Qualitative diligence is hard to compare.
Two analysts reading the same company arrive at different reads. Without structure, narrative becomes opinion.
// no shared schemaOntology-based market intelligence
See the story behind the company, structured, scored, and evidence-backed.
TOntoN analyzes public company evidence to evaluate vision, founder context, narrative clarity, strategic coherence, and comparable qualitative signals, producing transparent scores, explanations, reports, and benchmarks.
VISION ONTOLOGY
// vision_profile · acmerobotics.com
73/100
Vision score// evidence · (En)Vision · novelty
"We are building general-purpose dexterity, robots that learn the long-tail of human work, not a single task."acmerobotics.com/about · score driver · confidence: high
Report readyPDFVision report · 14 pp
The problem
Important company signals are visible before they are measurable.
They are scattered across websites, founder profiles, interviews, posts, and reports, and they resist the spreadsheets that diligence teams already rely on.
Founder and vision signals are unstructured.
Interviews, manifestos, and posts carry the strongest forward-looking signals, and the weakest data discipline.
// unstructured surfaceAI summaries aren't enough.
One-paragraph LLM summaries collapse nuance into prose. They tell you what was said. They don't tell you what it means.
// lossy abstractionTeams need evidence, not opinions.
Investment committees and clients don't accept unsupported assertions. Every claim has to be traceable to a primary source.
// source-or-it-didn't-happenThe solution
Three steps from public evidence to decision-ready intelligence.
TOntoN structures the unstructured. Each step is inspectable, and reviewable by a human before it informs a decision.
Collect and review sources
TOntoN crawls a company's public surface, site, blog, press, careers, founder profiles. An analyst confirms coverage and excludes the noise.
Analyze through the Vision ontology
Evidence is mapped to five primitives, Pre-Event Influences, Visioner, Visioning, (En)Vision, Statements, and scored against the (En)Vision dimensions.
Generate report, evidence, benchmark
Scores roll into a full report: radar, dimensions, evidence quotes, gap call-outs, peer benchmark, and a sourced PDF export.
Product workflow
From a company name to a sourced benchmark, in one linear flow.
Each stage is a distinct workspace screen, so analysts know exactly where they are, what was decided, and what the system did next.
- 01
Add Company
Drop a URL or company name. TOntoN discovers the public surface and prepares the run.
- 02
Source Review
AI-selected sources, ranked. Confirm, exclude, or flag, the analyst owns coverage.
- 03
Vision Analysis
Two-stage LLM pipeline extracts quotes and scores against the Vision ontology.
- 04
Result Page
Radar, dimensions, evidence drill-down, confidence bands, named gaps, all on one screen.
- 05
PDF Report
Sourced PDF export: shareable, citable, branded. Every score traces to a source.
- 06
Benchmark
Where this company sits in your workspace, percentile, distance from median, gap to leader.
What TOntoN measures
The Vision ontology: five primitives, scored independently.
Distilled from over a decade of vision-statement research. Every score in a TOntoN report ladders up to one of these primitives.
Pre-event
Influences
The context that produced the vision, heritage, founder history, prior art, and the conditions that made this company possible.
Visioner
Who holds the vision and how legibly they articulate it across surfaces, founder, leadership, and the named voices behind the company.
Visioning
The process of building, refining, and broadcasting the vision over time, how it evolves, where it sharpens, where it drifts.
(En)Vision
The vision content itself, scored across six dimensions: temporality, globality, novelty, specificity, complexity, popularity.
Statements
The artefacts, mission, manifesto, public commitments, that carry the vision externally and anchor it to a public record.
What you get
Every run produces six artefacts, all traceable to source.
Built for an analyst making a defensible call, not a stakeholder skimming a slide. Pick the artefact you need; the rest stays one click away.
Overall Vision score
A composite 0–100 score for the company, rolled up from the five primitives. The headline number behind every report.
73/100
High confidenceDimension scores
Six (En)Vision dimensions scored independently. Where the company over- and under-indexes is visible at a glance.
Temporality82
Novelty88
Specificity54
Evidence-backed explanations
Every score names its drivers and shows the quotes that moved it. Click any node in the report to reach the source page.
"Robots that learn the long-tail of human work…"acmerobotics.com/about
Source quality review
A reviewable list of every source used, with quality flags. Low-confidence material is marked, not silently scored.
acmerobotics.com/abouthigh
linkedin.com/in/founderhigh
techcrunch.com/2024/...low
Benchmark rank & percentile
Where the company sits inside your workspace, percentile, distance from median, gap to the leader. Updated as new runs complete.
P82 · workspace
Sourced PDF export
A branded, citable export: the full report compressed into a shareable artefact. Used for IC memos, client decks, and audit trails.
PDF
Built for
For teams who need qualitative signals to be comparable.
TOntoN sits next to the spreadsheet, not inside it. It hands an analyst the structure that quantitative tools never gave them.
Investors
Screen and compare qualitative signals during early diligence. Get a defensible read on vision and founder narrative before the first call.
// pre-IC screeningConsultants
Turn strategy, vision, and narrative into a diagnostic. Score the client's public posture against peers and surface the gaps to brief on.
// strategy diagnosticPlatforms
Enrich market intelligence products with a qualitative signal layer. Embed Vision scores alongside financial and traction data.
// data partnershipFounders
Understand how clearly your company vision is expressed. See what an analyst would conclude, and what they'd miss, from your public surface.
// public-surface auditWhy TOntoN is different
An ontology-driven intelligence platform, not another AI summary.
Generic AI tools summarize. TOntoN structures, scores, and benchmarks. The output is built to be reviewed, defended, and compared.
Criterion
Generic AI summary
TOntoN
Structural modelHow the analysis is organized.
·Free-form prose; no shared schema across companies.
✓Ontology-based: five primitives, six (En)Vision dimensions.
Evidence handlingHow claims tie to source material.
·Quotes blended into summary; sources approximate or absent.
✓Every score names drivers and links the exact source quote.
ComparabilityWhether you can put two companies side-by-side.
·Each summary is bespoke; comparison is qualitative at best.
✓Calibrated 0–100 scores per dimension, directly comparable.
ExplainabilityWhether a reviewer can audit the result.
·Black-box reasoning; hard to defend in front of a client or IC.
✓Named score drivers, confidence bands, gap call-outs.
BenchmarkingWhere this company sits relative to peers.
·No native peer set; benchmarks have to be rebuilt by hand.
✓Workspace percentile, distance from median, gap to leader.
Source curationWho decides what gets analyzed.
·Whatever the model retrieves, often opaque, often outdated.
✓Human review of every source before analysis runs.
Methodology
Transparent by design, a reading lamp, not a spotlight.
TOntoN tells you what the public record says, how confidently, and what it doesn't. It supports expert judgement, it does not replace it.
Public-source evidence only.
Analysis is built from public material, website, blog, press, careers, profiles. No private data, no leaked documents, no scraped paywalls.
Human source review before analysis.
An analyst confirms, excludes, or flags every source. The system runs against what the human chose to include.
Scoring tied to defined ontology.
Every score maps to a named ontology node. The model never produces a number without a defined dimension to put it against.
Supports decisions, doesn't replace judgement.
Reports surface structure, evidence, and gaps. The decision, to invest, to advise, to publish, remains with the human reviewer.
Where the crawler couldn't reach, where extraction returned nothing, where source confidence is low, the report says so. Insufficient evidence is a finding, not a failure.
Get started
Start seeing the qualitative signals others miss.
Drop a URL, confirm the sources, watch the pipeline run. Access is currently invite-only while we calibrate the scoring model with early partners.