Deputy + AI: 174 tools, and the pay run stops being a surprise
Deputy runs on a connector we build and maintain ourselves, with 174 tools across rosters, timesheets, leave, awards, people and locations. Ask what the fortnight cost while there is still time to do something about it.
The question arrives every fortnight, usually a day after it would have been useful. What did that cost us, and why was it more than we planned?
Answering it properly means the roster in one tab, timesheets in another, leave in a third, and a spreadsheet to hold them still. So it gets done after the pay run has closed, if it gets done at all.
Deputy is now live on FloConnector, on a connector we build and maintain ourselves, with 174 tools. The point of them is that the question becomes a question.
YouPay run closes Thursday. Where are we at?
Claude412 timesheets for the fortnight, 383 approved. Of the 29 outstanding, 22 are at the Northside site and all of them are Kelly's team. Six have no end time recorded, so someone did not clock off. Rostered hours were 3,140 against 3,317 worked, so you are 177 hours over, and 61 of those came from the two Saturdays.
That is four exports and a spreadsheet, done in the time it takes to read it.
What you can ask for
The roster, the clock and the pay rules are all in reach at once, which is what makes the useful questions answerable. A sample of what people ask in the first week:
- Who is clocked on right now, and who has not turned up to a shift that has already started
- Which timesheets are still unapproved, whose they are, and how long they have been sitting there
- What rostered hours cost against what was actually worked, by location, by area, or by person
- Who has leave approved next month, and which shifts that leaves uncovered
- Whose training or certification expires before a roster they are already on
- Which sites went over their scheduled hours last month, and by how much
- What this week looks like if two people call in sick on Friday
Then the follow-up, which is the part that is genuinely new. Every one of those can be acted on in the same conversation: publish the roster, approve the timesheets, move a shift, record the leave. You are not reading a report and then going to do the work somewhere else.
What “174 tools” covers
| Area | Tools | What it reaches |
|---|---|---|
| Awards, agreements, pay rules and pay cycles | 27 | What a person earns, and the rules that decide it |
| Tasks, training, sales metrics, memos and journals | 27 | What was assigned, certified, sold and recorded |
| Employees, addresses, contacts and workplaces | 21 | Who works for you and where they are allowed to work |
| Publishing, shift swaps, offers and approvals | 20 | The verbs Deputy's own app uses to run a week |
| Leave, leave rules, accruals and public holidays | 19 | Who is away, what they are entitled to, what accrued |
| Shifts, availability and events | 18 | The roster, and what stops someone being on it |
| Time clock and the shift verbs | 12 | Clock on, clock off, pause, who is running late |
| Locations, areas and geographic reference data | 11 | Where the work happens and how it is divided |
| Timesheets and their pay lines | 10 | Hours worked, and what those hours became |
| Identity, resource introspection and custom fields | 9 | Who the connection is, and your own fields |
| 174 tools | 88 read, 86 write | 70 flagged sensitive |
The awards row is the one that makes the rest worth having. Plenty of tools can tell you how many hours someone worked. Reaching the pay rules underneath is what turns hours into money, and turns “we were 177 hours over” into “and here is what that cost.”
Once you are past a couple of hundred timesheets the numbers stop fitting in a conversation. Ask about a quarter and the connector loads the set into a temporary, in-memory analytics workspace, the model queries the aggregate, and thousands of rows never enter the chat. There is more on how that works, but you do not have to do anything to get it.
What it will not do, on purpose
This is the first platform we have connected where the records are people, and that changed what we were willing to ship.
It cannot end anyone’s employment. There is no tool here to terminate, reactivate or delete an employee, and the fields that would amount to one, the active flag and the termination date, are not writable either. That second part is what makes it real rather than decorative. Whether someone is employed should not be one instruction away from a chat window, and that holds in both directions, because reactivating a person hands back their roster, their pay and their login.
It cannot touch your webhooks. Deputy’s event subscriptions are account-wide plumbing, and they are usually feeding a payroll or reporting system that nobody in the conversation is watching. Deputy does not replay missed events, so a wrong edit there breaks something downstream quietly. Not even read access.
It cannot change your custom field definitions. The values in your custom fields are fully readable and writable. The definitions themselves are schema, and deleting one abandons everything stored against it, so that stays a settings job for a person.
It never sees payroll identifiers. Deputy keeps tax file numbers and bank details behind a separately gated API that we have not asked for and do not want. That data is not reachable through FloConnector at all.
Eleven tools that could have been in the count are not, for those reasons.
You decide how far it reaches
Deputy has no scope system. There is no read-only mode to request and nothing on the approval screen to tick. What a connection can do is exactly what the Deputy user who authorised it can do in Deputy.
That sounds like a limitation and is closer to a feature, because it means the account you connect with is the dial. Connect as a manager scoped to one venue and your AI sees that venue. Connect as an administrator and it sees the organisation. If it ever refuses something, that is that person’s access, and the fix is in Deputy’s own permission settings.
Then a profile decides which of the 174 tools a given endpoint publishes and who on your team can use it. Describe what the profile is for in your own words and we switch on the ones it needs. Or tick the whole thing read only and hand out something that answers questions and changes nothing, which for most businesses is the right first month.
Your Deputy login never reaches the model either way. We hold the connection encrypted, the AI only ever gets to ask. That is how credentials work on every connector we run.
Connecting takes about a minute
You sign in to Deputy, approve the connection, and you are finished. Nothing is typed by hand, there is no key to generate and no form to fill in.
There is also nothing to buy. Every Deputy plan can use the API, which Deputy confirmed to us directly, so there is no tier waiting behind this and no partner program to join.
One Deputy account per workspace, because a connection is pinned to the install that authorised it and cannot wander into an organisation you did not name. If you run several installs, give each one its own workspace.
How to connect Deputy to Claude
- Create a FloConnector workspace and open the Deputy connector
- Click connect, sign in to Deputy as the person whose access you want your AI to have, and approve
- Create a profile, describe what it is for, and save
- Copy that profile’s endpoint into Claude as a custom connector
- Ask it something. “Which timesheets are still unapproved?” is a good first test
Common questions
Can it fire someone? No. There is no tool that can terminate, reactivate or delete an employee, and the employee record cannot reach the fields that would do the same thing indirectly. That was a decision, not a gap.
Do I need a particular Deputy plan? No. Every tier reaches the API, and our app is registered directly with Deputy rather than through a marketplace, so there is nothing to apply for.
What if it tries something it is not allowed to do? It is refused and you are told why. Access follows the person who authorised the connection, so connecting as someone more senior widens it and connecting as someone scoped to one location narrows it.
Does it work with ChatGPT as well as Claude? Your Deputy MCP endpoint is a standard connector, so it works with any client that supports them. If the term is new, we wrote a plain-English explainer.
Can I use it alongside our accounting? Yes, and it is where this gets interesting. A workspace holds more than one platform, so with Xero beside it, “what did last fortnight’s labour cost, and has it hit the accounts yet” is one question. The rest of the catalogue is at connectors.
What does it cost? Deputy is included on every plan, see pricing. New accounts get free credits to try it. It is in beta, which means connectable today, with every one of the 174 tools already exercised against a live Deputy account, and still proving itself in the wild.
174 tools. Ask it what the fortnight cost while there is still a fortnight left.
Keep reading
More from the blog
MCP Apps: fifty connectors, one widget
MCP Apps is the first official extension to the Model Context Protocol: a server can now render its own interface inside the conversation. Here's why a platform sitting in front of 50+ vendors should build exactly one of them, which one we picked, and what the protocol taught us on the way.
I gave Claude access to my entire business, and filmed what happened
A walkthrough of FloConnector doing real work in real systems: a lead in Gmail becomes a ServiceM8 job, a stale CRM pipeline gets chased and pushed through to a work order, and a supplier remittance marks off a month of invoices. Including the two things it got wrong.
AroFlo + AI: 72 tools, and an answer that isn't quietly a month old
AroFlo runs on a connector we build and maintain ourselves, with 72 tools across jobs, quotes, invoices, bills, purchase orders, scheduling, timesheets, inventory and assets. The interesting part is what we had to do so the answers are the whole answer.