British Healthcare Trades Association (BHTA)

Overview

We deployed 44 AI tools across all seven BHTA departments: 25 standard assistants covering operations, marketing, sales and customer service, plus 19 custom assistants built around specific pain points. The 14-person team now saves an estimated 2,500+ hours per year.

Marketing data consolidation dropped from six hours to minutes. Grant research runs automatically. The CEO has a structured weekly briefing for the first time.

Manual processes consuming hours

Trade associations typically operate with small teams covering broad responsibilities. Each team member manages multiple roles across compliance, marketing and governance. BHTA runs three brands and manages regulated processes, creating bottlenecks around individual expertise.

BHTA's 14-person team faced five major bottlenecks. Marketing staff consolidated data from five platforms across three brands monthly, consuming six hours. The research team updated government grant listings manually through three sources. Membership onboarding required 21 due diligence steps. The CEO reviewed 21 days of email and 14 days of calendar manually. Complaint data arrived unstructured and required hand-mapping into 23 columns.

Standard tools first, custom builds next

We started by deploying 25 standard assistants personalised with BHTA's core business documents. These covered everyday tasks: drafting content, analysing data, preparing meetings, reviewing documents, handling customer queries. Every staff member had a working AI toolkit from day one.

We then ran dedicated discovery sessions with each department. Staff described their workflows, tools and frustrations. We documented each challenge as a structured problem statement and built a custom assistant using the GRAFT framework (Goal, Role, Audience, Format, Tone). No new platforms were introduced. Staff paste data in, the assistant processes it, and the output drops into Excel, WordPress or Outlook.

AI now integral to every department

Within the first month, the Marketing Data Consolidator eliminated six hours of monthly manual work. The Complaints Organiser turned unstructured text into paste-ready Excel rows from day one. Standard assistants were already in daily use across every department.

By month four, all 44 tools were in production. The Grant Updater had refined through three specification versions. BHTA renewed for six months and added two more users. Combined productivity reached an estimated 2,500 to 3,000 hours per year across standard and custom assistants.

BHTA launched an AI Champion programme. Staff now identify new use cases independently. Five additional challenges were documented and built without our involvement.

From assistants to connected systems

The natural next step is connecting BHTA's assistants directly to the systems they draw from. Three areas are taking shape.

The Grant Updater is moving toward a direct connection to government data sources, publishing to the BHTA website without manual intervention. The marketing data tool is being lined up to pull performance data straight from Google Analytics instead of requiring a monthly paste-and-consolidate step.

BHTA's bespoke CRM is on the path to connecting with the AI workspace via MCP (Model Context Protocol), giving assistants live access to membership data for onboarding checks, renewal tracking and engagement analysis.

Each of these removes the remaining manual handoff between an AI assistant and the system it serves. The membership team will shift from data chasing to retention and engagement work.

What we learned

The standard toolkit gave every staff member immediate value while we ran discovery. By the time custom assistants arrived, the team was already comfortable working with AI daily. Department-by-department discovery ensured every assistant matched real workflows. The Marketing Data Consolidator was the first deployed and proved the model immediately.

The Grant Updater needed three specification versions. Government data sources use inconsistent formats, and edge cases kept surfacing. Each iteration improved accuracy, but patience from both teams was essential.

Organisations with small teams covering broad responsibilities benefit most from this approach. No two problems were alike at BHTA, and a generic rollout would have missed the highest-value use cases.