Legal Startup
15 Jan 2026
The Case of the Missing Lawyer
A legal startup wanted faster client work without hiring a room full of associates, but no one trusted a generic AI copilot to handle live matters.
The brief was not 'add AI to legal work.' It was to identify one defensible workflow, keep qualified humans in control, and make the system useful enough that busy lawyers would actually use it.
- The startup moved from vague 'AI for lawyers' language to one clear workflow with an owner, controls, and a believable rollout path.
- Lawyers spent less time on low-value intake and first-pass prep, while keeping sign-off where it belonged.
Public-Sector Prime
4 Dec 2025
The Case of the Fraying Bid
A public-sector bid had an AI section full of ambition, but the delivery story would not survive serious scrutiny.
The risk was not just losing marks on elegance. A weak AI section could create delivery doubt across the whole bid and leave the account team defending promises no one wanted to inherit after submission.
- The bid moved from broad aspiration to a believable service story with clearer scope and better operational credibility.
- Internal reviewers had something they could defend instead of hand-waving around 'future AI capability.'
Pilot Rescue
19 Sept 2025
The Case of the Stalled Pilot
An internal AI pilot looked clever in demos, yet every week it ran it created more questions than confidence.
Left alone, the pilot would become one more expensive proof point that 'AI is interesting but not ready.' The real problem was not model quality alone. It was muddled scope, weak acceptance criteria, and too many moving parts for one workflow to earn trust.
- Stakeholders could finally see what the pilot was for, what was being dropped, and what would count as a successful next release.
- The workstream regained credibility because the team was shipping against a narrower, more defensible target.
Retail Operations
8 Jul 2025
The Case of the Overloaded Operations Floor
An operations-heavy retail team was drowning in exception queues, vendor emails, and late-order investigations that stole hours from people who should have been fixing customer outcomes.
The trap was obvious. Many AI demos can summarise a complaint. Far fewer can deal with live operational mess: partial data, conflicting system records, edge cases, and the need for a clear audit trail when money, stock, and service recovery are involved.
- Operators spent less time hunting for context and more time resolving the cases that actually required judgment.
- Leadership got a narrower, more useful AI story: faster exception handling with visible controls, not vague claims about autonomous service.
Financial Services
11 Apr 2025
The Case of the Anxious Compliance Team
A regulated team wanted AI help with policy-heavy reviews, but every conversation ended in the same place: useful in theory, dangerous in practice.
The real issue was not resistance to AI. It was the absence of a workable control model. Without traceable evidence, bounded tasks, and clear escalation rules, the compliance team would never trust the output enough to put it near production.
- The team stopped debating abstract AI risk and started assessing one bounded workflow with visible safeguards.
- Reviewers gained time on evidence gathering and comparison work while retaining ownership of the actual compliance call.
Central Government
30 Jan 2025
The Case of the Vanishing Caseworker Hours
A service team handling evidence-heavy public casework was losing experienced staff time to routing, summarising, and preparing files instead of progressing decisions.
A careless build would only move the problem around. If the workflow could not show what evidence had been used, how a recommendation had been formed, and when a human had stepped in, the service would reject it regardless of speed gains.
- Caseworkers got back meaningful time because routine preparation work no longer began from a blank page.
- Service leaders had a more credible path to backlog reduction because the workflow respected auditability and review discipline.