Get early access
How it works

Three choices, and together they are the whole thing

A chatbot sends your question to one very large model in a data centre, which searches through a paid service and hands the finished paragraph back to you. You cannot point at a sentence in it and ask where that came from, because the paragraph carries no record of which page produced which clause. Three decisions take that apart.

Research Workbench Agents
  1. Discovery
  2. Orchestr.
  3. Task Giver
  4. Query Help
  5. Task Doer
  6. Evaluator
Six stages run in order. The three choices below are what makes that order possible.
Choice one of three

Your browser fetches the pages, so sites see a person

Today, an extension you install does the fetching. Every page request leaves your own machine, on your own connection, so a publisher sees the same request it would see if you opened the page yourself. Pages that turn away a data centre open normally for a person, so LoQuery reads sources a crawler cannot reach.

Because your browser does the fetching, no crawler of ours visits the pages you read, and no site has an address of ours to recognise. And the fair question about fetching through your browser gets a straight answer: LoQuery reads one page at a time, at the pace a person browses, defeats no paywall and no login, and the excluded-domains control keeps a run off any site you name. Your account and your run records do sit on our servers. The privacy page lists what is stored.

Research Workbench Sources

Sources Discovered 24 sources 8 domains

Most-used sources

  • en.wikipedia.org × 6
  • aviation-safety.net × 4
  • recherche-research.bac-lac.gc.ca × 4
  • www.canada.ca × 3
  • archives.gov × 2
  • nsarchive.gwu.edu × 2
  • www.baaa-acro.com × 2
  • ufocasebook.com × 1
The source dossier from one run over aviation records. Every page in it arrived through the browser of the person who started the run, and none of them was requested from an address of ours.

Cell values, sources and caveats here are real and open to checking. The scores, attempt counts and source totals show the shape of a run and get replaced once a captured session record lands.

Choice two of three

Extraction and judgment are separate calls, so nothing grades itself

LoQuery splits research into narrow jobs: one writes the searches, one pulls a single fact off a fetched page, one decides whether that page supports it. Every answer comes back in a fixed shape, checked by plain code, so you can tell which step produced a value and which step to check.

Every question put to the model is about a page that search already fetched, so the work is judgment over supplied evidence rather than recall from training. No value in the table comes from what the model remembers. Small open models are enough for that. A model grading its own work defends the answer it already committed to, which is why the judging step never sees how the extraction reasoned.

  1. 01

    Extract Task Doer

    Reads a fetched page and pulls out the specific value for one field, with the URL it came from. It does not score anything and it is not asked whether it did well.

  2. Verdict boundary

    One of exactly three places in a run where LoQuery stops to read what you typed, named for the verdict it leads to. The gate and the judge below are a single call, so whatever you sent while this item was in flight is taken here, all at once, before that call starts. Nothing is read once it is running, because acting on an instruction early costs more than making it wait.

  3. 02

    Gate Deterministic

    Plain code, no model. Six checks run over the extracted values. Dates in the future, dates outside the window you asked for, too many fields marked not-applicable, entities pulled off an aggregator page. Arithmetic, so it always fires the same way.

  4. 03

    Judge Evaluator

    A separate call with its own context. Is this actually the entity that was asked about? Is this source authoritative for this kind of claim? Does the date make sense for this event? It never sees how the extraction reasoned, which is the point.

  5. 04

    Verdict confirmed · with caveats · uncertain · mismatch

    A score, one of four verdicts, and typed caveats, each with a code and a severity rather than a sentence the model improvised. And verdict_source: evaluator.

  6. 05

    Override You, if you disagree

    Set the verdict yourself, or keep it and rewrite the reasoning. verdict_source becomes user or collaborative, and it stays that way downstream. The machine's original call is not overwritten, it is recorded as having been overruled.

Choice three of three

LoQuery writes its own instructions for every question

The code knows nothing about securities enforcement or aviation history. Before a search runs, LoQuery decides what it is looking for, which sources count as authoritative in that subject, and what each column has to contain. Nobody added a subject, a source list or a field rule.

Plans Task Plan

Aviation history · Task Plan 6 sections

  1. 1 · Research intent unspecified · event · aviation history
  2. 2 · Output fields date, official explanation, primary source, status, summary
  3. 3 · Research strategy DAG · 1 level
  4. 4 · Field limitations (none)
  5. 5 · Authoritative sources aviation-safety.net, gov, canada.ca, archives.gov, nasa.gov, mil
  6. 6 · Dynamic config summary domain: aviation history · denied_sources: 0

Enforcement · Task Plan 6 sections

  1. 1 · Research intent unspecified · company · enforcement
  2. 2 · Output fields regulator, penalty, date, filing, summary
  3. 3 · Research strategy DAG · 1 level
  4. 4 · Field limitations penalty: currency required
  5. 5 · Authoritative sources sec.gov, justice.gov, cftc.gov, gov, courtlistener.com
  6. 6 · Dynamic config summary domain: enforcement · denied_sources: 1
Two unrelated runs, one code path. Six slots each, both written before anything was searched, and the enforcement run required a currency on the penalty field where the aviation one had nothing to constrain. The source lists differ, and so do the denied-source counts.

Sources here are real; scores and attempt counts are illustrative until a session capture lands.

The subject decides which sources count as authoritative, and LoQuery works that out from the question before anything is searched. When a subject gives it no institutions to name, the plan leaves the source list empty and flags the subject as uncalibrated.

Beyond the choices · 1 of 5

Two ways in, and the one you pick decides what you type

A run starts one of two ways, and the choice decides what you type. Direct takes a list you already have. Give LoQuery a category instead and Discovery works the population out first, in waves, then hands the list back. You strike what does not belong, so you can settle who is in the set before anything is researched.

You give it a list

Direct

Vitol Inc, Glencore Ltd, Freepoint Commodities

It researches each item against your criteria and returns one row per item.

  1. You paste the list
  2. It researches each item
  3. Table with a source on every cell

Use it when You already know the population and need evidence about it.

You give it a category

Discovery

Aerial incidents over Canada or the Great Lakes that a government or military body formally investigated

It finds the population first, in waves, then hands the list back for you to approve before it spends anything researching it. Set a discovery budget and it can run straight through.

  1. You describe the category
  2. It collects candidates in waves
  3. It validates and removes duplicates
  4. You approve, strike or send it back
  5. Then it researches the approved list

Use it when You do not know who is in the set, and the answer depends on getting that right.

Approve a discovered list yourself →

Beyond the choices · 2 of 5

Two ways the configuration gets written, and neither is a template

Filling that gap is your job, and there are exactly two ways it gets filled. LoQuery writes the whole configuration from your description, which is what happens if you do nothing. Or you write it yourself. There is no menu of domains to pick from.

Writing it yourself is not all or nothing. Set the two fields you have an opinion about, so you can correct one run without teaching LoQuery a subject.

Path one · the default

Auto-detect. Describe the question in plain language and a model writes the whole configuration for this run. Most people never leave here.

Path two · yours

Build Your Own Type. You supply the configuration. Whatever you set is locked and never overwritten.

There is no third path, and specifically no library of prebuilt domain templates. Six of those existed once and were deleted, because a template list is where a general system quietly becomes a specialist in whatever is on the list.

Custom job type 4 of the 8 shown here · the rest fall through
  • entity_type What counts as one row clinical_trials
  • discovery_preferred_sources Specialist sources to reach for first clinicaltrials.gov, ema.europa.eu, who.int
  • tier_overrides Promote or demote specific domains for this run clinicaltrials.gov: 0.95
  • entity_confidence_threshold How sure it has to be before a candidate survives 0.70
  • additional_field_types Field names it would not recognise auto
  • dedup_key_pattern What makes two names the same entity auto
  • task_doer_context Extraction rules for this subject auto
  • field_guidance_overrides Where to look for a novel field type auto

Field names are the product's own. The values beside them are illustrative of a subject nobody configured, rather than a captured run.

Blanks are not defaults, they are deferrals. Anything you leave empty is generated for this run, and if that generation fails the static foundation underneath still answers, with a more generic result, not a broken one.

Beyond the choices · 3 of 5

Every cell carries a source or a reason

The source sits on the cell itself, next to a verdict on how well it supports that claim, so you can defend one number without re-reading the run. Any warning LoQuery raises against its own answer sits there too. When no record resolves, every cell reads Not Found and the row says why. Nothing is filled in with something plausible.

Every item gets up to three attempts, and each one searches wider while accepting weaker sources. The full ladder is in Making an open model return Not Found.

Research Workbench Table View
Item name Score Verdict Why Caveats DateOfficial explanationPrimary sourceStill unexplained
Kinross incident 2 attempts 0.71 confirmed with caveats Score 0.71; a USAF determination is on record and is contested, and no government or military file establishing it was located: the safety board's analysis, findings and recommendations were withheld under FOIA. source_contested uncalibrated_authority_domain 23 November 1953 aviation-safety.net/wikibase/161691 USAF: the F-89C was tracking an off-course RCAF C-47 and was lost over Lake Superior en.wikipedia.org/wiki/Felix_Moncla None located. The safety board's findings were withheld under FOIA; ASN WikiBase entry 161691 is user-submitted, not the USAF report aviation-safety.net/wikibase/161691 No. A determination was issued, and the pilot of the RCAF flight named in it, Gerald Fosberg, denied being intercepted en.wikipedia.org/wiki/Felix_Moncla
Lake Michigan 1994 incident 3 attempts 0.18 mismatch Low score (0.18); zero of 2 evidentiary field(s) filled. The 1994 reports are a cluster of sightings across several towns, so no single incident record resolves. Not Found Not Found Not Found Not Found
Two rows from the aviation run. The Primary source cell on Kinross links the ASN WikiBase entry as evidence about itself: a user-submitted record standing where the USAF report should be. The row beneath it scored 0.18 and filled nothing, because the 1994 reports are a cluster of sightings across several towns with no single incident record.

Sources here are real; scores and attempt counts are illustrative until a session capture lands.

Beyond the choices · 4 of 5

What a chatbot does, what deep research does, and what LoQuery refuses to be

A deep research mode is a chatbot given a longer leash. Rather than answer from one round of search, it plans, runs many searches, and hands back a written report with citations in it. Those citations sit inline, next to the claims they support, and none of them says how well the page supports the sentence. Whether a verdict travels with the source decides if you can defend one number later, and LoQuery puts one on every cell.

The differences fit in one table, and the last row is what a class of thirty pays. The pricing rule behind that row →

A free chatbotA deep research modeLoQuery
Where sources attach Nowhere, or the answer as a wholeInline, per claim, with no verdict on fitPer claim, with a verdict on fit
When the record is missing Answers anywayNearly always answers, citedThe cell says Not Found and why
Who judged, on the record NobodyNobodyEvery verdict names the judge: machine, human, or both
What a class of thirty costs Free, sold below costA subscription per seatThe runs those thirty students make, at cents a run

The same argument, stated in the negative.

  • Not a chatbot

    There is no conversation with the pipeline. You configure a run, watch it work, steer it while it goes, and get a table. Nothing here is trying to sound like a person.

  • Not a search engine

    It orchestrates search engines, today through your browser and your connection. We do not have an index and we are not trying to build one.

  • Not trained on a private library

    It reads the live open web, per run. Nothing is retrieved from a corpus we curated in advance, so nothing is as stale as the last time we updated it.

  • Not a cache

    Nothing carries over between sessions. Run the same question twice and it is researched twice, because a verdict reused from last month is a verdict nobody checked.

  • Not autonomous

    A run expects you in the loop, and today it needs your browser attached. The person in the loop is the design: a tool that finished the thinking while you were away would be the thing we built this to replace.

Beyond the choices · 5 of 5

Where we lose: reasoning, speed, testing

LoQuery loses on open-ended reasoning. A small open model under hard contracts beats a large one on verifiability, and that is the trade LoQuery is built on.

If your question needs a long chain of original thought, this is the wrong tool. We would rather say so here than have you find out on your third run.

The first answer takes longer than a chatbot's, because today searching happens in your browser at the pace a person would browse. We check quality by hand: LoQuery runs only with a browser attached, so no automated end-to-end evaluation exists yet for anyone to check.

Get early access