DocfuAI
Sign in
For lead analysts & chief decision-makers

The questions you walk in with —
answered with receipts.

Demand and supply shocks, regime shifts, supplier concentration, what-if scenarios. Docfu turns a curious question into a certified answer with the data lineage to back it.

Sign in
Provenance-tracked · fitness-certified
Reviewer gate before findings merge
Supplier risk · rare-earth concentration
CRO
Chief Risk Officer
at a regional bank
The question I couldn't answer
How concentrated is our SE Asia loan book in rare-earth processing — and if a single province tightens export permits for sixty days, which facilities lose input redundancy first?
Outcome
A facility-level concentration ranking with time-to-redundancy-loss per site — something defensible for the next stress-test review.
Inside Docfu · sample run
~ 14 min · 12 sources · 4 assets
›SE Asia rare-earth facilities · export permit shock · 60d redundancy
Query analysisAsset discoveryCurationCertificationHypothesis test
Purpose locked up front · reviewer gate before findings land
Asset
Fitness
Top dim.
Rare-earth processing facility registry
fully fit
completeness 0.96
Provincial export permit history
conditionally fit
temporal 0.77
Input-source concentration index
fully fit
accuracy 0.92
Substitute supplier lead times
marginally fit
single source
Hypothesis · M8 · supported
3 facilities lose redundancy within 60 days of permit tightening
fit-rate
81%
Regime · EU industrial electricity
LMA
Lead Macro Analyst
at a sovereign wealth fund
The question I couldn't answer
Has EU industrial electricity cost behaved differently from the US and China since 2022 — and does the gap explain any of the manufacturing output divergence we're seeing?
Outcome
A side-by-side price series with a tested co-movement hypothesis — enough to frame the allocation conversation with evidence.
Inside Docfu · sample run
~ 14 min · 12 sources · 4 assets
›EU vs US vs China industrial electricity · 2022–2025 · output divergence
Query analysisAsset discoverySynthesisCurationHypothesis test
Purpose locked up front · reviewer gate before findings land
Asset
Fitness
Top dim.
EU industrial electricity prices · monthly
fully fit
temporal 0.97
US industrial energy cost index
fully fit
completeness 0.93
China manufacturing power tariff panel
conditionally fit
geo 0.72
Industrial output index · 3-region
fully fit
accuracy 0.91
Hypothesis · M8 · supported
EU electricity premium explains 30–40% of output gap vs US post-2022
fit-rate
76%
Disruption pattern · chokepoint co-movement
HGR
Head of Geopolitical Research
at a reinsurer
The question I couldn't answer
Red Sea, Panama, Black Sea — are port disruptions across these three chokepoints becoming correlated? If two activate simultaneously, how does that change our cargo exposure model?
Outcome
A conditional co-occurrence estimate for dual-chokepoint activation — a number to put into the PML model alongside the marginal rates.
Inside Docfu · sample run
~ 14 min · 12 sources · 4 assets
›Red Sea · Panama · Black Sea · disruption co-occurrence · 2022–2026
Query analysisAsset discoveryEvidence triageCurationCertificationHypothesis test
Purpose locked up front · reviewer gate before findings land
Asset
Fitness
Top dim.
Port disruption event log · 3 routes
fully fit
completeness 0.94
Transit volume index · weekly
fully fit
completeness 0.89
Cargo rerouting cost panel
conditionally fit
completeness 0.68
Geopolitical incident register
fully fit
accuracy 0.92
Hypothesis · M8 · supported
Chokepoint disruptions are not independent — dual-activation P > 0.3
fit-rate
84%
Supply risk · API export controls
DoS
Director of Strategy
at a global pharma company
The question I couldn't answer
If India tightens API export controls for the molecules in our top-20 formulary, which products lose their primary source — and how long until a qualified alternate is ready?
Outcome
A product-level redundancy map with time-to-alternate per molecule — something to take into the next supply risk committee.
Inside Docfu · sample run
~ 14 min · 12 sources · 4 assets
›India API export controls · top-20 formulary · alternate source readiness
Query analysisAsset discoveryCurationFallback retrievalCertification
Purpose locked up front · reviewer gate before findings land
Asset
Fitness
Top dim.
India API export policy history
fully fit
completeness 0.98
Formulary → API source mapping
fully fit
accuracy 0.94
Alternate supplier qualification log
conditionally fit
temporal 0.73
Time-to-qualify estimates · DMF status
marginally fit
single source
Hypothesis · M8 · supported
6 of top-20 products have no qualified alternate within 90 days
fit-rate
88%
Who reaches for Docfu

The person who has to be right by Tuesday.

Heads of research, lead analysts, strategy directors, portfolio leads. The people whose hands stay on the data — because the next decision depends on the answer being defensible, not just plausible.

They don't need a chatbot. They need a way to validate a hypothesis against fast-changing macro, geo-political events, and a series of small signals — and walk into the room with a defensible asset list, not a paragraph.

Validate a what-if
Stress an input-cost or supply tightening across the segments in your book without rebuilding the model.
Pattern-detect
Test whether four disruptions are one story or four. Get a conditional probability, not a vibe.
Re-anchor a thesis
Re-test a half-life or a regime when the macro moves. Same query, fresh certificates.
Defend the answer
Every asset arrives with a fitness level, conditions, and a certificate ID an auditor can resolve.
What runs underneath

A pull request, for research.

Workspace = repo. Session = branch. Conversation = PR. Reviewer gates the curation and modelling stages. Findings merge, or they don't.

Query analysis
Purpose locked
Domain, geography, temporal scope, hypotheses. The fitness baseline against which every asset will be scored.
Asset discovery
Discovery → triage
Search expanded against the historical compliance pattern for this domain. Triaged for relevance and provenance.
Curation gate
Hypothesis parameters
Each asset receives a fitness level, per-dimension scores, and any conditions. Reviewer must approve before merge.
Certification
Certificate ID stamped
Durable, externally-resolvable. Your auditor can verify a finding without ever logging into the workspace.
Modelling gate
Hypothesis tested
Hypotheses backed primarily by marginally-fit assets are flagged. Reviewer signs off, or asks for changes.

The next what-if you turn over — answer it with a certificate.

Sign in