What the analysis produces
Political intelligence / For the whole party
AZIMUTH
Political intelligence
before the next move.
Explore what each side could do.
Find the questions that need answers while there is still time.
For the party’s field, research, strategy, and media teams.
Proposed system / Existing foundationsFrom scattered signals
Keep scrolling ↓To a clearer picture.
01 / DSC through the four rooms
One issue.
Four rooms.
What is the strongest move each side could make? How could the other side answer? Azimuth explores all four views, including the ones that challenge your own position.
In this exercise, A is the government and B is the opposition. “Worst” means the strongest plausible move the other side may face. These are possible arguments, not findings about current recruitment. The actors can change with the issue.
Each possibility keeps its assumptions, warning signs, and reasons to reject it. People decide which action to take.
02 / The question changes with the moment
What needs
an answer now?
Prepare the questions before an event. When new information arrives, check which question matters and what its answer would change.
Ask at this moment
- What to check
- Who follows up
- Why the answer matters
03 / Belief, trust, and possible damage
Which stories
could change trust?
An issue can get attention without changing anyone’s mind. Azimuth asks what people might come to believe, why it matters to them, and what could show a real change.
The damage question
“Which belief could this issue change?”
A quality concern, a claim of unfairness, a question about competence, and a threat to identity may affect trust differently. The effect depends on the people and the situation.
For a case such as Tirupati laddu, a quality question and an allegation concerning a sacred practice raise different concerns. Their effect on public belief remains a hypothesis to test. This is not a finding about the case, intent, or voting behaviour.
The event
What happened, according to the available records?
Claimed intent
Why do people say it happened? What supports that explanation?
Claimed scale
How widespread or serious is it claimed to be?
Proposed remedy
What do people say should happen next?
Existing narrative
Which earlier story shapes how this event is understood?
A document may answer one question while leaving the others open. Keep those questions separate.
Comments show the discussion we captured. Suitable surveys or field research are needed to establish whether beliefs actually changed. Attention, loss of trust, and changed behaviour are different outcomes.
04 / Intelligence across the party
Your teams bring
what the feed cannot.
Field experience, research, and political judgment enter the same case. Azimuth connects them to the source material and keeps the questions, options, and decisions together.
Field teams
Record what they observed, heard, or inferred. Each keeps its own source and level of certainty.
Researchers
Check the records and compare explanations, including ones that challenge the preferred account.
Strategy & media
Examine possible developments and review what the evidence supports saying.
Organizers
Give each follow-up an owner and check what remains unanswered.
Decision-makers
Choose the action and keep a record of the reasons.
05 / Sources and field knowledge
One case.
The sources behind it.
Connect public posts, official records, and field reports. See where they agree, where they conflict, and what is still missing.
An operator’s account can open an inquiry before documents are available. Keep direct observation, secondhand information, and inference distinct.
01 Collect02 Connect03 Review
What did the operator observe?
Source basis, date, and confidentiality
Reported observationWhat does the document say?
Order, clause, and effective date
Source attachedWhat is being said?
Post, exact claim, and capture date
Awaiting reviewONE CASE Sources, claims, and open questions
06 / Timing and priority
A useful question
has a deadline.
Ask which belief or decision is at stake, why the issue matters now, and whether the team can act on the answer. The loudest issue does not automatically come first.
What changed?
Find a new document, a conflicting account, or a change in how people describe the issue.
What could follow?
Explore possible developments and warning signs. Record any forecast separately, with its assumptions, deadline, and a rule for checking the result.
What could answer it?
Test the strongest challenge and possible responses. Keep what supports and weakens each one.
When does it matter?
Link the question to an event, an owner, and the decision deadline. Check old material before reuse.
Before the announcement, hearing, or review
“If the answer changes,
what should we do differently?”
Scenarios explore possibilities. A forecast states an expectation before the outcome, with a deadline and a clear rule for judging the result.
07 / Decisions and memory
Every decision
keeps its reasons.
Brief
Explain what changed, what the evidence supports, and what remains uncertain.
Assign
Give each check an owner and a due date.
Review
Compare new evidence. Revisit the decision.
Remember
Keep the original expectation, actual outcome, corrections, and lessons for the next case.
What did we expect?
What actually happened?
People approve decisions and communications. The system does not publish autonomously.
08 / What already exists
What exists today.
What comes next.
4,174Videos in the recorded corpus
182,556Comments in the recorded corpus
336Channels in the recorded corpus
Research sample—not a representative measure of voters or public opinion.
Collection & research
Video and comment datasets, issue briefs, and a bounded Instagram collection pilot. Collection coverage varies by source.
CortexSignal / Meta collection
Scheduling and worker controls, access roles, recovery tools, and dashboard foundations. Sustained multi-account operation and Facebook coverage still need proof.
EvidenceDesk
Case, source, claim span, review, and verdict records demonstrated with synthetic examples. Collector integration remains partial.
The connected Azimuth workflow
Manual examples and the four-room method exist. Connecting belief analysis, field reports, timely questions, and team decisions into one working system remains planned.
The next proof
Run one issue through collection, analysis, a team decision, and a checked outcome. Compare it with the team’s existing process.
09 / The complete working scope
What Azimuth
is being built to do.
01Collect and organize
Connect permitted public-source collection, official documents, consented field reports, transcription, OCR, and multilingual research. Preserve links and dates. Deduplicate repeats. Flag coverage gaps.
02Understand the issue
Build timelines and compare claims. Separate the event from claims about intent, scale, and remedy. Examine possible effects on trust and who could benefit. Keep facts, reported accounts, and assumptions distinct.
03Challenge the scenario
Use Worst of A, Worst of B, Counter A, and Counter B to examine each side’s strongest plausible move and the other side’s response. Keep the assumptions, warning signs, and evidence that would reject each possibility.
04Ask and act
Prepare questions and source packets before expected events. When the event arrives, check which branch applies. Name the evidence needed, the owner, the deadline, and the decision that depends on the answer.
05Work across the party
Prepare role-appropriate briefs, field checklists, review queues, and follow-up records. Keep permissions scoped. Retain source context as information moves between teams.
06Learn and verify
Record expectations before outcomes. Keep results and corrections without rewriting the original. Historical records contain five forecasts and zero resolved outcomes, so prediction accuracy remains unproved. Evaluate language quality and collection gaps too.