browolf

Broadcom have ditched the small resellers and we have to renew with a larger UK supplier. The cost of Vsphere standard has gone up to ~ £60 per core minimum 96 which we are renewing this month. After October they will only offer Cloud Foundation licenses which costs around £125 per core. No one is going to agree to that. Our service partner that supports Vmware in schools is already developing a migration to Hyper-V solution.

My Head has an idea that ai could be used to condense send data into summaries for teachers. Apparently one of the problems in teaching SEND students is that there's so much data that it's impossible for teachers to know their send students very well nor how they should adapt their teaching. I'm not sure there's an off the shelf ai that could handle this kind of thing and building/vibing our own LLM/system is probably impractically costly and way out of our skillset . This sounds like an actual product. Anyone seen a product that sounds anything like this?

Just looking for data transcription tools - slt are waking up to it. Suggested Scriba and Neurameet to them. Both look good

I use pfftools to extract the contents of PST files and then a python script that uses beautiful soup to extract emails into text messages and then some other scripts to get rid of irrelevant messages and perform redacts. The process will often start with around 3000 emails that i can reduce to 500-1000. A lot of emails are lists are that aren't specifically about the subjects. The advantage of this kind of thing is is being able to design the final output which is 1 text file where individual emails are split with dividers containing dates and subject lines. If you have some familiarity with python and time for vibe coding check out my code https://github.com/browolf/Subject-Access-Requests-from-Microsoft-Copilot-365 The digital officer is using acrobat pro to redact pdf documents from cpoms and SEN with redaction lists

how do i stop staff from seeing other staff's personal data? Edit: I think it's 1 slt who already had those permissions in SIMS and didnt realize it

What is the support like?

I'm looking for a supplier of kyocera copiers to quote for a lease.

I us The M365 copilot app in win11 signed in with my 365 account then paste the code into vs code to test. I also have a chatgpt account the school paid for because previously copilot was really rubbish at coding and persistence but it's got massively better in the last 6 months so probably dont need this one as much.

According to an exam inspector that randomly turned up today, the future online exams will be marked by ai and written by ai. I'm not sure if the exam boards will be saving much money though as they will be spending it on ai compute instead I imagine.

We looked into moving once but it's an incredibly complicated endeavor with a big school. In the end they decided to get Synergy which reads and writes to sims but the Ui is a million times better. https://www.schoolsynergy.co.uk/

I don't think you need agents for something like this. That's overcomplicating it. You need to run the script locally to see what happens. If there's an error paste it in the chat and it tells you what to fix. Often it's easier to just ask to get the full updated script again. First I made a proof of concept in chatgpt - that it was possible to redact pdfs using python - with this prompt write a python script to redact text in pdf files. Redacted text will be replaced with the text <redacted> the script will ask for an input then redact that input in all the pdf files in the same folder. The script will show the names of the files in which the text was redacted and how many times. The script will keep asking for inputs or (q)uit You can see that process here; https://chatgpt.com/share/6a0c137f-2ebc-8386-aece-e27035353d6e Then I switched to copilot because I thought I would need to paste screenshots and you don't need to redact data with copilot because it's held within the 365 tenant enclosure. In copilot I started with the python file and the prompt "read this file comment ok after" then the real prompt New script with the same principle and style as this script script that redacts all the pdf, xlsx or docx files in a folder from a text file of redaction words or phrases this ui only has 2 columns first column asks for the text file the 2nd file is the output 1. list of files then what redactions are made to each file backups occur as before when the text file is loaded it needs to be sorted in the order, phrases then words, phrases may include numbers split so really the sort is items per line ask me any questions before giving me the code answered questions got code tested code, go through a bunch of errors and iterations however by the end I decided to drop docx and xls and focus on pdf started a new chat with the file and the prompt "redacted words are added to a txt file called redacts.txtwhen the program loads it looks for redacts.txt and asks to redact files using this file straight off. This process needs to run in a thread so that output can still be generated" go through another load of iterations until it operates how you like it.

I just got this book on Audible Jamie Bartlett - How to talk to Ai (and how not to) - He's done various podcasts about tech and if you wanted an intro he was just on the Bottleneck podcast taking about various things and Ai with Rory Sutherland. Never a dull moment when RS is involved.

I vibe coded a tool to redact PDFs - it's way faster than Acrobat Pro - I have converted it to an exe file so the data officer can use it. https://github.com/browolf/Subject-Access-Requests-from-Microsoft-Copilot-365/blob/main/PDF Redactor/pdf_redactorV3.py

Now you mention it ediscovery premium does deduplication. By default it's at 65% similarity but I haven't used it enough to know if it can be higher. I vibe coded our own processing for emails but redacting documents is still a massive pita and we are looking at products to do that

I don't think that is an option. You could potentially do that later, depending on how you are processing the exports