When are Deep Research and Thinking Mode worth it?

When a question needs several sources and a comparison. Not when there's only one answer.

In 3 steps

  1. Name the purpose: Describe not just the topic but what you need the report for – "for a make-or-buy decision, cost and data protection matter most".

  2. Set the criteria: Say what should be compared. The criteria give the report its structure.

  3. Start it and let it run: amber keeps working in the background; you can leave the page.

What comes out

A structured report. Depending on the question, with a SWOT analysis, competitor comparison and strategic roadmap – which is why Deep Research suits market overviews and decision papers particularly well.

Every statement in the report is linked to its source, and you export the result directly as a PDF or Word document.

Thinking Mode

In Thinking Mode, amber thinks the task through first before answering, instead of starting to write straight away. That takes a few seconds longer than the regular amber chat, but it pays off whenever an intermediate step could go wrong and you want amber to work through the problem with you rather than just give you a quick answer.

When it's worth it

  • Calculations and figures you're unsure about

  • Tasks with several conditions ("only the contracts signed before 2024 and renewed automatically")

  • Reviewing code, following logic, finding contradictions in a document

  • Decisions where you want to see the reasoning, not just the result

When it isn't: for short questions, summarising and wording. There, Thinking only costs time without making the answer any better.

Thinking Mode or Deep Research?

Thinking Mode

Deep Research

Helps with

tricky tasks, multi-step reasoning, code, calculations

questions that need many sources brought together and compared

Duration

a little longer than a normal answer

several minutes to more than ten

Result

one answer, worked out more carefully

a structured report with source references

Example

"Check this calculation step by step and tell me where the mistake is."

"Compare three providers on cost, data protection and integration effort."

→ Thinking Mode when the question is hard.
→ Deep Research when the answer is scattered far and wide.

For everything else, the normal chat is faster and does the job perfectly well.

Good to know

Deep Research takes time. A few minutes is normal; with complex requests amber may still be working after ten minutes. That's not an error or a stuck progress bar – so don't start it if you need an answer in two minutes.

Both use more than a normal request. Longer processing means more tokens. If you want to know what that means for your daily allowance → How credit usage works at amber

You don't have to start recurring research yourself every time. A market overview every Monday morning runs as a scheduled task → How do I run tasks automatically?

Up next: How do I create my first agent?