Week 1
Baseline
Pick one priority process and measure what it currently costs in capacity, delay, rework, missed opportunity, or margin. Agree the result and the threshold that counts as success.
How we work
A free call. A paid assessment. Then only what the case supports.
The introductory call is free. The assessment is paid, and its report is useful whether or not Halcroft builds anything. Implementation is proposed separately, and support only where it is needed.
One accountable owner. One agreed measure. Each stage its own decision.
Using AI is not the same as operating an AI system.
Ad-hoc AI helps one person complete a task. A governed delivery system changes how the firm runs the process.
Ad-hoc AI use
Governed delivery system
Helps one person complete a task
Runs a defined part of an accountable process
Depends on individual prompts and habits
Uses approved inputs, instructions, and ownership
Output is checked inconsistently
A named person reviews defined decisions
Activity is difficult to trace
Inputs, outputs, and approvals are recorded
Time saved is assumed
Capacity, turnaround, quality, conversion, or margin is measured
Halcroft finds the workflow most likely to constrain growth, implements AI or automation where the business case is strong, and measures the operating result. The same person remains accountable through assessment, build and measurement.
From first call to handover
Assess broadly. Build narrowly. Scale on evidence.
The assessment looks across the workflows. The first build stays deliberately focused, creating evidence before a wider programme is recommended.
Introductory call
45 minutes, free. We look at where capacity is constrained and whether an assessment is worth doing. It is a conversation, not an audit.
Paid operational AI assessment
Fixed fee, scoped before it starts. Interviews, process and systems review, a baseline where the data allows, ranked opportunities with the financial assumptions, risks, and what not to automate. The report stands on its own, whether or not Halcroft builds anything.
Scoped implementation
Separately proposed and only where the assessment supports it. Agreed problem, acceptance criteria, data and security arrangements, human review points, training, handover, and measurement of the result.
Ongoing support
Optional, and only where there is a real recurring responsibility. Separately agreed, with its own fee and service scope. Many engagements end at handover.
The system build
Four weeks, with an operating baseline at both ends.
Week 2
Design and configure
Redesign the process, assign ownership, connect the required data and tools, set permissions, and place the human review gates.
Week 3
Test and launch
Test output quality against representative documents, onboard the delivery team, then run the system on live or representative work.
Week 4
Stabilise and report
Monitor usage, adjust, then report measured impact against the agreed result alongside a roadmap for the next process or team.
Built to survive growth
Process, ownership, controls, and technology move together.
A system is only scalable when the team knows who owns it, how decisions are made, and where human judgement remains essential.
Scoped permissions
The system reaches the repository and the matter it was built for. Nothing wider.
Human review gates
Nothing reaches a client without a named person approving it. The gate is part of the workflow, not a policy note.
Audit trails
What went in, what came out, who approved it. Answerable to a risk committee, not just to a demo.
Approved tools only
A written starter policy on what staff can and cannot put into AI tools, so the informal use already happening has somewhere to go.
What the firm keeps
- Constraint baseline
- Shows where the current process is restricting capacity, consistency, or growth.
- AI use and data policy
- Clarifies what staff can and cannot put into AI tools.
- Operating playbook
- Documents the new process so it can be repeated.
- Team training and adoption materials
- Training, examples, and office hours for the delivery team.
- Results review
- Explains the measured result and scale-up recommendation.
- Scale-up roadmap
- Shows how the first system can expand across teams, services, or portfolio companies.
The System Build Guarantee
The first system has to earn the next one.
If the system does not meet the operating success criterion agreed at kickoff, we will continue working on that process for up to 30 additional days without charging professional-services fees.
Conditions: the client provides the agreed documents, data, system access, subject-matter experts, user testing, and process usage. The guarantee applies only to the agreed system and operating result. It does not guarantee revenue, valuation, or new client wins.
A useful first conversation
Bring the growth plan. Leave with a clearer operating decision.
In 45 minutes, we will identify the process most likely to constrain growth, test whether a system would help, and say plainly if it would not.