Productivity content has a credibility problem. Every list claims its tools save you "20 hours a week" - usually written by someone who has not personally used most of them. This is the small list of AI tools I actually use every week, the rough hours each one saves me, and where I would not bother.
The headline figure (8 hours) is honest. It is what I clawed back over the first two months of integrating these tools into recurring work, measured not by stopwatch but by the much simpler test: I now finish on Thursdays what used to take me until Friday afternoon.
What "Productivity" Actually Means with AI
Most "AI productivity" tools fail because they target the wrong layer. They speed up tasks you should not be doing in the first place: writing meeting notes nobody reads, drafting emails that should have been a Slack message, summarising documents you did not need to open.
The right targets for AI productivity are the activities that satisfy three conditions: recurring, structured, and below your pay grade. Meeting notes from a call you led. The first draft of a quarterly report. Sorting your inbox into "respond now" and "respond Friday". Compressing a 40-page PDF into a one-page brief. These are tasks AI does competently and you do not need to.
The 10 tools below all hit those criteria.
1. Meeting Capture: Granola (~2 hours/week saved)
Granola joins your calls in the background, transcribes them, and produces a structured summary the moment the call ends - action items, decisions, key quotes. The interface is just a notes app; you take notes during the call as normal, and Granola merges your notes with the transcript afterwards.
I tried Otter, Fireflies, and Read AI before Granola. Granola wins on three things: it does not send a bot to join the meeting (privacy and friction win), the merged note format is genuinely usable, and the action item extraction is accurate enough that I trust it.
Cost: £15-18/month after a generous free tier.
Where it earns its keep: 4-6 calls a week × 10-15 minutes of post-call admin = 1-2 hours.
2. Email Triage: Superhuman AI (~90 minutes/week saved)
Superhuman is a paid email client with native AI features - auto-categorisation, AI-drafted replies in your voice, and "split inbox" rules that surface the 5 emails that actually matter while silencing the 95 that do not.
The price tag (~£25/month) puts most people off. The honest read: if email is a daily bottleneck, it earns its keep within two weeks. If you are someone who checks email 3 times a day and clears it in 20 minutes, save your money.
Free alternative: Shortwave has a generous free tier with similar AI features and is genuinely good. I switched between the two for months; Superhuman shaves another 5-10 minutes a day off Shortwave for me but it is close.
3. Research: Perplexity (~90 minutes/week saved)
Perplexity replaces Google for any research task that needs current information with sources. Ask it a question, get a summarised answer with the actual citations underneath. You stop opening 7 tabs to triangulate.
For deeper research projects, Perplexity has a "Pro Search" mode that does multi-step reasoning over sources. For factual lookup, the free tier is enough.
Cost: Free tier solid; Pro at £15/month if you research daily.
Where it earns its keep: Any task where you would have spent 20 minutes Googling, gone in three directions, and ended up with one usable paragraph. Now: 5 minutes, properly sourced.
4. Writing: Claude (~2 hours/week saved)
Claude is my writing partner for anything longer than a paragraph: blog drafts, long emails, course content, briefs, proposals. It writes in something close to a natural register and follows constraints well, which matters when you need a 280-word cover letter rather than a 700-word essay.
I keep a small library of "system prompts" - pre-written instructions that frame Claude for specific recurring jobs (Hus voice, Highland Code voice, course module structure). Loading the right system prompt is the difference between a useful first draft and a generic one.
Cost: £15/month for Claude Pro.
Companion: ChatGPT Plus (£20/month) for tasks where Claude is weaker - image generation, custom GPTs, structured outputs. See our full comparison.
5. Note-Taking: Notion AI (~45 minutes/week saved)
Notion AI sits inside Notion documents and does the small writing tasks: summarise this page, turn these notes into a memo, draft an action plan from this brainstorm. Because it is in the place where my notes already live, I actually use it. Standalone "AI note tools" tend to gather dust.
Cost: £8/month add-on to a Notion plan.
Alternative: If you live in Obsidian, the Smart Connections and Copilot plugins give you similar functionality with local-first privacy.
6. Calendar & Scheduling: Reclaim AI (~60 minutes/week saved)
Reclaim plugs into Google Calendar and automatically defends focus time, schedules recurring tasks, finds meeting slots that respect everyone's preferences, and reschedules when conflicts arise. The "habits" feature blocks 90 minutes a day for deep work and quietly reorganises around it.
The genuine win is fewer scheduling-by-email loops. Reclaim's scheduling links are smart enough that 80% of meetings book themselves.
Cost: Free tier usable; £8/month for the full feature set.
Alternative: Motion is similar with stronger task-priority logic but is overkill if you just want better calendar defence.
7. Document Analysis: NotebookLM (~30 minutes/week saved)
Google's NotebookLM lets you upload up to 50 sources (PDFs, docs, websites, YouTube videos) and ask questions across them with proper citations back to the original passages. It is the tool I reach for whenever I have to read a long document and produce a brief.
The killer feature is the "audio overview" - it generates a 12-15 minute conversational podcast about your sources that you can listen to on a walk. For pre-meeting prep, this is the highest-leverage use of AI I have found.
Cost: Free.
8. Coding & Automation: Cursor or Claude Code (~varies)
Caveat: only relevant if you write code or run technical workflows. Cursor is a code editor with a built-in AI assistant. Claude Code is Anthropic's terminal-based coding agent. Both have shifted what one engineer can ship in a day.
For non-coders this section is irrelevant. Skip it. For anyone who works in spreadsheets, Excel's Copilot integration is the closest equivalent and worth turning on.
9. Voice Capture: Otter.ai or Whisper (~30 minutes/week saved)
For one-off voice memos and conversations that I want transcribed but where Granola is overkill (a chat in the kitchen, a voice note to myself in the car), Otter's free tier is fine. If you want best-in-class transcription with no upload limits, OpenAI's Whisper running locally is free and excellent.
10. Connective Tissue: ChatGPT Custom GPTs (~30 minutes/week saved)
I have built five small custom GPTs that handle recurring jobs: a "weekly review" GPT that asks me the same five questions every Friday, an "email cleaner" GPT that rewrites my drafts in a warmer register, a "meeting prep" GPT that builds a one-pager from a calendar invite plus relevant docs.
None of these are products. They are small, persistent prompts I do not have to retype. The aggregate saving is the largest of any single tool on this list because they cover the long tail.
My Weekly Stack
Here is roughly how it fits together across a normal week.
- Monday morning: Reclaim builds my week. Granola captures the kickoff calls. NotebookLM briefs me on a long doc I needed to read.
- Tuesday-Wednesday: Claude drafts long-form work. Superhuman triages email twice a day. Perplexity handles ad-hoc research.
- Thursday: Catch-up. Custom GPTs run weekly reviews and tidy admin.
- Friday: Reclaim auto-rolls unfinished tasks into next week. Granola summaries get filed. Claude drafts the week-ahead notes.
None of this is heroic. It is just one small tool removing one specific friction at each step. The 8 hours a week is the cumulative effect, not any single tool's contribution.
Tools I Stopped Using
- Notion AI as a standalone writing tool. Useful inside Notion, weak as a Claude/ChatGPT replacement.
- Mem.ai. Smart concept, but I never built the daily habit.
- Heyday and Rewind. Solid privacy concerns and the "search your life" promise oversold the daily value.
- Most browser extension AI assistants. They duplicate what Claude/ChatGPT already do, with worse interfaces.
How to Get to 8 Hours/Week Yourself
The mistake people make is buying all the tools at once and then feeling overwhelmed. The compounding works in the opposite direction.
- Pick the bottleneck. Where does your week leak the most time? Email? Meetings? Research? Writing? Pick one.
- Add one tool. Use it daily for two weeks before adding anything else.
- Measure. Did this give you back 30+ minutes a day? If yes, keep it. If no, drop it without shame.
- Repeat. Add the next tool only when the previous one has become invisible - a habit, not an effort.
For broader workplace context, see our guide to using AI at work. And if you are starting from zero, our learn AI from scratch roadmap covers the underlying skills.
Frequently Asked Questions
How long until I see real time savings?
Honest answer: week 3 or 4. The first two weeks are setup and learning the prompt patterns. By week 4 most people see 3-5 hours back; by month 3, 6-10 hours if they have integrated the tools into recurring workflows.
Are these tools privacy-safe for client data?
It depends on the tool and the plan. For client data, use enterprise or business plans and check the data processing terms. Default-deny: do not paste client data into anything until you have read its privacy policy.
Should I pay for all 10 tools?
No. Most professionals pay for 3-5 and use free tiers for the rest. Start with the tool that solves a daily bottleneck. Add more only when you can name the specific weekly task they will accelerate.
What if my company blocks AI tools?
Many companies have approved enterprise AI tools and blocked unmanaged consumer ones. Find out what is approved before installing anything. Using a blocked tool with company data is a fireable offence in some sectors.
The Skill Behind the Stack
Tools change. The thinking does not. AI Mastery teaches you the durable skill of getting useful work out of any AI tool - 20 modules, 130+ lessons, 500+ tested prompts.
View Course & Pricing →