How we charge
We agree the number we are measured against before anything is built. The clock starts the day the system goes live. If it misses the bar, we keep working at no additional cost until it clears.
Run your own numbers
Put your own headcount, hours and hourly cost in. Whatever this prints is the kind of figure that goes on the signed page, once we have measured it with you rather than guessed it.
3 people spending 6 hours a week on manual work costs about $34,560 a year. That is 864 hours you could put back on the floor.
3 people x 6 hrs/week x 48 weeks = 864 hrs/year 864 hrs/year x $40/hr = $34,560/year
An estimate from numbers you enter, not a promise. 48 working weeks assumed. Your real figure depends on the process we replace.
On the Uniserve engagement the target is roughly 1 hour of a technician's day given back by the customer-facing chatbot and voice agent. One named process, one number, measured the same way before and after. That is the shape of a figure we will sign.
1 hr/day x about 250 working days x $40/hr = about $10,000 per technician per year. Scales with team size. This is a projection agreed as a target, not a booked result.
This module runs on numbers you typed. The signed baseline runs on numbers we measured together in the audit, on the process we are actually replacing, using a method we both agreed to before the build.
Same arithmetic. Different evidence. Only the measured version goes on the page.
The terms
Most AI work is sold on a promise and invoiced on effort. We would rather be invoiced on a number you signed off on.
A signed baseline before any build
We measure the process as it runs today and write the numbers down. Hours per week on a named task, requests per month, how long a quote takes. You sign off on the exact figures we are measured against. Nothing is built before that page exists.
The clock starts at deployment
Not at signature, not at kickoff. The measurement window opens the day the system is in your team's hands and doing real work. Build time is our problem, not yours.
If it misses the bar, we keep working at no additional cost
Same scope, same team, no new invoice, until the agreed bar clears. That is the whole of the guarantee. We are not offering a cash refund, because a refund pays you for a system that still does not work.
Cancel at any time
No minimum term and no exit fee. If you want to stop, you stop, and you keep what has been built and delivered to that point.
A free operations audit up front
Before money changes hands we map where time and money leak across your functions, score the candidates on return and feasibility, and show the arithmetic. You keep that map whether or not you hire us, and it is the document the baseline is built from.
You own the code
The repository, the data and the deployment are yours. No licence to renegotiate, no hostage fee to leave. Two founders build it and two founders hand it over.
The bar and the timeframe are set per engagement, in writing, because a support queue and a multi-site stock count do not move on the same schedule. What never changes is that you set the number with us before the build, and you can read the arithmetic.
The sequence
Five steps. Nothing is built until you have signed the number it will be measured against.
We sit with the people doing the work, map where the hours go, and score the candidates on return and feasibility. You leave with the map either way.
One page: the process, the current numbers, the bar, the timeframe, and how we will measure it. Both names on it. This is the document the guarantee is enforced against.
We build the two or three things that move the signed number, not a platform. Uniserve's programme runs three months, and the customer-facing chatbot and voice agent are in build now.
Live, in your team's hands, on your real workflow. This is the day the clock starts, not the day you signed.
Same method, same process, arithmetic printed. The fee is the fixed one quoted before the build. If the bar has not cleared, we keep working at no additional cost until it does.
Why we can offer it
These are the published numbers on the industry we work in. They are the reason a guarantee is worth anything, and the reason ours is narrow.
of enterprise AI pilots show no measurable P&L return.
MIT NANDA, The GenAI Divide, 2025of AI projects fail, about double the failure rate of other IT projects.
RAND, 2024of AI budgets go to technology and about 7% to the people expected to use it.
Deloitte, 2025Read those three numbers together and the failure is not the technology. It is scope set too wide to measure and adoption nobody paid for. Deloitte's split is the tell: money on the tool, almost nothing on the people who have to use it.
Carrying the risk changes how we work. If we are on the hook to keep working until the number moves, we cannot afford a twelve-month platform with a vague benefit. So we scope narrowly, pick the two or three problems with the clearest arithmetic, and measure the process before we write code. That is also why the audit is free: it is how we decide whether we can stand behind a number at all. Sometimes the honest answer after an audit is that the leak is a process problem, not a software problem.
Two founders do the work, so there is nobody to hand the risk to. Shiv and Vishal built Fennec, which runs live across Harbour Event Centre, TradeX, The Pit UBC, The Show Ottawa and 20+ event companies, and shipped the tip-out software that ended gratuity disputes at Palapa Tours. When the bar is missed, the people who built it are the people who come back.
What we need from you
The guarantee only works if the baseline is real. These are the conditions that make it real.
Time with the people who actually do the task, and a look at the system they do it in. Not a description of the process in a meeting. The Uniserve plan exists because we sat with the work itself, function by function.
A sanitised version of the work is a guess, and we cannot guarantee a guess.
One person who can approve the baseline, settle a scope question in a day, and say no. Owner-led businesses are our whole practice because that person is usually in the room already.
No account managers on our side. We ask for the same on yours.
You read the numbers, you challenge them, and then you sign them. If we cannot agree what today looks like, there is nothing to measure and we will say so before you spend anything.
Also worth naming: real constraints, early. Uniserve's data and models all have to be hosted in Canada, and that shaped the build from day one.
Questions we get asked
If a term below is not clear enough to hold up in a disagreement, tell us and we will rewrite it.
A named process, its current numbers, and the method we will use to read those numbers again later. Hours per person per week on a specific task. Requests per month, and how many of them repeat. Days to turn a site visit into a quote. Whatever the process actually produces, counted the way your systems already count it.
It is one page, it is written before anything is built, and you sign it. The bar and the timeframe sit on that same page and are set per engagement.
It happens. Volume doubles, a location opens, a team shrinks. When the underlying volume moves materially, we re-measure the same process with the same method and write the change down, with the reason, next to the original figure.
The rule is that the bar never moves quietly and never moves in only one direction. If your volume tripled, the honest baseline is the new one. If it halved, that is on the page too.
This is what the method clause is for. Because we agreed how to measure before deployment, a disagreement is about arithmetic rather than opinion, and arithmetic can be re-run. We re-run it, in front of you, from your own data.
If you still think the bar was missed, the guarantee applies: we keep working at no additional cost. And you can cancel at any time, so you are never arguing from a locked-in position.
You do. The repository, the data and the deployment are yours, and it is written into the agreement before anything is built.
Where a build sits on a partner platform we name it plainly at the start. On the Uniserve programme, all data and models are hosted in Canada, and that requirement decided which platforms were on the table at all. You will always know what you own outright and what you are running on.
Your choice, and there is no default that quietly renews. You can take the code and run it in-house, since you own it. You can keep us on a defined next scope with its own signed baseline. Or you can stop, and we hand over documentation and access.
Palapa Tours is the shipped version of that: the tip-out software calculates, splits and tracks staff gratuities, the disputes stopped, and the tool keeps running without us standing over it.
A fixed fee, quoted in writing before the build, on the same page as the baseline. No hourly billing, no change-order drip, and no percentage of a number we calculated ourselves.
We quote in plain figures, and the figure sits in the agreement next to the baseline it will be measured against, before the build starts. We do not publish client fees and we do not publish pricing tiers, because a tier is a guess at your problem. Yours comes out of the free audit, in writing, and you can read the arithmetic behind it.
Start with the audit
Thirty minutes with the two people who would do the work. You leave with where the hours go and what it would take to get them back, whether or not you hire us.
The plain version of the qualifier: the bar and the timeframe are agreed per engagement, in writing, before anything is built. Nothing on this page guarantees a particular dollar amount, a multiple, or a cash refund. What is guaranteed is that if the system misses the bar you signed off on, we keep working at no additional cost until it clears, you can cancel at any time, and you own the code either way.
shiv@exdsconsulting.com
