AI agent training for charities, at 30% off

Charity teams are usually short of time rather than ideas, and AI is most useful on the work around the mission rather than the mission itself. Charity teams use it to research funders and draft grant applications, assemble board papers from the finance report and risk register, turn monitoring spreadsheets into impact reports, answer common supporter emails and chase overdue invoices. The biggest change comes when AI moves out of a chat window and becomes an agent connected to your files and tools, because it can then fetch the information itself instead of waiting for someone to paste it in. A person still checks everything that goes out. The time a small team gets back goes to the work only people can do: funders, supporters, volunteers and the people you serve.

A chatbot answers questions. You paste text in, it replies, and you carry the result to wherever it needs to go. An AI agent has three things a chatbot lacks: context about your organisation (your policies, past applications and templates), access to tools (a shared drive, a spreadsheet, your inbox or your accounts) and instructions to follow a process step by step. So you can ask an agent to read this quarter’s figures, compare them with last quarter and draft the finance section of the board pack, and it does the whole job rather than giving one reply. ChatGPT, Claude and Microsoft 365 Copilot can all work as agents once they are set up with the right context and connections. Setting them up that way is what the Agent Accelerator teaches.

AI can draft them, and it is very good at the research around them, but it shouldn’t write them alone. An agent can read a funder’s priorities, past awards and guidance, score how well your work fits, and produce a first draft in your voice from your previous successful bids and latest outcomes. Funders now receive many applications drafted with AI, so a competent but generic answer no longer stands out. The applications that win still carry what only your team knows: a named local example, a figure from your own monitoring data, an honest account of what you learned last year. Use AI to apply to fewer, better-fitting funders and to reach a strong draft faster, then spend the time you save on the substance. Check each funder’s guidance on AI use before you submit.

You don’t need a developer or an IT team. Start with one repeatable task that costs someone a few hours a week and does little harm if a draft is wrong, such as sorting the shared inbox, drafting thank-you notes or summarising volunteer feedback. Use a work account on a business or non-profit plan rather than a free personal one. Write down what data the AI may see and who reviews its output, then give it three or four good examples of the finished task. Run it alongside your current process for a week or two before you rely on it. The Agent Accelerator is built for non-technical teams: at DebateMate, a team member with no technical background built an invoice-chasing agent connected to Xero during the cohort.

Most charities start with the assistant that fits the software they already run. Teams on Microsoft 365 can begin with Copilot Chat, which Microsoft includes in its non-profit Business Basic and Business Premium plans, and add Microsoft 365 Copilot and Copilot Studio agents later. Teams that want agents working across documents, spreadsheets, the web and other tools often choose Claude, which Anthropic offers to eligible non-profits at up to 75% off its Team and Enterprise plans. Whichever assistant you pick, the agents connect to the tools you already have: Google Drive or SharePoint, Excel or Google Sheets, a CRM such as Salesforce or HubSpot, accounting software such as Xero or QuickBooks, and email. The skills of briefing an agent and checking its work carry across all of them.

It can be, with the right set-up. Never paste case notes, health information or anything that identifies a service user into a free consumer chatbot. Use a work account on a business or non-profit plan, check the provider’s terms on data retention and model training, and give each agent only the access it needs, starting read-only. Run a data protection impact assessment before any agent touches personal data, and follow the Information Commissioner’s Office guidance on AI and data protection. Keep a person reviewing anything that leaves your organisation. For decisions that affect a beneficiary directly, such as eligibility or safeguarding, the agent prepares the information and a person decides. On the course you build on your own accounts and choose what each agent can see.

No, and the charities that get the most from it don’t try to make it. They hand agents the admin that pulls staff away from people: assembling reports, chasing paperwork, answering the same supporter question for the tenth time. The work a charity exists for depends on human judgement and relationships, whether that is building trust with beneficiaries, supporting volunteers or keeping funders close. What changes is where staff time goes. A small team can take on work that used to need more hands, and the people on it spend more of their week on the parts of the job they came to the sector to do.

Start with a short written AI policy and take it to the board. It should say which tools are approved, what data may be used, who reviews what, and how staff report a mistake. Trustees usually ask about data protection, reputation and accuracy first, so show them the guardrails before the gains: work accounts, the least access each agent needs, human review and a data protection impact assessment for anything involving personal data. Then show one concrete pilot, the time it took before and after, and the review process around it. Trustees can use agents themselves too, for example to get a briefing on the three items in a long board pack that need a decision before the meeting.

Staff and trustees of a charity registered with the Charity Commission for England and Wales, the Scottish Charity Regulator (OSCR), the Charity Commission for Northern Ireland or the Charities Regulator in Ireland. Social enterprises, CICs and other non-profits are not covered by the rate, but email us and we will talk.

Every training format: seats on the live Agent Accelerator cohort, licences for the self-paced course, and team bootcamps in person or online. Bootcamps need at least 10 people. Apply, and James will book a call to plan the day and take 30% off the quote. The charity rate replaces any other discount rather than adding to it.

Yes. The course teaches how agents work: giving them context, connecting them to your tools and handing them real tasks. That carries across whichever assistant your charity uses. Most sessions are built in Claude, and Copilot’s agent tools have moved on quickly this autumn, so James runs Copilot-focused training for charities that live in Microsoft 365. Say so when you apply.

Usually one working day. James checks every application against the charity register himself.

The charity rate is paid by card for the live cohort and the self-paced course. Bootcamps are invoiced. If your charity needs an invoice or a purchase order for anything else, email james@agentaccelerator.ai and we will sort it out.