my 2026 ai stack
seven tools that changed how i work
👋 Hey, I’m Suhas and welcome to this week’s edition of the newsletter. Each week I tackle questions about building products, driving growth, and accelerating your career.
Over the past year, AI has stopped being optional for everyone in tech.
Be it designers, engineers or PMs - everyone is vibe coding and shipping prototypes faster.
I experimented aggressively.
Copilots, assistants, roadmap generators, vibe coding tools and AI meeting bots.
Most fell out of my workflow within weeks.
They never helped with the hard parts of product management.
But some stuck around and have become part of my workflow coming into 2026.
The lens I now use:
Where does this fit in my workflow?
Why is it better than alternatives?
What behavior did it change?
That filter left me with seven tools.
Perplexity Comet: My personal assistant
Why I find it useful: I was spending 2-3 hours weekly researching competitors, screenshotting sections, and pasting into docs.
I now say: “Go to Product Hunt, find the top 5 PM tools launched this month, extract their taglines and pricing, put it in a sheet.”
It opens the tabs, extracts the data, formats everything, done in 5 minutes.
Why alternatives didn’t work: ChatGPT Atlas struggles with complex workflows. Dia can’t do automations yet and its Skills feature doesn’t fit my workflows.
What changed: I have started using my browser like an agent. A lot of the grunt work is automated now.
Best for: Anyone drowning in repetitive browser tasks or manually aggregating information across multiple websites.
Claude Code: Claude with hands
Why I find it useful: I accumulate a lot of local data: meeting transcripts, research docs, project files, and exported CSVs.
Getting it and doing something with it is tedious, like summarizing trends, finding patterns, and organizing files.
Why alternatives didn’t work: Zapier and Make.com don’t touch your local files.
Claude Code runs locally in your terminal, can read your files, run commands, and execute workflows directly.
What changed: If I can describe a task clearly, I can usually automate it in minutes by simply asking.
Best for: Anyone who wants local automation without engineering help or is tired of repetitive computer tasks.
Gamma: idea to presentation in 5 minutes
Why I find it useful: Gamma handles the entire formatting and design work automatically.
What I love most is that it generates presentation-ready slides with proper structure, visual hierarchy, and professional polish.
Why alternatives didn’t work: Google Slides means hours of manual formatting. Canva’s templates are built for social media, not business presentations.
What changed: The time saved creating a deck has gone from hours to minutes.
It frees up your bandwidth to focus on the presentation material and not worry about its appearance.
Best for: Anyone who needs professional presentations fast without spending time on design and formatting.
n8n: When tools refuse to talk
Why I find it useful: I am subscribed to at least 10 other AI and product newsletters.
Every week, I’d manually skim through them, copy interesting bits into Notion, and try to remember what was relevant.
It was scattered, inconsistent, and I’d often forget key insights by the time I needed them.
I built an n8n workflow that monitors my inbox, pulls the new newsletters, sends them to Claude to extract and summarize, stores it in a Notion database with tags and posts a weekly digest to our Slack.
Why alternatives didn’t work: Zapier would’ve cost $80+/month, writing custom scripts meant maintenance hell and Make.com was close but n8n’s visual debugging made it easier to troubleshoot, plus the self-hosting option gave me control.
What changed: Information actually gets captured and used instead of lost. Now I automate everything: customer feedback flows, bug reports, auto-create tickets, etc.
Best for: Anyone doing repetitive data transfers between tools or building custom workflows that would be expensive on per-task pricing platforms.
Wispr Flow: writing at the speed of thought
Why I find it useful: My typing speed is ~80 WPM but I speak at ~160 WPM.
That’s 2x faster, and Wispr captures it with extremely low error rates.
I now dictate specs while walking, capture ideas between meetings, and clear my inbox without touching a keyboard.
Over 50% of my writing happens through voice now.
Why alternatives didn’t work: Apple’s dictation is a nightmare. It has constant errors, no punctuation, and requires heavy cleanup. Other tools had similar accuracy issues.
What changed: Ideas get captured immediately instead of being lost to context switches.
Best for: Anyone in meeting-heavy roles who loses ideas waiting to type them, or spending hours daily writing and wants that time back.
Emergent: Vibecode fast, and have fun with it
Why I find it useful: Emergent excels at product building.
It handles complex user flows, multi-step onboarding sequences, authentication systems, and form logic very well.
Why alternatives didn’t work: Bolt and Lovable break down with product complexity.
They’re optimized for simpler websites, not the kind of multi-step flows and state management that actual products require.
What changed: Product validation happens with working prototypes instead of static decks. Teams can experience and give feedback on real functionality.
Best for: Anyone validating product ideas before committing engineering resources or building MVPs that need real functionality, not just visual design.
Granola: A personal favourite
Why I find it useful: This is a personal favorite.
I live in meetings and the challenge is in remembering its context and recalling the right insight from it weeks later.
Granola remembers context across conversations and lets you query that memory later.
Ask “what did we decide about the pricing model?” and it pulls from every relevant discussion.
Why alternatives didn’t work: Otter gives transcripts i.e, walls of text I never revisit. Basic transcription tools lack the contextual layer that makes information actually retrievable and useful.
What changed: Pre-meeting archaeology stopped being necessary. Context switching between meetings became effortless because the memory layer is always accessible.
Best for: Anyone in meeting-heavy roles or managing long-running initiatives with multiple stakeholders across many conversations.
How to find yours
Don’t adopt a stack. Look for friction.
Notice where you hesitate before making a call, where discussions keep looping, and where insights fail to become action.
Pick one tool that attacks that moment.
Use it for real work for seven days.
If it makes decisions easier, keep it. If not, drop it.
That’s how my stack shrank from dozens to seven.
Upcoming Events
Behind the Scenes at Lovable: Scaling AI Products in the Real World
📅 Jan 17
Learn How Voice AI Products Are Built — Then Build One Live 🚀🎙️GC Hyderabad Edition
📅 Jan 24
From 0 → IPO: Real Product Lessons from a PW PM (GrabChai Delhi)
🗓️ Jan 31
Jobs Jobs Jobs
Senior Product Manager
Lead Product Manager
Director Product
Roles open across top companies like PhonePe, Swiggy, Groww & many more.
Heard anything cool in the world of PM that I should know about?
Drop me a note, always curious to learn.
Suhas 👋
Love the filter criteria: workflow fit, superior to alternatives, and actually changes behavior. That's the real test - does the tool stick?
Claude Code definitely passed that test for me. Started using it for coding, but ended up building an entire personal AI agent system (Wiz) that handles email, job searching, blog writing, and relationship tracking. It runs 24/7 on my Mac. The "local file automation without engineering expertise" you mention is exactly right - though I'd add that it grows with you as your ambitions expand.
Interesting that n8n and Claude Code are both on your list. I've been thinking about whether to add n8n for certain workflows, but Claude Code handles most of what I need. Curious if you've found use cases where one clearly beats the other.
The Granola mention is intriguing - maintaining context across conversations is exactly what I built manually with markdown memory files. Might be worth checking out.
Wrote my honest take on Claude Code (what actually works vs the hype) here if you want another perspective: https://thoughts.jock.pl/p/claude-code-review-real-testing-vs-zapier-make-2026
Good one about AI Tools - helps with personal productivity.
Curious about no mention on NotebookLM!