Free tools Windows power users keep installed
One-click scans. No signup required.
Retrieve whole notes when a rule only makes sense with its reason or its exception attached. Retrieve chunks or snippets when documents are short or queries are narrow. That is the conditional position Tom Jones takes in “Whole notes, not fragments: the retrieval half,” published 2026-09-18. His evidence supports the condition more firmly than it supports a blanket claim that whole notes are better.
What the described system retrieves
The design stores each note as plain Markdown with a short header. Every whole note is embedded as one unit. At query time, the user’s question is embedded and notes are ranked by cosine similarity, so the object returned is the full note, not the passage that happened to match.
As an Amazon Associate I earn from qualifying purchases.
The implementation details the article reports are as follows. These describe one author’s setup, not a compatibility guarantee for any later version of the tools:
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
- Embedding model:
nomic-embed-text, run locally through Ollama. - Vector store: a SQLite table using the
vec0extension. - Keyword path: a separate full-text index, aimed at exact tokens such as function names, command-line flags, and error strings.
- Fusion: dense results from the vector table and keyword results are combined into one ranked list.
The split between the two paths is deliberate. Dense retrieval handles paraphrase, where the user describes a problem in different words than the note uses. Keyword matching handles exact strings that embeddings can blur.
#1 Best Overall
- Stay present in every scenario: Every conversation is covered, in person, on calls, and online. 4 MEMS + 1 VPU microphones with AI beamforming capture every voice across the room. Smart Dual-Mode Recording switches automatically between phone calls and in-person. The free Plaud Desktop captures online meetings without a bot
- Walk out of every meeting with notes ready to act on: Plaud Intelligence transcribes in 112 languages with speaker labels and turns each recording into action items, decisions, and follow-ups, structured and ready to use. Choose from 10,000+ customizable templates tailored to your role and industry
- AI summary ready before you reach your desk: Auto Transfer moves each recording to the Plaud app automatically, and AutoFlow transcribes and summarizes so your notes are ready before you are back at your desk. Upgrade anytime to Pro (1,200 min/mo) or Unlimited
- Access your AI workspace anywhere: One connected workspace across Plaud Desktop, Plaud Web, and the Plaud mobile app, so your conversations and finished work follow you everywhere
- Your conversations stay private and yours: Compliant with ISO 27001, ISO 27701, SOC 2, HIPAA, GDPR, and EN 18031, with zero data used to train AI models. Trusted by 2.5M+ professionals, including legal, medical, and business professionals handling sensitive information
Why the retrieval unit changes what the reader gets
A chunk is a fixed-size or passage-level slice of a document. When a rule is split from its exception, the slice can be accurate and still misleading. Consider a note that says a migration must be run with a safety flag, then, two paragraphs later, says the flag can be skipped on an empty staging database. A chunk that matches the first sentence returns the instruction without the exception. A whole-note hit returns both, along with the header that says what the note covers.
The cost is payload. A whole note carries neighboring text that may not answer the question, so each retrieved token carries less relevance on average. Whether that trade pays off depends on how long the notes are and how much context the reader can use.
The public SciFact comparison: a tie
On the SciFact benchmark, the article reports that whole-note retrieval and chunked retrieval scored effectively the same on nDCG@10. The author’s explanation is structural: SciFact abstracts are short, so a chunk is already close to the full document. When the unit of retrieval and the unit of meaning nearly coincide, returning the larger unit adds little.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Rank #2
- AI-POWERED TRANSCRIPTION & SUMMARIES: Plaud Note Pro is your professional voice transcriber, delivering high-accuracy transcription in 112 languages with auto speaker labels. Powered by top AI models and thousands of templates, Note Pro instantly creates structured summaries, mind maps, To-Do lists, and proposals tailored to your role and industry
- ENHANCED CONTEXT WITH MULTIMODAL INPUT: Capture audio, type notes, add images, and press to highlight key moments for richer context. During recording, instantly mark key moments with a single button press. Simultaneously enrich your audio by snapping photos of important documents or typing in ideas
- CHAT WITH YOUR RECORDINGS USING "ASK Plaud": Unlock deeper insights with this interactive AI. Ask questions, extract key points, draft emails, and get next-step suggestions—all grounded in your original audio for reliable, ready-to-use answers
- INTELLIGENT RECORDING WITH AI DIRECTIONAL AUDIO: Enjoy seamless, intelligent recording with Plaud Note Pro. Its AI automatically switches between call and meeting modes while recording, while directional audio and real-time spatial awareness minimize noise to capture voices with crystal clarity
- Everything Included: Includes Plaud Note Pro, magnetic case, magnetic ring, charging cable, and a free Starter Plan with 300 transcription minutes per month. Upgrade anytime in the Plaud app to Pro Plan (1,200 min/mo) or Unlimited Plan(Up to 24 hours of transcription per user per day)
This matters because it rules out an unconditional whole-note advantage. A reader who adopts whole-note retrieval for short scientific abstracts should not expect a measurable gain from the benchmark alone. The article also reports a keyword-search control on SciFact, which scored lower than both dense arms (exact figures in the table below).
The NFCorpus comparison: whole notes against keyword search
On NFCorpus, the article compares whole-note retrieval with keyword search across 323 queries. Whole-note retrieval scored higher. That is a different comparison from the SciFact one: the baseline is keyword search, not chunked dense retrieval, and the article does not report a chunked figure for NFCorpus. A reader should not read the NFCorpus result as evidence that whole notes beat chunks on that dataset.
The internal 14-task evaluation
The article also reports an internal evaluation of 14 tasks, scored by a language model, comparing whole-note retrieval with standard snippet retrieval. Whole-note retrieval scored 52% and snippet retrieval 27%. The author states that this evaluation cannot be rerun externally. Treat it as an author-reported internal result on a small, private task set. It shows that the approach performed well on those tasks; it does not establish a general performance gain, and it should not be summarized as “nearly doubling” accuracy.
Rank #3
- YOUR AI PERSONAL ASSISTANT FOR EVERYDAY PRODUCTIVITY: More than a voice recorder, Pocket works as your AI personal assistant to capture, transcribe, and summarize meetings, calls, and ideas instantly. Core features are included out of the box, with optional advanced tools available for power users.
- ONE-TAP RECORDING FOR REAL-LIFE MOMENTS: Capture meetings, phone calls, and in-person conversations instantly with a simple tap, no typing, no interruptions, just effortless note-taking anywhere you go.
- SMART AI INSIGHTS & ORGANIZATION: Pocket automatically turns recordings into clear summaries, key action items and structured conversation maps so you can quickly review what matters without digging through audio.
- TURN CONVERSATIONS INTO ACTION WITH “ASK POCKET”: Don’t just record, understand. Instantly ask questions across your meetings, extract key insights and generate next steps in seconds. All grounded in your recordings, so answers stay accurate and reliable.
- MAGSAFE COMPATIBLE FOR SEAMLESS USE: Easily attach Pocket to your iPhone or other MagSafe compatible devices for convenient, hands-free recording on the go. Perfect for capturing meetings, calls, and ideas without needing to hold your device.
Reported figures at a glance
| Evaluation | Metric and scope | Whole-note result | Comparison | Reproducibility |
|---|---|---|---|---|
| SciFact (public) | nDCG@10 | 0.7014 (author’s three runs of the public arm: 0.7014, 0.7019, 0.7014) | Chunked retrieval: 0.7016. Keyword-search control: 0.6644 | Public benchmark; the author’s harness described in the article |
| NFCorpus (public) | nDCG@10, 323 queries | 0.3417 | Keyword search: 0.3098. Chunked figure not stated | Public benchmark; the author’s harness described in the article |
| Internal (14 tasks) | Task success, scored by a language model | 52% | Standard snippet retrieval: 27% | Not independently rerunnable; the author says so explicitly |
All figures are from Tom Jones, “Whole notes, not fragments: the retrieval half,” published 2026-09-18. The SciFact and NFCorpus runs are described by the author; no independent replication of any figure in the article is available as of this writing.
Where the condition holds
The author’s proposed boundary is long documents in which a rule’s exception, rationale, or scope sits apart from the rule itself. Short documents have little room for that to happen, so chunking loses little. This is the article’s interpretation of its own evidence, and the article presents it that way.
The article also reports a difference by reader. A large-context model did better with whole notes. A small local model did better with compact records. The source does not establish this as a universal rule, so it is best treated as a reason to check the reader’s context budget before choosing a unit, not as a fixed setting.
Rank #4
- Plaud Intelligence: Capture conversations in 112 languages and generate accurate transcripts with the Plaud App and Web. Plaud Intelligence uses leading models like GPT-5.5, Claude Sonnet 4.6, and Gemini 3.1 Pro to transform raw audio into structured insights. Choose from over 10,000 professional templates to generate mind maps and to-do lists, turning hours of discussion into immediate clarity
- Multiple Ways To Wear With Included Accessories: Adapt Plaud NotePin S to any workflow instantly with four included accessories. Wear your device effortlessly as a necklace, wristband, clip, or pin. Plaud NotePin S features a dedicated physical record button for precise, tactile control. Stay professional and keep your intelligence within reach all day
- Enterprise-grade Privacy: Built to the highest standards with ISO 27001/27701, SOC 2, HIPAA, GDPR, and EN18031 compliance. Every conversation is secure and protected. It is the trusted choice for creative, medical, and business professionals handling sensitive info
- Multimodal Input & Multidimensional Summaries: Capture audio, type notes, add images, and press/tap to highlight for richer context with multimodal input. Press the record button to mark key moments in real time. Plaud transforms a single conversation into multiple perspectives, providing faster, clearer insights, and unifies these inputs to deliver role-specific summaries that reflect your intent and priorities
- Lightweight Power and Peace of Mind: Weighing only 0.61 oz, Plaud NotePin S delivers 20 hours of continuous recording and 40 days of standby time. Store up to 64GB of audio locally, ensuring you capture every insight even without an internet connection
Rules that must always apply
The author separates two kinds of content. Knowledge can be found by similarity search. Rules that must always apply should not depend on search at all. The article’s position is that such rules are loaded unconditionally into context. As the author puts it: “Safety rules are never retrieval-gated.”
This is a design position from one author, not a formal safety standard. If a rule failing to appear would cause harm, the article’s recommendation is to remove the decision from the retrieval step.
A decision framework
- A rule depends on a reason or exception in a long note: retrieve whole notes.
- Notes are short abstracts or self-contained entries: chunks or snippets are likely enough, and the SciFact tie suggests you will not lose much.
- Queries depend on exact names, flags, or error strings: keep the keyword index in the fused pipeline rather than relying on dense similarity alone.
- The rule must apply every time: load it unconditionally instead of retrieving it.
- The reader is small and local with a tight context budget: compact records may serve better, according to the article’s observation of one local model.
Keep the evidence limits in view when adopting any of these. The public benchmarks are reported by the author with the harness he describes, and the internal result cannot be checked by others.
The article’s core claim is conditional, and that is its strongest feature. The author writes: “The condition is the useful half, so the condition is what we are handing you.”
Source: Tom Jones, “Whole notes, not fragments: the retrieval half,” published 2026-09-18.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Recommended Free Tools
