CLEARDESK
We build it. Your company’s manual work, done by software.
AI and automation for the rule-bound knowledge work of a company, connected to the systems you already run.
ONE PROMISE
The documents, requests, approvals, and records that eat your skilled people’s time get done by AI-driven software your people actually use. It is connected to the systems you already run, “done” is written down before we build, who uses it, on which task, and every hour and every AI token is on the invoice, at cost.
THE PROBLEM WE SOLVE: every company has work that follows rules and still takes skilled people days, writing an RFP, routing an approval, answering the same request, keeping the record straight. AI vendors sell pilots that impress in a demo and never reach production. Integrators sell hours. Nobody defines “working” before the build starts, so nobody can be held to it, and the company is left with a tool it pays for and doesn’t use.
HOW WE DELIVER IT
1
Is It Worth Doing?
One process, one week. How much skilled time it eats, what the systems around it will allow, and a verdict, including “don’t automate this, fix the process first.”
2
Requirements and Design
Detailed requirements, the solution design, the integration points with your existing systems, and the written definition of “done”, who uses it, on which task, measured how. The build estimate comes out of this, not before it.
3
Build It to Production
Build, test, connect, deploy, against the design and the definition of “done” from step 2. Hours only for days actually worked, under a plan and ceiling you sign.
4
Run It
Licence, support, and improvement after go-live. You see usage; you decide yearly: continue, adjust, or stop.
Already built, for the process we know best. Our first product is for the organizations that issue RFPs. It reads your templates and past procurement documents, builds your library of clauses and company wording, and drafts an RFP, RFI, or RFS section by section with AI , while your team works in it together, verifying as it goes.
“Done” is written down first. Step 2 ends with a signed document: who will use the software, on which task, what the measure is — cycle time, error rate, hours returned — and what it connects to. The build runs against that document, under a plan and a budget ceiling you sign, billed for days actually worked. AI usage appears on your invoice at cost. There is nothing to argue about at go-live, because the test was written before the first line of code.
“The desk is clear when the work is done — not when the software is installed.”
ClearDesk is a LUKAPS branch · Toronto, Canada