Conference Prep: How I Read 8 Talks Worth of Docs Before DevDay

DevDay is in two weeks.
The schedule has 47 talks across three tracks.
I want to actually understand 8 of them.
The rest, I'll happily watch the recordings later. The 8 I want live: those, I want to be in the room for, prepared, with my brain on, asking the right question afterwards.
That requires prep. Here's how.
Table of Contents
- The Conference Documentation Gap
- The System
- A Real Conference Prep List
- Pre-Fetching Speaker Docs
- Tablet-Only Conference Workflow
- The Note-Taking-During-Talks PDF Setup
- What If I Only Have 3 Days Before the Conference?
- Build Your Conference Reading List
The Conference Documentation Gap
Walking into a talk cold means:
- The first 10 minutes are jargon you don't understand
- The middle is interesting but you can't follow the depth
- The end is "okay so what?" and you missed the takeaway
Walking in prepared means:
- The first 10 minutes confirm what you know
- The middle is the new thing you came for
- The end is your question to the speaker
The difference is reading the relevant docs in the two weeks before.
For most conferences, this isn't realistic by web browsing alone. Too many tabs, too many context switches. PDFs change the math.

The System
Two weeks out from the conference, I do this:
- Print the schedule
- Pick the 8 talks I want to be in the room for
- For each talk, identify the related libraries, frameworks, or papers
- Generate a PDF for each
- Read them on my commute / lunches / evenings
That's it. The whole prep system.
The catch is step 4. If I had to manually fetch and bundle each one, I wouldn't do it. With OfflineDocs, each PDF is a 90-second paste-and-generate.

A Real Conference Prep List
Here's an actual list from an AI conference I prepped for last year:
| Talk | Related docs to read |
|---|---|
| "RAG in production at scale" | LangChain core, vector DB primer |
| "Streaming agents with the AI SDK" | Vercel AI SDK reference |
| "Anthropic's Contextual Retrieval" | The Contextual Retrieval paper, Anthropic API |
| "Bun for AI workloads" | Bun docs |
| "Postgres for vector search" | pgvector docs |
| "Eval frameworks compared" | Each framework's eval docs |
| "Multi-modal pipelines" | Whisper docs, image embedding docs |
| "AI infrastructure at $bigco" | Their open-source SDK docs |
8 talks. ~12 PDFs (some shared across talks). Total reading: ~600 pages.
In 14 days, that's 40-50 pages a day. Roughly an hour of reading. Doable on a commute.

Pre-Fetching Speaker Docs
If a speaker has shipped a popular library, the docs for that library are usually the prep material.
Most conference talks are 30 minutes of "context everyone in the room already has, plus 10 minutes of new thinking." If you're not in the "everyone already has" group, you've started 20 minutes behind.
Reading the speaker's library docs in advance closes that gap.
For talks based on papers, read the paper. For talks based on internal company tooling, read whatever they've open-sourced. For talks based on industry trends, read the canonical write-ups.
Generate each as a PDF. Save in a "DevDay 2026 prep" folder. Read on commute.

Tablet-Only Conference Workflow
I don't bring a laptop to conferences anymore.
Tablet only. iPad mini with an Apple Pencil.
| Tool | Use |
|---|---|
| Reader app (Books / GoodNotes) | Read prep PDFs, take notes during talks |
| Slack mobile | Async coordination |
| Email mobile | Quick replies |
| Phone for everything else | Hotspot if needed |
The laptop is too tempting. You open it for "one quick thing" and miss the talk.
Tablet-only also forces lighter packing, less battery anxiety, and more conversations between sessions. All good.
For a deeper take on tablet-based reading, the e-ink devices guide covers Boox / Kindle Scribe / reMarkable alternatives if you want zero distractions.

The Note-Taking-During-Talks PDF Setup
I open the PDF for the relevant prep material before each talk.
When the speaker references something, I jump to it in the PDF. Margin note: "she mentioned this re: rate limiting." When the speaker says something I don't follow, I write the question I'd ask: "did they tune the embedding model on their own data?"
Half the speakers will answer my question themselves later in the talk. The other half: I ask during Q&A, or I find them at the coffee break.
This works because the PDF is right there. If I had to switch to a browser to look something up, the talk would move on.
What If I Only Have 3 Days Before the Conference?
Cut the list.
3 days = read for 4 talks instead of 8. Or read shallower for all 8 — skim the docs instead of reading them carefully.
Either way, something is better than nothing. Going in with even surface-level prep beats going in cold.
This is similar to interview prep — perfect prep doesn't exist, enough prep does.
Build Your Conference Reading List
Two weeks before your next conference:
- Pick your top 8 talks
- Identify the docs for each
- Paste each URL at offlinedocs.ai/new
- Generate
- Load on tablet
- Read on commute
That's the system. Three minutes of setup per PDF. Two weeks of better-prepared commutes.
You'll ask better questions. You'll understand more. You'll have better conversations between sessions.
The conference will feel different. Try it next time.
Ready to Get Started?
Start creating your Offline Docs Now! Reduce screen time and save your eyes.