Chinese start -up challenges the facts challenging Chat GPT in data ventilation: which businesses should we use?

This promise to be true Almost almost Almost looks great: Leave the file separated from a dirt comma (CSV) file in the AI agent, wait two minutes, and return polished, interactive charts ready to offer your next board.

But the Chinese startup is exactly Manus.im This month is supplying its latest data ventilation feature.

Unfortunately, my early hand -on testing with bad datases shows a basic enterprise problem: impressive capabilities that combine inadequate transparency about data changes. Although Menus Chat handles better dirt data than GPT, no tool is yet ready for the board room ready.

Spreadsheet problem affecting enterprise analytics

Rasms’ 58 % of the 470 finance leaders survey still rely mainly mainly on Excel for monthly PI, yet they own a BI license. One more Ticker The study estimates that the overall spreadsheet affects about 90 90 % of the organizations-which creates a “last mail data problem” between government warehouses and hasty exports, which occurs several hours before important meetings in analysts’ inboxes.

Menus targets this same space. Upload your CSV, explain what you want in a natural language, and the agent automatically clears data, chooses the appropriate Vega light grammar and returns the PNG chart ready for export-no axis tables are required.

Where Monos beat Chat PPT: 4x with dirt data slowly but more accurate

I tested both Datases (113k-Row (113k-Row) modern data analysis of both Menos and Chat GPT E -commerce orders200 k-r Marketing chimney 10K-Row Sauce mr r), First clean, then damaged with 5 % error injection, including NULLs, mixed format dates and duplicates.

For example, testing the same prompt — "Show me a month-by-month revenue trend for the past year and highlight any unusual spikes or dips" — across clean and corrupted 113k-row e-commerce data revealed some stark differences.

For context: Dupic can handle only 1 % of the file size, while Claude and Grook took more than 5 minutes each but developed interactive charts without PNG Export Options.

The results:

Chatra 1-2: Chart-outpts with the same tax trend indicator on dirt e-commerce data. Menus (below) produces an integrated trend despite the misconduct of data, while the chat shows the GPT (above) the unclean history in the form of unclean history.

Menus behaved like a cautious junior analyst – Automatically clear the data before charting, successfully analyzing contradictions in history and handling NULL without clear guidelines. When I requested an analysis of the same revenue trends on bad data, Manas took about 4 minutes but created an integrated concept despite data quality issues.

Chat GPT acts like a speed coder – Priority to faster output over data hygiene. The same request only took 59 seconds but misleading concepts were created because it did not automatically clear the contradictions of formatting.

However, in terms of “executive preparation”, both tolls failed. Nor prepared a label for the board ready for the board without a signal. Data labels were often over -liping or very small, the bar chart lacked proper grid lines and the number formatting was contradictory.

Transparency crisis businesses cannot overlook

This is the place where the Menus causes anxiety to adopt an enterprise: The agent never takes cleaning measures that applies to. An auditor of the final chart review has no way to confirm whether out -ry -out people have been dropped, given or changed.

When someone offers a quarterly results based on a CFO Menus chart, what happens when someone asks, “How did you handle duplicate transaction with Q2 system integration?” The answer is silence.

Chat GPT, Claude and Grook all show their codes, though the lack of transparency programming experience through the Code Review is not extended to business users. What the enterprises need is an easy audit trailer, which promotes confidence.

AI racing from the warehouse

Although the Menus is focused on CSV uploads, major platforms are directly building chart generation in enterprise data infrastructure:

In Google’s Gemini Big Curi Usually became available in August 2024, which enabled the generation of SQL questions and inline concepts on direct tables, respecting the protection of the row level.

Microsoft’s fabric copyl In May 2024, Power BI reached the GA in the experience, working directly with leakhouse datases, creating visuals inside the fabric notebooks.

AI Assistant of Good DataLaunched in June 2025, works in consumer environment and respects existing cement models, which allow consumers to ask questions in simple language, while receiving answers that agree with default matrix and business terms.

These warehouses-local solutions completely eliminate CSV exports, preserve full data lineage and take advantage of existing security models-the benefits of struggling for the struggle for the struggle of the menus such as file upload tools.

Critical gaps to adopt an enterprise

My test revealed several blockers:

Direct data connectivity Absent – Menus only supports the file uploads, which do not have the Asnaphilic, Big Cory or S3 connectors. Manus.im says the connectors are “on the road map” but do not offer any timelines.

Audit Trail Transparency Completely missing. Enterprise data teams need change logs, showing how AI has cleared their data and whether its fields are correct.

Export flexibility PNG is limited to results. Although the Adequate of the Quick Slide Deck is suitable, businesses need customized, interactive export options.

Decision: Infectious Tech, Premature for Enterprise Use Matters

Ad hoc CSV analysis of SMB executives, MANS ‘Drag and Drop Visionalization seems to be working.

Cleaning autonomous data handles real -world dirt, which will otherwise require pre -processing, when you have the appropriate data, it is cut from hours to minutes.

In addition, it offers an important advantage of the run -time compared to Excel or Google Sheets, which requires manual axis and has enough time to load due to local computing power limits.

But with the ruled data leaks, regulated businesses should wait for warehouse -related agents such as Gemini or Fabric Coal, who keep the data in the security sector and maintain full lineage.

Down line: Manas proved a prompt charting work and handled the dirt data impressively. But the business of businesses, the question is not whether the charts look good – that is, can you put your career at stake with data changes that you cannot audit or confirm. Unless the AI agent can plug directly into the Government Tables with strict audit tracts, Excel will continue to play its role in quarterly quarterly presentations.

Daily Insights on Business Use Matters with Daily VB

If you want to impress your boss, the VB Daily covers you. We give you internal scope what companies are doing with Generative AI, from regulatory shifts to practical deployments, so that you can share insights for more and more ROIs.

Read our privacy policy

Thanks for subscribing. Check more VB Newsletter here.

There was a mistake.