Zedly AI

Launched our setup & security product for OpenClaw: zedly.ai/zedly-setup

Just launched our open-source plugin for OpenClaw for PII redaction, shell command blocking, prompt injection detection, and a tamper-evident audit log in agent workflows. One install, zero code changes. zedly.ai/zedly-shield?v…

We replaced fixed threshold retrieval with an LLM-driven planner. For each query it chooses the strategy: section-first, cross-ref expansion, iterative evidence. Retrieval that adapts to the question instead of always doing the same thing.

New: Zedly now auto-classifies documents on upload and runs type-specific extraction (starting with lease abstraction). Results are cached so repeat queries are instant, no re-processing.

Added an overdraft risk analyzer to the bank statement pipeline. Flags balance volatility, sustained low balances, overdraft history, automatically. What used to take 2 hours of manual review is now a 30-second report.

@HeckerImel Most real world data is messy. The interesting part is how you make it usable without oversimplifying it.

Models suggest I'm most myself when my data is messy

@UdemeOkono This is so real. VLOOKUP rarely fails because of logic. It usually breaks because of hidden spaces, mixed data types, or slightly different column structures. Cleaning the schema first changes everything.

A thorough data cleaning saves a lot of time with the rest of the work. VLOOKUP & his family members stressed my life with a few messy columns today 🤦🏾 This data analyst sontin, is this how you people did it to travel the world? 🙆🏾‍♀️🙆🏾‍♀️🙆🏾‍♀️

@jerryjliu0 This is such a real problem. Most Excel files in the wild are full of merged cells, inconsistent headers, and multi level structures. It is not the data that is hard, it is the layout.

LLMs/general agents still struggle to make sense of messy and complex Excel data. You can't easily dump all cells into the context window, and using the code interpreter is inefficient. LlamaSheets is one of my favorite releases from last year. We've embarked on an effort to build state-of-the-art algorithms and models to segment and parse complex Excel tables - including merged cells, hierarchical rows/columns. This includes both sheet-level and table-level understanding. We think there's a ton of use cases that this can help solve (simplest example: structuring your income/P&L/cash statements to be LLM-ready), and we'd love to get your feedback. Come check it out and let us know your thoughts! Sign up: cloud.llamaindex.ai Docs: developers.llamaindex.ai/python/cloud/l…

@CasmirCodesData This is such a real breakdown of what messy data actually looks like. Most of the work is not analysis, it is fixing inconsistent spellings, missing values, and structural gaps before anything meaningful can happen.

A friend challenged me with a messy healthcare dataset: surgeries, patient outcomes, and AI flags. At first glance, it was intimidating: inconsistent spellings, missing dates, and numbers all over the place. But I knew: every messy dataset has a story waiting to be told. I duplicated the data to preserve the original, then started tidying it up — fixing formats, standardizing spellings, and handling missing values. Already, the hidden structure was starting to emerge. Once cleaned, I dug deeper. The dataset had no admission or discharge dates, just surgery dates and length of stay. I created calculated metrics: • Discharge dates •Success flags for surgeries •Categorized patients as Recovered or Follow-up Required •Converted raw surgery minutes into readable hours and minutes Suddenly, patterns started appearing: •AI-assisted surgeries often had smoother outcomes •High-risk surgeries showed higher complication rates •Some surgeons’ experience clearly affected surgery duration It wasn’t just numbers anymore — it was a story about patient safety, efficiency, and technology impact. Finally, I built a dashboard-ready framework: •KPI cards for quick insights: Total Surgeries, Success Rate, Complications, Avg Surgery Duration, Length of Stay, Patient Satisfaction •Charts for deeper analysis: Success by Surgery Type, Complications by Risk Level, AI vs Non-AI comparisons, Monthly Trends What started as messy rows became actionable insights: •Where AI really helps •Which surgeries need more attention •How patient experience is affected by complications and care efficiency Healthcare Analyst in the chat what do you think about the KPI and what the project overview?

@memovirginia Observing the data first usually reveals hidden spaces, mixed date formats, duplicated rows, or inconsistent headers. Cleaning structure before transformation makes everything more reliable.

One thing data analysis has taught me is: slow down. Before cleaning, transforming, just look at the data. Observe patterns.Notice gaps.Ask questions. Rushing leads to wrong insights. Patience builds better analysis!

@Eyowhite3 The biggest mistake is jumping into formulas before fixing the schema. Clean columns, consistent types, no hidden characters. Once the structure is solid, analysis becomes much easier.

How to deal with very messy dataset in Excel.

@DrubSilvan Messy data really does hold everything together. Clean in theory, chaotic in reality.

@we4v3r Nice. Most AI tools look impressive on the surface, but the real value is when they can handle messy uploads and inconsistent formats without breaking.

today my ai built: • a discovery checklist for criminal defense attorneys • a strategic plan dashboard with gantt charts • a drag-drop upload system to refresh it then it cleaned up messy data, fixed routing bugs, and is now writing this tweet. what did your ai do? 🐙

@AlaniJoshua_ Messy and inconsistent data is the silent bottleneck. A lot of teams think they have an analytics problem, but they really have a formatting and structure problem upstream.

Industry Challenges for Data Analysts and Practical Ways Forward Entering the data analytics field is exciting, but beginners often face real-world challenges that aren’t obvious from courses or tutorials. 1) Messy, Incomplete, and Inconsistent Data Challenge:🧵

@Mineveverexist1 @PerleLabs Data quality is the quiet bottleneck in most workflows. People underestimate how much damage messy formatting and inconsistent structure can cause downstream.

Hello Perle fam! Here is my new post about why human feedback matters in AI and why @PerleLabs is focusing on it. AI is not just about bigger models and faster training. If the data is messy and nobody checks the results, the model will keep making the same mistakes. That is why I like the direction Perle Labs is taking. Perle Labs keeps humans in the loop and treats validation and data quality as the core of the whole system, not an extra step. To me, this feels practical and future proof. Perle Labs is building AI infrastructure that people can actually trust, with clear standards and a community that has real impact. @Meet_Perle @xlordiot @alivinex2 @AhmedZRashad

@joebasshd Messy data is almost always a normalization problem. Mismatched headers, weird date formats, inconsistent encoding. Cleaning that up properly makes everything downstream easier.

I found one of the dirtiest data I've ever come across. Who wants to do some data cleaning?

@SaniNasara_ @TDataImmersed @DabereNnamani @JudeRaji_ @JacobAjala @msexcel This is such an underrated problem. One invisible space or inconsistent header can break a whole pipeline. Most CSV headaches are really formatting and schema issues, not data problems.

Data cleaning is crucial in Data Analysis. An invisible space can ruin your dataset. Always clean your data no matter how clean it appears to be.Last image sheds light on the functions I used in cleaning the dataset😉 @TDataImmersed @DabereNnamani @JudeRaji_ @JacobAjala @msexcel

@RandolfKt Very true. The real value usually hides in messy, unstructured data. Clean benchmarks are rarely what teams actually deal with.