BellData Intelligence
Data · Live

The country,
alive.

Static directories rot the moment they ship. Bell's data doesn't. Every record on the graph is polled continuously, every change is detected and timestamped, every fact carries its own freshness — down to the 60-second air-traffic ping, up to the weekly sector report.

Twenty-two record types. Five refresh tiers. Continuous, every day, on Bell-owned infrastructure.

Live data plane · pulse of the country
1.2 B
Datapoints kept current
every day, every record
The refresh-cadence atlas

Every record type, on its own clock.

The pulse of the country isn't uniform. Air traffic moves by the second; macro indicators move by the week. Bell respects that. Below: every record type and the exact cadence Bell polls, change-detects, and updates it.

T1
Real-time
60-90 secondsThe pulse layer.
Live6 types
60s
Air traffic
every commercial movement
Refreshing
60s
Road traffic
every major Doha segment
Refreshing
60s
Weather
all Qatar stations
Refreshing
60s
Signals feed
~3 / minute, country-wide
Refreshing
90s
News stream
~8 / hour, Qatar coverage
Refreshing
90s
People-density heat
anonymized aggregates, by district
Refreshing
T2
Live
5-15 minutesThe market-rhythm layer.
Live4 types
5 min
Tenders & RFPs
new posts as they land
Refreshing
10 min
Job postings
~14,800 / month, country
Refreshing
10 min
Regulator bulletins
every QFC / QCB / QFMA notice
Refreshing
15 min
Public conversation
LinkedIn, press, podcasts
Refreshing
T3
Frequent
1-3 hoursThe leadership & licence layer.
4 types
1 h
Leadership changes
CEO / CFO / CTO / board moves
1 h
Licence updates
issuances, renewals, revocations
2 h
Filings & registrations
MoCI registry & updates
3 h
Ownership changes
cap-table edges, UBO shifts
T4
Hourly–Daily
6-24 hoursThe structural layer.
5 types
6 h
Company core fields
employees, sector, revenue band
12 h
Regulations
new laws, circulars, decrees
12 h
Government datasets
ministry publication cadence
24 h
Court & tribunal records
commercial dispute filings
24 h
Sector aggregates
volume / heat / momentum
T5
Weekly+
weekly cadenceThe slow-moving layer.
3 types
weekly
Industry reports
sector trackers, multi-source
weekly
Academic & policy
Education City, think-tanks
weekly
Macro indicators
capital flow, trade, employment
The live change stream

Field-level diffs, as they land.

Not events — field-level changes. This is what ‘alive’ actually looks like, inside the graph: every datapoint compared against its previous value, every diff captured with its source attached. Watch a new change land at the top every few seconds.

Live change stream · whole graph·field-level diffs · showing the last 8
every diff cited at source
  • Tayyar Fintech
    Employee count
    3847
    LinkedInjust now
  • Doha Health Network
    CFO
    vacantYousef Al-Mannai
    MoCI filing8s ago
  • QTerminals
    Open tenders
    23
    QFC bulletin22s ago
  • Industries Qatar
    Family-office LP %
    22%24%
    MoCI registry41s ago
Why static directories fail

Same record. Six months later.

The same record on Doha Health Network, seen through a static directory frozen six months ago versus seen through Bell's live graph today. Where they diverge is what you missed.

Field
Static directory
Bell.qa live record
Employees
350
380
CFO
vacant
Yousef Al-Mannai
Ownership clarity
not tracked
Founder 62%, LP 22%, ESOP 11%, Strategic 5%
Live signals
0
4 (licence, CFO, expansion, LP liquidity)
Board members
3 known
4 known
Sector tag
Healthcare
Healthcare · private clinic operator
HQ address
West Bay
West Bay
Founded
2014
2014
6 of 8 fields diverged in 6 months · static loses 75% of its accuracyMultiply by 500+ datapoints per record. Every record.
The change-detection engine

Catching change without lying to chase it.

Detecting that something changed is the easy part. The hard part is knowing it actually changed — not just that one source said so. Bell's change-detection engine corroborates, weights, cites, and resolves every change before it promotes to the record.

Field-level diffs

Bell diffs every record at the field level, not the record level. Knowing what changed is more useful than knowing that something changed.

Granularity matters: a CFO change is not a registration change.

Multi-source corroboration

A change only lands when corroborating sources agree. Single-source disagreements are flagged for review, not promoted to the record.

A rumour from one feed is a rumour. Three sources agreeing is a fact.

Confidence-weighted promotion

Each change carries a confidence score derived from source strength, agreement count, and historical reliability. High-confidence changes promote instantly.

Low-confidence changes wait. The record never lies for speed.

Source-level provenance

Every change in the record carries the source that caused it. Click any datapoint — see exactly what fired, exactly when, exactly where.

No black box. Every diff has a citation.

Full change history

Every change a record has ever had is retained. Replay the record at any past date and see what it looked like, what it knew, what sources it cited.

The graph remembers everything.

Bella-augmented resolution

When sources disagree or a change is ambiguous, Bella reviews the candidates, evaluates the evidence, and proposes the resolution — with reasoning attached.

Edge cases get a second pass. Records get human-grade judgment.
You've seen what stays alive
Now query records that never went stale.
Activation runs from 1 to 24 hours from your access request. The first record you open is already as fresh as the country.
Three lenses on the same liveness

What Bell.qa changes when data stays alive.

For the data engineer

Cadence is observable per record type. Every diff is timestamped, sourced, and replayable. You can audit the freshness of any field at any point in history — the graph remembers.

For the data buyer

The records you procure aren't snapshots. Every field carries its last-verified timestamp; every change carries its corroboration; every datapoint stays as fresh as the country. No quarterly stale-data write-off.

For the partner / investor

Live is the defensibility moat. Anyone can license a dataset; nobody else has built the change-detection engine that keeps a Qatari country graph current, on Bell-owned servers, with full provenance.

Records that don't age.

Twenty-two record types. Five refresh tiers. 1.2 billion datapoints kept current every single day. Every field cited, every change replayable, every record as fresh as the country it represents.