Timings are typical for a firm of 20 to 100 staff and depend on scope.
Five stages, in this order
Whether we are automating one approval or preparing a firm of a hundred for Copilot, the work follows the same sequence. Skipping a stage is the most common reason automation projects disappoint.
Discover
Two to four weeks
We learn how the work really happens, which is rarely how the procedures manual says it does.
Short interviews with leadership and with the people who do the work day to day. We map the processes that matter end to end, review where the data lives and who can reach it, and check what you are already licensed for.
What comes out of it
- Process maps for the workflows under review
- A data and access review: what is sensitive, where it sits, who can open it
- An inventory of the AI and automation tools already in use, approved or not
What we need from you
A few hours of leadership time, and 30 to 45 minutes each from the staff we interview.
Design
One to two weeks
We pick the two or three workflows worth changing first and agree what success looks like before anything is built.
Value, risk and effort decide the order, not novelty. For each chosen workflow we design the new version, decide where a person must review the output, and agree the measures we will report against: hours, turnaround, errors, adoption.
What comes out of it
- A short design for each workflow, in plain language
- Success measures and a baseline taken before the change
- Decisions recorded on data handling, residency and human review
What we need from you
One decision meeting with the workflow owners.
Build
Varies by workflow
Built in your systems, secured, documented and owned by you.
We build in your Microsoft 365 or Google Workspace tenant and your existing business systems, using service accounts and least-privilege access rather than someone’s personal login. Each workflow is tested with real cases alongside the old process before the old process is retired.
What comes out of it
- Working automations in accounts your firm owns
- Documentation of what each one does, what it touches, and how to switch it off
- Short training for the people who use it
What we need from you
Testers from the team that does the work, and sign-off before go-live.
Measure
Thirty to ninety days after launch
We report against the measures agreed in design, including when the answer is disappointing.
Hours returned to senior staff, turnaround, error and rework rates, and whether people are actually using what was built. Where something is not earning its keep, we say so and recommend whether to fix it or retire it.
What comes out of it
- A short results report for leadership
- A recommendation for the next workflow, or a reason to stop
What we need from you
Thirty minutes to review the results.
Support
Ongoing
KDTS keeps it running and secure.
Automations break quietly when a vendor changes an interface, a password expires or a staff member leaves. KDTS monitors them, keeps the documentation current, and reviews the security of every connection as part of its managed services. When your platform adds a new AI feature, we tell you whether it changes anything.
What comes out of it
- Monitoring and fixes for the automations we built
- Access reviews when staff join or leave
- Policy updates when the tools change
What we need from you
Nothing routine. A call when something in the business changes.
Nothing we build should be something you cannot switch off, understand or hand to someone else.
Begin with discovery
Tell us what is taking too much of your senior people’s time, or what your board has started asking about AI. We will reply within one business day and suggest whether a short call is worth it.