Token counter
Exact GPT-4o token counts, per-model cost, and how much of each context window a prompt actually fills.
/tokens
Pre-launch · 18 free tools, no account
The AI research workspace that remembers everything.
FlexContext turns documents, web searches, websites, videos and datasets into persistent, source-backed knowledge — so a research project accumulates instead of restarting in a new chat every morning.
The app: drop in files, say what you want, and one of the 18 tools does it
What do you want to do with your files?
Say it the way you would say it to a person. Your words are matched to one of 18 tools by readable rules, and the tool runs right here in your browser. No upload, no account.
esc
The workspace is not open yet. Try the free tools and labs in this tab, including a voice studio and a document chat that run their models on your own device.
- 18
- free tools and labs
- 28
- in-browser voices
- 0
- accounts required
Free, forever, no account
18 free tools for research, RAG and SEO
The same operations FlexContext runs internally, given away as standalone tools. Every one of them opens in your browser and processes your files locally, including the text-to-speech studio. Each has its own page if you want the detail.
Context & retrieval
Count tokens, split documents, and watch a retrieval ranking explain itself — the mechanics behind every RAG pipeline.
Text chunker
Compare fixed, sentence and Markdown splitting on your own text, with every chunk drawn to scale.
/chunk
RAG playground
A BM25 ranking with the arithmetic left visible: chunk map, term stats and per-term score contributions.
/rag
Ask your documents
Cited answers from your own PDFs and Word files, written by an open model running on your graphics card.
/ask
Summarizer
The sentences that carry a document, with their pages, or a table told in a few lines. Nothing is reworded or uploaded.
/summarize
Report builder
Summaries, tables, JSON and images, combined into one printable HTML page and a Markdown copy.
/report
Data & documents
Profile, filter, convert and validate the files research actually arrives in: CSV, JSON, TSV and plain text.
CSV profiler
Column types, nulls, uniques, distributions and correlations from a CSV, in one pass.
/analyze
Analytics builder
Select columns, filter rows, group by a field, aggregate a metric and chart the result.
/analytics
Format converter
Move a table between CSV, TSV and JSON, or encode and decode Base64, then download it.
/convert
JSON validator
Validate and format JSON, browse its shape, and get the exact line and column of any fault.
/json
Web & media
Turn pages, images and PDFs into something you can read, cite and reuse.
TTS voice studio
Turn text into natural speech with Kokoro, running in your browser. Save your favourite voice presets.
/tts
HTML extractor
Readable text, every link and any tables, pulled out of a page’s HTML source.
/extract
Image converter
Change format, dimensions and quality, compare the file size, and download — without uploading.
/image
PDF page editor
Merge documents, extract page ranges, reorder and rotate — on a PDF that never leaves your machine.
/pdf
Search & algorithm labs
Interactive models of the algorithms behind search, ranking and research agents.
Research agent simulator
Turn the dials a deep-research agent actually has and watch coverage, cost and confidence move.
/agent
PageRank simulator
Describe a link graph, set the damping factor, and watch authority converge iteration by iteration.
/pagerank
Graph explorer
Run breadth-first and depth-first traversal over the same graph and compare the journeys.
/graph
Trie visualizer
Build a prefix tree from your own vocabulary and follow autocomplete narrowing, character by character.
/trie
Why it exists
Information → evidence → knowledge → deliverables
Research is not a single question. It is a body of material that should get more valuable each time you return to it — and in a chat window, it does the opposite.
-
01
Collect
Information
Documents, scanned PDFs, web pages, search results, YouTube transcripts and datasets land in one project instead of six browser tabs.
-
02
Verify
Evidence
Claims get attached to the passage they came from, contradictions between sources become visible rather than averaged away.
-
03
Structure
Knowledge
Notes, annotations, entities and datasets accumulate into a project that is worth more in month three than it was in week one.
-
04
Produce
Deliverables
Cited reports, structured datasets and — further out — dashboards, knowledge graphs and explainer video from the same material.
What we are building
The tools above, pointed at your own library
FlexContext is a workspace where sources, tools and conversation live together — and where you can always see exactly what the model was given.
A library that is actually yours
Documents, images and video in your own space, browsable as a file tree. Every type opens in place — PDF, video, images and agent output.
Citations onto the real document
Answers cite your sources, and clicking one opens the actual PDF or Markdown at the chunk it came from — not a paraphrase of it.
Tools that run under visible limits
Crawling, media work and analysis run in a bounded sandbox — a fixed core and memory ceiling per job, enforced by the OS, with the CPU cost reported back to you.
Every run stays in the conversation
A tool run is not a disposable message. It keeps its settings and its result as a block you can read later, or clone into a new run with one click.
Slash commands, including your own
Call any tool with / from the composer. Save a configuration you use often as your own command and it is there next time.
The model gets no surprises
It answers on what you selected, and it cannot reach your filesystem or run anything unasked. When a tool would help, it suggests the command — you decide.
Who it is for
Built for research-intensive work
Anyone whose output depends on sources they have to keep, compare and cite.
Analysts and consultants
Market sizing, competitive intelligence and due diligence, where the deliverable is only as good as the sources behind it and every figure has to be traceable.
Journalists and researchers
Document sets that grow for months, contradictory sources that need comparing rather than blending, and citations that must survive a fact-check.
Founders and product teams
Continuous competitor, pricing and customer research that accumulates into something the whole team can query instead of one person’s notebook.
SEO and content strategists
SERP behaviour, ranking structure, internal linking and content research — several of the labs on this page exist because that work needed them.
Students and academics
Literature that has to be read, cited and returned to, with the reading of month one still usable in month six.
Developers building RAG
Chunking, retrieval, context budgets and evaluation — the tools on this page are the same operations, exposed rather than hidden.
Not open yet — and we would rather say so
FlexContext is in active development. Rather than collect your email for a launch with no date, the tools above are simply free to use right now, with no account. Here is where things honestly stand.
- Per-user storage, sandboxed tools, slash commands and response blocks — built and tested
- Chat with a real model, answering only on the sources you pick
- The 18 tools and labs on this page — free, and staying free
- RAG citations onto the source document — in progress
- Hosted accounts and sign-up — not yet open
- Dashboards, knowledge graphs and explainer video — on the roadmap, not built
Questions
Frequently asked questions
What is FlexContext?
An AI research and intelligence workspace. It collects documents, web pages, search results, transcripts and datasets into a persistent project, keeps every answer tied to the source it came from, and turns that accumulated material into reports, datasets and other deliverables — instead of losing it at the end of a chat.
Are the free tools really free?
The tools need no FlexContext account or subscription. Every tool runs in your browser, including the TTS studio, so there is no per-use cost.
Do the tools upload my files?
No. Data and document tools process files in your browser, and the TTS studio generates speech in your browser too; its voice model is downloaded once from Hugging Face. Voice recipes are saved on this device.
How is this different from ChatGPT or a chat assistant?
A chat assistant starts from nothing each time and decides for itself what context to use. FlexContext is built the other way round: your sources persist across sessions, you control exactly what the model is given, and every claim carries a citation that opens the original document at the passage it came from.
Can I sign up or join a waitlist?
Not yet. The product is not open, and a signup form implying otherwise would be dishonest. The tools are what is on offer today.
Why give away 18 tools before launch?
They are the same operations the product runs internally — chunking, retrieval scoring, token accounting, extraction, profiling. Giving them away is a more honest preview than a mockup, and it is useful to people whether or not the product ever ships to them.