI well starts with knowing where it helps and where it does not. In my last post, I wrote about how we ask AI questions. The next step is knowing which tasks to give it in the first place. AI is very capable in some areas and unreliable in others. Knowing the difference saves time and avoids mistakes. Where AI is strong AI is good at working with language and information we give it. In practice, that means: •          Drafting: first versions of emails, reports, proposals and presentations. •          Summarising: turning long documents or meeting notes into key points. •          Restructuring: reorganising rough notes into a clear format, such as actions, risks or decisions. •          Explaining: making a complex topic easier to understand, at the level we need. These tasks share something in ...
Read more