About Tector

Honest research, not just fast research.

Tector is a research platform for teams who have to stand behind their answers. It turns scattered raw material — documents, pasted text, web pages and captured AI conversations — into structured, evidence-backed findings your team can act on and defend.

Every finding is cited. Every disagreement between sources is shown, not hidden. And every gap is labelled as a gap, so nobody mistakes missing evidence for certainty.

Our mission

Help teams replace scattered notes and unverified AI summaries with organized, cited and honest research — where what's known, what's contested and what's still unclear are visible at a glance.

Our vision

To become the trusted system of record for team research: a place where evidence stays traceable to its source, disagreement is treated as signal rather than noise, and AI accelerates analysis without replacing judgment.

Most tools pick a winner. Tector shows you the disagreement.

Market entry, pricing, competitive and diligence teams drown in disconnected PDFs, articles, chat exports and notes. The usual fix makes it worse.

What most tools do
Summarize blindly, with prose that isn't tied to any source
Quietly pick a "winner" when sources disagree
Fill gaps with confident-sounding answers — false confidence
What Tector does instead
Organizes raw sources into a structured taxonomy of research questions
Extracts cited insights instead of ungrounded prose
Surfaces contradictions explicitly, rather than hiding or auto-resolving them
Flags what's still unknown, so you see gaps instead of false confidence

From a loose question to a defensible answer

Eight steps, one chain of evidence. Every finding can be traced back to the source it came from.

01
Scope the project
Describe the problem in plain language. Tector asks a few clarifying questions, then recommends a research type — market entry, pricing, competitive, diligence or GTM launch.
02
Generate the taxonomy
A structured tree of themes and sub-questions, tailored to the project type. Every source and answer is organized against it.
03
Add sources
Upload files, paste text, drop in a link, or capture an AI conversation with the browser extension. Each source stays linked to wherever it is used.
04
Extract insights
Sources are broken into discrete insights and auto-classified into the right sub-question — no manual tagging.
05
Run research
Each sub-question gets a grounded, cited answer. No citation, no answer. Disagreements become flagged conflicts; unknowns become explicit gaps.
06
Generate summaries
Reports at document, branch or project level, with question coverage tracked. When evidence changes, they flag as stale and refresh.
07
Read the synthesis
A decision-first narrative: situation, what we know, what is contested, what is thin, and the implications.
08
Compare & chat
See several models answer one question side by side, or keep asking in Research Chat — grounded in your project, with citations.
The synthesis view

Ordered by what matters for the decision — not by folder structure

Above the full report sits a narrative pane that gets you oriented in minutes.

1
Situation
The question and the context around it
2
What we know
Findings corroborated across sources
3
What's contested
Both sides of a disagreement, and the mechanism behind it
4
What's thin
Gaps, labelled as gaps
5
Implications
What it means for the call you're making

Everything inside Tector today

Thirteen capabilities, grouped by where they sit in your research workflow.

Bring it in
Scoping and sources
Guided project scoping
Turns a loose business problem into a defined project and research plan.
Multi-format ingestion
Files, pasted text, web pages and AI conversations, all in one pipeline.
Browser extension
Capture pages and chats straight into a project, no copy-paste.
Reusable templates
Save a taxonomy and skip scoping next time.
Make sense of it
Extraction and research
Insight extraction
Raw material becomes discrete, taxonomy-tagged insights.
Cited research runs
Mandatory citations and explicit “we don't know yet” flags.
Conflict detection
Both sides of a disagreement, with the reasoning behind it.
Ship it
Reports and answers
Multi-scope summaries
Document, branch or project reports that know when they are stale.
Synthesis view
Known, contested, thin and implications — ranked by relevance.
Multi-LLM Compare
One question, several models, side by side.
Grounded Research Chat
Ongoing Q&A tied to your project's insights and citations.
Run it as a team
Access and governance
Organizations & roles
Owner, admin, editor and viewer roles per organization.
Analytics & governance
Usage, AI spend by action type, user management and audit logs.
Provider-agnostic

Not tied to one model

Tector is a research layer that sits above the models. Compare runs one question across a fixed roster of cost-efficient models, side by side:

GPT-4o mini · OpenAI
Claude Haiku · Anthropic
Gemini Flash · Google
Llama 3 8B · Meta

Every generated report and answer records which model produced it — so the platform improves as models do, without locking you to a single vendor.

Scalability & reliability

Built to scale with your team

Model-agnostic — a dedicated AI gateway means new models slot in without touching the product.
Multi-tenant by design — organizations, projects and permissions are isolated at the data layer.
Cost-aware — heavy research and compare runs are triggered deliberately, with spend tracked per action type.
Staleness-aware — summaries notice when their evidence changes and refresh in the background.
Auditable — organization-level audit logs and super-admin visibility support governance.

AI should speed up the analysis. It shouldn't replace the judgment.

Launching September 2026 · Early access for the first 100 teams