AI root-cause investigator for hardware engineers
Find the root cause nobody else could.
Bring the machine problem nobody can explain, and the tool's own manuals. Cala works it like an investigation: every kind of cause on the table, the evidence for each quoted from the page it came from, and the one check that decides it. It runs on your own computer.
Cala is a desktop app that investigates machine problems with hardware engineers. Give it the manuals for one tool and the symptom as you would tell a colleague. It maps every kind of cause, ranks the hypotheses against the evidence, cites every number to the page it came from, and names the one check that separates them. When you confirm the fix, the root cause is recorded for the next time. Everything runs on your own computer.
01 · The situation
The tool is down, and everyone is looking at you.
Nobody holds a machine this size in their head. Not you, and not the engineer who has been on it since install. What experience gives you is a feel for where to look first.
02 · The loop
Escalate it, and what comes back is a list.
Cala gives you ranked hypotheses in about a minute, each with its causal chain written out and the page it rests on. From there the loop runs at your speed, not the queue's.
03 · One case, start to finish
What an investigation looks like.
Not a longer checklist. A ranked cause with the evidence behind it, the one check that decides it, and the root cause recorded once you confirm the fix. This case is real, retold on a different machine so nobody can be identified.
- 01 Problem
Off by a little, in one corner, now and then
A precision laser drilling machine placed features slightly off, only in one region of the field and only sometimes. Most parts measured clean, so the escapes reached the next step first.
- 02 Clean checks
Two weeks, every subsystem in spec
Optics, motion, power, environment, alignment: all normal. No alarm, no trend that moved with the issue. By every measurement it could make of itself, the machine was perfect.
- 03 Hypothesis
Not a component. A configured term
Cala got the symptom as you would tell a colleague. Its top hypothesis was a software compensation term across the field: wrong by an order of magnitude, the machine looks healthy everywhere and is wrong in one region.
- Field compensation reference, rev D · p. 31
- Machine configuration guide · p. 12
- 04 The check
Read one value, compare one record
Read the configured value and compare it with the commissioning record. Minutes of work. No teardown, no parts.
- 05 Confirmed
Ten times its intended value
Restored, the issue disappeared and the machine requalified. The first hypothesis on the list was the one, on a machine where nothing measured abnormal.
- Open before Cala
- Two weeks
- Hypothesis to confirmation
- First check
- Signals showing abnormal
- None
A real case from early testing on production equipment, retold on another class of machine with the same causal shape.
04 · What you get
The cause is hiding two subsystems away.
Cala reasons in causal chains. It starts from how the machine actually works, looks at every kind of cause at once, and follows the physics across subsystem boundaries that no one person owns. You get the chain written out end to end, in minutes, with the evidence for every link.
05 · See it work
One issue. Candidates across every kind of cause.
Here is one issue on a tool. Cala does not jump at the first likely answer. It works through every kind of root cause the machine has, and each candidate names the page it came from and what to go and measure next.
Load lock B reaches transfer pressure 38 s late, but only from the third wafer of a lot onward. Load lock A on the same tool is unaffected. No alarm, no error code.
Kinds of root cause
Candidates · Thermal
Every kind above is a different thing that can go wrong, not a different way of saying the same thing. Cross one off and the problem is smaller.
06 · Check it yourself
Every number comes with the page it came from.
Values show up the way they're written in your document, with the document name and the page, so you can go and check. Cala doesn't invent numbers. If a number isn't in anything you've loaded, it tells you the evidence is missing instead of filling the gap.
07 · The other ending
Some nights end differently.
The causal chain is on the screen in minutes, the check comes back positive, and everyone in the room hears what it was and why. The tool comes back up, and the night ends at a reasonable hour.
08 · Hand it over
Hand it over with the citations still attached.
When it's time to hand it over, write it up: a service report for the ticket, a summary for the next shift, or a hand-over for the customer. It goes out with what you checked, what each check ruled out, and the page you read it from.
09 · Where things live
Your documents stay on your computer.
Cala runs on your computer, and your documents live there. When Cala reasons, the pages it's reading pass through our proxy to the model provider and come straight back. Nothing is stored on our servers.
10 · Questions
The questions people ask first.
Written the way they're usually asked, because those are the ones worth answering. The long versions are on the security and privacy pages, and the proof is in the case studies.
Can it actually solve something my whole team is stuck on?
It has. A machine down for two weeks, every sensor in spec, no alarm, no trend. Cala’s top hypothesis was a software compensation term nobody had questioned. Reading the configured value took minutes, and it was about ten times what it should have been. Restored, the machine requalified.
First hypothesis on the list, confirmed on the first check. Two more investigations are in the case studies.
What does it do when I hand it a real problem?
It works the problem, not the chat. Every kind of possible cause goes on the table: thermal, mechanical, electrical, contamination, control and software, alignment and metrology. Hypotheses come ranked. Evidence is quoted with document and page, with what it read kept separate from what it inferred.
Checks are ordered by what each one rules out, so a quick check that kills three hypotheses comes before a slow one that confirms a favorite. It ends in a report file you can attach to the ticket as is.
Nobody has ever seen my problem. There is nothing to search.
That has happened. A dimensional oscillation in dense sections with no write-up anywhere. Cala proposed a material-dependent mechanism nobody had put together, designed experiments to separate it from the alternatives, and attached a prediction to each one. The experiments ran. Every prediction held.
When there’s no page to find, it reasons from the physics in your documents.
It will just tell me what the manual already says.
The manual is where it starts, not where it stops. One process model had been trusted for years, but the fits never settled on one parameter set. Cala asked for the one thing nobody had put on the table, the model’s implementation, and found a quadratic where the physics calls for an exponential. Corrected, every structure collapsed onto one parameter set.
It has no loyalty to what everyone already believes.
It’ll make up a part number. They all do.
Don’t trust it. The product isn’t built on the idea that you will. Every value that decides anything carries a citation with the document and the page, quoted as written. Inferences are labeled as inferences. Missing evidence is called missing, not papered over with a number of the right shape.
Test it the cheapest way there is: load a manual you know cold and try to break it. There’s a ready-made protocol for exactly that.
I have search, and I have ChatGPT.
Search gives you what you thought to ask for. A chatbot answers the pages you pasted in. Cala tells you what else it could be, and it remembers the machine: read the documentation once and the knowledge base for the whole tool sits on your disk, there every session after. No re-uploading PDFs at 3 a.m. on shift.
My documents are OM and PM manuals, BKMs and vendor training decks, half in Chinese.
Load them as they are. That pile of vendor PDFs for one tool is exactly what Cala is built for: it reads them into a knowledge base for that machine, mixed Chinese and English without being told which is which, and answers in whichever language you ask. Citations name the document and the page either way.
The first users are semiconductor equipment and process engineers. Any machine with real documentation works the same way.
What leaves my computer?
Your documents, and everything Cala learns from them, stay on your computer. Our servers hold your email, plan and usage: no document content, no chat content, and we train nothing on any of it. While Cala is reasoning, your question and the pages it’s reading go to the model provider through a proxy that keeps nothing. Full detail on the security page.
Do I need my company’s permission?
That’s between you and your employer’s rules about software and documents. Cala is a personal subscription, not a way around anyone’s policy. If that conversation needs to happen, the security page says exactly what leaves your computer and what never does, and it was written to be shown to the person who approves things.
A hundred a month is real money, and it’s my money.
20, 100 or 200 USD a month, depending on how much troubleshooting you do, plus prepaid credits when a bad week uses the plan up. Personal subscription on your own card, no procurement. Cancel any time from your account page; it stops at the end of the paid period, and everything on your disk stays yours. Read a case study first and decide from that.
What do I need to run it?
macOS 13 or newer on Apple silicon (M1 or newer). Windows 10 or 11, 64-bit; the Windows build isn’t code-signed yet. Internet only while Cala is working a problem. The knowledge base is local and stays local.
Bring the one that's still open.
Load the manual for the tool it's on and give Cala the issue that's been sitting there with no root cause. Read the list. Check the citations against the pages they point at. See if there's something on it nobody has tried yet.