The practice.

Practice

Four areas. One practice.

Every engagement is small, senior-led, and directly delivered. A hard ceiling on how much work we take on in a year.

§ 01 / 04

AI

AI work spans two distinct registers. We design and build AI enrichment, scoring, and automation inside CRMs and operational platforms, where the system needs to be trustworthy as well as clever. We also help engineering organisations change how they build software in the agent era, moving teams from writing code to orchestrating agents.

AI-integrated business systems

The work pairs large-language models with real business logic: governance, retries, cost controls, audit trails. The system needs to be trustworthy as well as clever.

  • AI opportunity enrichment: engagement scoring, activity rollups, 90-day tracking, FranklinCovey indicators, and a tabbed LWC insights panel with a live refresh button.
  • AI account intelligence: five-phase programme taking account-level understanding to parity with opportunity intelligence, including external data (Companies House, Google Custom Search, D&B) and scheduled re-enrichment.
  • AI lead enrichment: public web-context gathering with confidence scoring, engagement tone detection, and GDPR-compliant source tracking.

AI Transformation

Helping engineering organisations move from writing code to orchestrating agents. Spec-driven delivery design, cultural transition for teams used to traditional implementation, evaluation of the agent tooling landscape, and the measurement that proves the new model is moving the needle.

  • Methodology design: partnering with engineering leadership to design the spec-driven, agent-first delivery model for a team or function. Architecture of the operating model, definition of standards, the cultural change programme.
  • Tooling evaluation: short-form engagement to assess which agent tools (Claude Code, Codex, agent frameworks, MCP-exposed integrations) earn their place in a given engineering context. Pragmatic, vendor-neutral. Output is a recommendation with reasoning.
  • Engineering culture change: the human side of the transition. Addressing engineer anxiety, framing the change as opportunity, coaching engineering managers through the shift, retaining and growing the team rather than reducing it.
  • Metrics and risk design: what to measure to prove the new model is working. Cycle time, throughput, architect-versus-implement ratio. Premortems for tooling dependency, skill atrophy, governance under regulated contexts.

Stack

Salesforce · HubSpot · n8n · Azure OpenAI · Apex · Lightning Web Components · Spec-driven delivery · Agent-first development · Engineering culture change · Tooling evaluation

§ 02 / 04

Technology forensics & investigation

Independent technical investigation of suspected unauthorised access, data exfiltration risk, and integration-layer exposure. Deliverables written for two audiences simultaneously: commercial (for decision-making) and legal (for defensibility).

Recent work

  • Salesforce unauthorised-access investigation: classified a suspected breach as an existing authorised OAuth integration path, assessed exfiltration risk, and produced recommendations suitable for legal and commercial use.

Stack

Salesforce audit · OAuth & Connected App review · API access pattern analysis · Event Monitoring · Log retention forensics

§ 03 / 04

Integration & data architecture

Cross-system workflow, bidirectional binary and attachment sync, API design, and version-controlled automation. Workflows that survive real-world failure modes: missing fields, credential changes, rate limits, silent schema drift.

Recent work

  • Salesforce ↔ Jira bidirectional sync: comments, binary attachments, deduplication by JIRA attachment ID, error isolation, scheduled FeedItem polling with duplicate suppression.
  • Version-controlled n8n: CLI tooling for pulling, pushing, diffing, and activating workflows so 80+ workflows can live in git and be edited by humans or AI agents.

Stack

n8n · Jira · Salesforce SOQL · Azure integrations · Apex · Node.js CLI tooling

§ 04 / 04

Product development

Our own product work, currently PicoPouch. Strategy, architecture, build, iteration: the same arc we walk through for clients, run on a product we own. The discipline keeps our advice on client product work grounded in what actually happens: pricing decisions, release cadence, the long quiet middle between launch and traction, the things that do not appear in case studies.

Talk about an engagement →