71–80 of 131 posts
Earlier quoted context omitted.
But what if the third bookkeeper is malicious? You need at least four bookkeepers to achieve byzantine fault tolerance with f=1.
does this assume a maximum of one malicious bookkeeper?
This shouldn't be ignored in the discussion here: The job performed by the humans was broader than what was requested of the model in this benchmark: humans also had to find the relevant invoices (searching through mailboxes, or requesting them from providers) and reason through any circumstances which cannot be inferred from the bank feed and invoices/receipts on their own. In the benchmark these circumstances are p…
If and when a large number of companies blindly turn over their accounts payable workflow to some AI agent system, it'll be very interesting to see the "social engineer the LLM" methods that fraud people use to get money sent to them. Basically the same idea as the ancient "send a fax with a bill for an unsolicited delivery of copier toner to 30,000 businesses" but taken into the modern era. edit: There's already a n…
That's why we have purchase orders that can only be entered by buyers. Product is received and approved by buyer. Invoice goes to accounting, who can't approve it unless there's a matching purchase order and receiver.
Yes, letting agents do whatever they want leads to disaster. But humans are gullible stochastic token generators as well. And that's why the problem is already solved.
Earlier quoted context omitted.
If and when a large number of companies blindly turn over their accounts payable workflow to some AI agent system, it'll be very interesting to see the "social engineer the LLM" methods that fraud people use to get money sent to them. Basically the same idea as the ancient "send a fax with a bill for an unsolicited delivery of copier toner to 30,000 businesses" but taken into the modern era. edit: There's already a n…
You can fix this simply by using normal controls. That's why we have purchase orders that can only be entered by buyers. Product is received and approved by buyer. Invoice goes to accounting, who can't approve it unless there's a matching purchase order and receiver. Yes, letting agents do whatever they want leads to disaster. But humans are gullible stochastic token generators as well. And that's why the problem is…
I've seen so many reports of humans with the authority/ability to execute an outgoing SWIFT transfer who've been social engineered into sending money to fraudsters... Or even just the basic low level "Hey I'm your boss sending you an SMS, please go buy some gift cards and scratch them off and send me the codes". No AI involved whatsoever.
The danger exists where some true believer AI evangelist type of management person tries to fully automate the entire purchasing and AP workflow, which I'm sure some people will attempt soon, with varying degrees of success.
This shouldn't be ignored in the discussion here: The job performed by the humans was broader than what was requested of the model in this benchmark: humans also had to find the relevant invoices (searching through mailboxes, or requesting them from providers) and reason through any circumstances which cannot be inferred from the bank feed and invoices/receipts on their own. In the benchmark these circumstances are p…
If and when a large number of companies blindly turn over their accounts payable workflow to some AI agent system, it'll be very interesting to see the "social engineer the LLM" methods that fraud people use to get money sent to them. Basically the same idea as the ancient "send a fax with a bill for an unsolicited delivery of copier toner to 30,000 businesses" but taken into the modern era. edit: There's already a n…
This shouldn't be ignored in the discussion here: The job performed by the humans was broader than what was requested of the model in this benchmark: humans also had to find the relevant invoices (searching through mailboxes, or requesting them from providers) and reason through any circumstances which cannot be inferred from the bank feed and invoices/receipts on their own. In the benchmark these circumstances are p…
AI helps automate things that didn't already have rigorous formatting and structures available as input... and that's really all it does (99% of the time). Doesn't matter how many more nines you add, rigorous formatting is still required . In some cases, it has teeth with compliance standards. Those standards cannot be compromised because there are already a lot of other layers contributing inaccuracy. It all adds up…
I don't get this argument. People say the same about autonomous driving.
But humans also have some number of nines. If you can get it better than humans, that's better!
This doesn't surprise me at all. You can really constrain this problem, give very narrow context, and get pretty reliable and reproducible results. I've gotten very good results with some vibe-coded deepseek book keeping. https://github.com/traverseda/beansync Parses emails or other sources, extracts numbers, correlates different transactions, web search, asks questions, stores notes (regex based, very simple). The h…
Note to self: traverseda doesn't have 2factor auth on his email and his LLM seems to have full access. Hmmm
Also this isn't really an agent. At least not a long lived one. Each email or transaction gets it's own session, the llm can make a few tool calls but must emit json as the final result. Very very short context lengths, very predictable results.
The bot does not have access to my passwords, there are pre-defined scripts that fetch the data that the run an LLM call for each transaction discovered.
Earlier quoted context omitted.
Hey, the author of the benchmark here. The benchmark data was prepared in April 2026 (when I was manually doing our VAT return with my co-founder). The invoices were indeed found manually. Currently we're using a custom "invoice searcher" built on Kimi 2.6 (in our testing several weeks ago it outperformed Opus 4.7; it was just more persistent). Ultimately, I still verify everything manually after the model is finishe…
adam, i'd like to get in touch and would love to run the benachmark with mixedbread as a search backend. we are doing this right now with a lot of compliance companies. would be very curious how it improves quality/cost e2e
Earlier quoted context omitted.
You can fix this simply by using normal controls. That's why we have purchase orders that can only be entered by buyers. Product is received and approved by buyer. Invoice goes to accounting, who can't approve it unless there's a matching purchase order and receiver. Yes, letting agents do whatever they want leads to disaster. But humans are gullible stochastic token generators as well. And that's why the problem is…
Indeed so, a fairly mundane RFP, RFQ, buyer, receiver, accounts payable process will stop a lot of problems. If an agent is inserted at some stage in the process with a clear path to make a ticket/escalate to a human if it sees something it doesn't understand, the risk isn't absurdly high , in my opinion. I've seen so many reports of humans with the authority/ability to execute an outgoing SWIFT transfer who've been…
With the SWIFT incident I saw, there was a rule that no payment can go to a vendor's bank that isn't a current, approved vendor. But the rule was not enforced in software: it was an internal accounting rule that humans were supposed to follow. The AP person "thought it had been approved" because there was a similar transaction with a different company that was a new vendor at a similar time. The other transaction was legitimate, the fraudulent spoofer wasn't. The wire got sent to a party in China.
With AI agents, if you approach it from the perspective that it will be gullible and trickable by fraudsters, you build in these hard guardrails. With humans, it's much easier to believe that "we trained Lucy on this procedure" will work in all circumstances, even if Lucy still has the technical ability to bypass the official procedure.
In these cases, it starts looking a lot more like traditional software, with your little AI chaos monkeys constrained in little boxes within the software chain.
This shouldn't be ignored in the discussion here: The job performed by the humans was broader than what was requested of the model in this benchmark: humans also had to find the relevant invoices (searching through mailboxes, or requesting them from providers) and reason through any circumstances which cannot be inferred from the bank feed and invoices/receipts on their own. In the benchmark these circumstances are p…
It's not completely reliable and the human cannot be taken out of the loop, but the number of menial tasks she's been able to automate has been really cool. A lot of processing data that arrives in non-standard formats, generating documents based on that data, etc.
She still has to review everything, but her workload is way down, and when her assistant quit she automated away his whole position.