Chinese startup Manus challenges chatgpt in data visualization: what enterprise should they operate?

The promise sounds almost too good to be true: drop the file separated with the comma values (CSV) to the AI agent, wait two minutes and recover the polished, interactive chart ready for the next presentation of the album.

But this is the Chinese startup Manus.im Provides the latest data visualization function, launched this month.

Unfortunately, my initial practical testing with damaged data sets reveals the company’s basic problem: impressive possibilities combined with insufficient transparency about data transformation. While Manus supports messy data better than chatgpt, no tool is ready for slides ready for the conference room.

Problem with spreadsheets harassing enterprise analyst

Rossums’ A study of 470 finance leaders showed that 58% still consisted primarily on the Excel program for monthly KPIs, despite having a BI license. Other Techradar The study estimates that the overall dependence of the spreadsheet affects about 90% of the organization-creating a “problem with the data from the last mile” between the regulated warehouses and the hasty export of CSV, which land in the inboxes of the receiving analysts of analysts a few hours before critical meetings.

Manus is focused on this exact gap. Send your CSV, describe what you want in natural language, and the agent automatically cleans data, chooses the appropriate Vega-Lite grammar and returns the PNG table ready for export-does not require pivot tables.

Where Manus overcomes chatgpt: 4x slower, but more true with messy data

I tested advanced manus and chatgpt data analysis using three data sets (113K-ROW Orders E -commerce200 thousand Trench marketing funnel 10k in order Saas MRR), first neat and then damaged with 5% errors injection, including zero, 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 neat and corrupted 113k-row e-commerce data revealed some stark differences.

In the case of the context: Deepseek could only support 1% of the file size, while Claude and Grok took over 5 minutes, but produced interactive charts without PNG export options.

Outputs:

Figure 1-2: Chart results from the same monitor of the revenue trend for messy e-commerce data. Manus (Dolny) produces a coherent trend despite data damage, while ChatgPT (TOP) shows distorted patterns from unclean dates.

Manus behaves like a careful younger analyst – automatically slowing the data before the chart, successfully analyzing the inconsistency of the date and service of zero without clear instructions. When I asked for the same analysis of income trends on corrupt data, Manus took almost 4 minutes, but he created a coherent visualization despite data quality problems.

CHATGPT works like a speed coder – prioritizing a quick output before data hygiene. The same request took only 59 seconds, but it caused misleading visualizations, because it did not automatically neat the formatting of inconsistency.

However, both tools failed in terms of “executive readiness”. Neither has developed a label of ready -axis scaling or legible without control prompts. Data labels often overlapped or too petite, the bar charts lacked the appropriate mesh lines, and the formatting of the numbers was inconsistent.

Crisis enterprises cannot ignore

Here Manus becomes problematic for the adoption of the company: The agent never revolves the surface that it uses steps. The auditor browsing the final chart is not able to confirm whether the protruding values have been dropped, assigned or transformed.

When the financial director presents quarterly results based on the graph generated by Manus, what will happen when someone asks: “How did you deal with the duplicate transaction from the integration of the Q2 system?” The answer is silence.

Chatgpt, Claude and Grok show their Python code, although transparency through the code is not scalable for business users who do not have experience in programming. What the enterprise needs is a simpler audit trail that builds trust.

Warehouse-Native Ai races before

While Manus focuses on sending CSV, the main platforms build charts directly in the infrastructure of the company’s data:

Google’s Gemini in Bigquery It became generally available in August 2024, enabling the generation of SQL queries and visualizations built into live tables, while respecting security at the level of the poem.

Microsoft’s Copilot in fabric He reached GA in Power BI in May 2024, creating visualizations of internal fabric notebooks while working directly with LakeHouse data sets.

Gooddata, you have an assistantIt was launched in June 2025, operates in customer environments and respects existing semantic models, enabling users to ask questions in a regular language, while receiving answers that are in line with predefined indicators and business terms.

These solutions in the native magazine completely eliminate CSV exports, retain the full data line and operate existing security models-free-paying files tools such as Manus Wassing to match.

Critical gaps in the adoption of the company

My tests have revealed several blockers:

Live data connectivity It remains absent – Manus only supports files, without snowflakes, Bigquery or S3. Manus.im claims that the connectors are “on the road map”, but they do not offer a schedule.

Transparency of the audit trail It is completely missing. Company data teams need transformation dailies, showing exactly how AI cleaned their data and whether its interpretation of the fields is correct.

Export flexibility is circumscribed to PNG outputs. Although suitable for brisk slide decks, enterprises need adapted, interactive export options.

Verdict: Impressive technology, premature for cases of company operate

In the case of SMB staff in the AD-HOC CSV analysis, the visualization of Manus Drag-i dropping seems to perform the task.

Autonomous data cleaning is supported by a real mess, which would otherwise require manual pre -processing, reducing time for minutes when you have reasonably complete data.

In addition, it offers a significant advantage of the executive environment compared to Excel or Google sheets, which require manual torque and significant load times due to local computing energy restrictions.

But regulated enterprises with adjustable data lakes should wait for native warehouse agents such as Gemini or Fabric Copilot, which store data in security circuits and maintain full tracking of the line.

Conclusion: Manus proves that one convoluted charts and impressively supports messy data. But in the case of enterprises, the question is not whether the charts look good – whether you can put your career in data transformations that you cannot audit or verify. Until AI agents are able to connect directly to adjustable tables with strict audit routes, Excel will continue to play its role in quarterly presentations.

Daily observations in matters of business operate with VB daily

If you want to impress your boss, VB Daily is covered by you. We give you an internal measure about what companies do with generative artificial intelligence, from regulatory changes to practical implementation, so you can share insights for the maximum roi.

Read our Privacy Policy

Thanks for the subscription. Check out more VB newsletter here.

There was a mistake.