Business automation

Automation projects fail when they automate the process as described rather than the process as performed. So we start by measuring where the time actually goes, which is almost never where the process document says. Then we automate the highest-volume, lowest-judgement steps and leave the judgement to people. Calder & Voss automated intake, conflict checking and engagement-letter generation; the partners kept the judgement calls and the administrative team stopped spending eleven hours a week on document assembly.

• 01

  Measured before and after, in hours and error rate. An automation nobody measured is an automation nobody can defend at budget time.

• 02

  Exception handling designed first. The 15% of cases that do not fit the rule are where the value leaks, and routing them to a person quickly is part of the build, not a gap in it.

• 03

  People are moved off the repetitive step, not out of the process. Automations that quietly hide their failures cost more than the manual work they replaced.

• 04

  Auditable by design: every automated action is logged with its inputs and its trigger, which matters as soon as a regulator or a customer asks what happened.

• 05

  Built on the systems you own where possible — the workflow engine you already licence beats a new platform nobody owns.

1. 01

   Measure the process

   Two weeks of observation and system-log analysis to find where time and rework actually accumulate. This routinely disagrees with the process documentation, and the disagreement is the finding.

2. 02

   Pick the candidate

   High volume, low judgement, clearly measurable. We rank the options and start with one, because a first automation that works buys the mandate for the next four.

3. 03

   Design the exception path

   What the automation refuses to handle, who it goes to, and how quickly. Agreed with the team who will receive those exceptions, before the build.

4. 04

   Build, pilot, compare

   The automation runs alongside the manual process on live work for a fortnight, and the outputs are compared. Where they disagree, the rule changes.

5. 05

   Roll out and re-measure

   Full deployment with a dashboard showing volume handled, exception rate and time saved, reviewed monthly against the original baseline.

• Process baseline: measured time, volume and error rate per step, before anything changed.
• The automation itself, deployed, with source in your repository.
• Exception routing configuration and the service expectation agreed with the receiving team.
• Audit log of every automated action, with inputs and trigger retained per your retention policy.
• Dashboard covering volume, exception rate and hours saved.
• Post-implementation report comparing measured outcome against the baseline.

• 01Does this mean redundancies?

  That is your decision, not ours, and it is worth being straight about. What we can tell you is what actually happens in most of these engagements: the automated step was the part nobody wanted, the team absorbs backlog that had been permanently deferred, and headcount stays flat while volume grows. If your intent is a headcount reduction, plan it openly — the team will work out the purpose of the project in week one regardless.

• 02Do we need AI for this?

  Frequently not. A large share of what gets pitched as AI automation is a rules engine, a scheduled job and a well-designed form — cheaper to build, cheaper to run and far easier to audit. AI earns its place where the input is unstructured: reading varied documents, classifying free-text enquiries, drafting a reply for a human to approve.

• 03What happens when the process changes?

  Rules live in configuration your own team can change, with the boundary between configurable and code documented explicitly. Anything an operations manager should reasonably be able to adjust is not buried in a deployment.

• AI

  AI and generative AI solutions

  Assistants, chatbots, voice agents and generative features that are wired into your actual systems and measured against a baseline.

  /services/ai-and-generative-ai
• API

  API and system integration

  Making the systems you already own talk to each other — payments, CRM, ERP, messaging and the third-party APIs in between.

  /services/api-system-integration
• DGT

  Digital transformation

  Multi-year programmes that digitise a business and modernise what it already runs — sequenced so each stage pays for the next.

  /services/digital-transformation