Enterprise AI

Analyze Thailand Election 66 Data Using Aerogram

Analyzing Election Data with Aerogram AI - 68 constituencies where the candidate won, but the party lost the party-list vote - 20 seats decided by a margin of just 500 votes - Which party came in second among Bangkok voters?

Cleverse Team · Feb 7, 2026 · 12 min read

**Disclaimer:**This is an experiment. We used AI tools to dig into Thailand's 2566 election data, in preparation for the coming 2569 election. This isn't intended to predict or influence any election outcome—it's about what the data is actually telling us.

46.5% of Thailand Voted Against Themselves

In approximately 70% of all constituencies in the South, the first-place party in the constituency MP vote and the first-place party in the party-list vote were different parties, whereas voters in Bangkok placed the same party first in both the constituency and party-list ballots.

Infographic: https://app.aerogram.ai/p/084148c3-ec44-4fea-b4ba-35e1d12d2d31


Who Won Constituency Seats but Lost Party-List

Bhumjaithai Party won 68 constituencies where they lost the party-list vote. Sixty-eight. That's individual candidate power winning despite the brand. Palang Pracharath Party had 39 such districts. Pheu Thai Party had 23.

Infographic : https://app.aerogram.ai/p/d80f881c-32e6-45bc-9ed5-0df650b57bc3

The United Thai Nation Party received more votes from Bangkok and 1st Constituency in other provinces than from other provincial constituencies.

More than half of the United Thai Nation Party's party-list votes came from voters in Bangkok and Constituency 1 in other provinces.Interestingly, while most people might think the Move Forward Party received the majority of its votes from urban areas, in reality, approximately 58% of the party-list votes it received came from non-urban constituencies (provincial constituencies other than Constituency 1).

Infographics: https://app.aerogram.ai/p/efed24af-4474-4141-8938-1dfbb8df0a11

How We Use AI and Aerogram to Create Infographic Generator Workflow

AI redesign this traditional process into an AI workflow. And the tool that makes this possible in practice is Aerogram.

Analyzing election data requires a political reporter who understands the narrative, a programmer who can clean the mess, and an data analyst who can find the patterns — all working together as one team with data analysis process.

AI redesign this traditional process into an AI workflow. And the tool that makes this possible in practice is Aerogram.

The Traditional Data Analysis Process

Step 1: Understand Context → Role: Reporter. Identifies the story angle, the key trending topic.

Step 2: Define Problems → Role: Data Analyst. Translates the reporter's needs into analytical questions.

Step 3: Clean Data → Role: Programmer. Spends hours wrangling messy raw data into usable format.

Step 4: Exploratory Data Analysis → Role: Data Analyst. Looks for patterns, generates charts, tests hypotheses.

Step 5: Interpret → Role: Political Expert. Determines what the insight.

Step 6: Present → Role: Reporter. Assembles everything into charts and dashboards for publication.

Six steps. Four roles. Too many handoffs. Too slow for election cycles. And the tools are completely disconnected — data in one place, analysis in another, AI chat conversations disappearing into browser tabs no one can reference later.

Redesign Process with AI Workflow

Step 1: Understand Context → Role: Reporter. This stays the same. Human judgment about what's newsworthy is irreplaceable.

Step 2: Define + Clean + Explore + Interpret → Role: Political Expert. Instead of four handoffs across four roles, the reporter works directly with AI that defines questions, cleans data, runs analysis, and interprets results in a single loop. Reporter reviews. AI iterates. Minutes, not weeks.

Step 3: Present → Role: Reporter. An interactive, shareable dashboard generated directly from the analysis.

Three steps instead of six. One loop instead of a relay.

How This Works in Aerogram

Step 1: Understand Context

Your data lives on persistent cards on the board. No re-uploading. No version confusion. When you need it, drag a connection to whatever analysis you're building. AI immediately runs code to describe the dataset

Step 2: Do Four Things in One Step

Create a card to generate analysis ideas. AI proposes analytical angles — turnout patterns, demographic shifts, anomaly detection — defining problems, exploring data, and interpreting findings in a single pass.

Create a second card to critique those ideas. Challenge assumptions. Check for confounding variables.

Merge both cards. The generative thinking and the critical review combine into a refined analysis. The reporter iterates this loop as many times as needed. Each cycle takes minutes.

Step 3: Present

The interactive dashboard is generated directly from the workflow on the same board. Shareable via link.

AI Workflow

Why AI Workflow

Redesign the whole process, don't layer AI onto existing process. The old way used AI as an assistant in each roles. Same process, slightly faster. The new way redesigns the process entirely. Four roles compact into one loop. The reporter drives the whole thing. AI handles the technical execution.

The reporter (non-coder) can perform a data analyst task. Once the workflow is built, the reporter operates it directly. No advanced prompting skills. No Python. The AI workflow encodes the methodology — the cleaning data, the analysis, the interpretation.

From individual AI work to a collaborative AI workspace. No more asking ChatGPT, copying the response, pasting into a doc, sharing with a colleague. That's individual AI work — everyone in their own chat window, stitching results together manually. In Aerogram, the data, analysis, critique, and output all live on one board. Everyone works on the same workspace. No lost context. No version control nightmares.

The real transform isn't "AI analyzes data faster." It's that the entire workflow lives in one place, and the people who need the answers can operate it directly.

Aerogram : https://www.aerogram.ai/

Cleverse Team

What would you like to achieve with AI?

Bring a business goal. We’ll work out where to begin.