Practical AI for Grant Writers
Give Your Attention Back to Your Mission.
You didn't join the non-profit sector to rekey boilerplate, decode compliance rubrics and format budgets. Bid Desk, our free and open-source app for Claude, does that admin: a fit check on your own criteria, funder rules with a source on every line, and a narrative, results framework and budget that match.
Set it up yourselves for free, or have us set it up with your funders' rules and your own content and train your team. Your time goes back to judgment, storytelling and impact.
Free and open source · Runs in your own Claude account
Bid Desk for Grants
Bid Desk is our free, open-source app for Claude, and the one we use in the workshop. It keeps every funding call on one board, shared with your whole team, and starts each one with a fit check: your own criteria, filled from the call with a source on every line. It pulls out the funder's rules with every point linked to its page in the call, and keeps your narrative, results framework and budget in step through one set of application facts.
3-minute walkthrough on a sample application, with subtitles.
10 min Go/No-go
Every call on one board by stage and deadline. Bid Desk fills your own fit check from the call and your content library, with a source for every answer and the points worth discussing, so you decide whether to apply before anyone writes a word.
Divide the Labour
For each section, a team member or the AI drafting from your own content library. A live counter keeps every section within the funder's character limit. Drafts follow your own voice, using past winning applications as style references.
A results framework that holds up
Outcomes, outputs and indicators with baselines, targets and means of verification. Bid Desk flags targets that are missing from your narrative.
A budget that follows the rules
Built from your own cost norms. Bid Desk does the sums and checks co-financing, in-kind caps, overhead rates, category caps and ineligible costs.
Everything matches
One set of application facts (grant requested, budget, co-financing, targets). Change a figure once and Bid Desk lists every section, target and budget line that still uses the old one. Every output has costs in the budget, and every target appears in the text.
The funder's own forms
Fills the funder's Word application form and Excel budget template, keeping their headings and formulas.
One application, one team
Every application is saved for your whole team, with a clear view of who still owes what: partner letters, budget figures, programme input. A chase message is one click away.
Learn from every decision
Record each decision and the reviewers' scores. Bid Desk shows your success rate and where applications lose points, and suggests what to change in your content library.
A fit check you can share
Download the filled fit check as an Excel file for the team meeting where you decide whether to apply.
See what Bid Desk produces:
Download a sample fit check (Excel)Filled by Bid Desk for a fictional applicant on a fictional funding call.
Who Is This For?
Bid Desk and our support are built for professional grant writers, development consultants and research coordinators who are ready to take back their time.
No one should spend their one wild and precious life manually processing grants. If you are exhausted by the mechanical friction of fundraising, hunting down old impact stats, aligning multi-year work packages with rigid spreadsheets, and wrestling with endless paperwork, this session shows you how to delegate that toil to a machine.
You will learn practical workflows for setting up a secure organisational knowledge base, running automated compliance audits, and letting AI handle the heavy drafting residue. That leaves your team free to focus on what machines can never replicate: strategic vision, deep donor relationships, and the core mission of your organisation.
It also fits grants and partnerships teams at NGOs and international NGOs who apply to institutional donors such as the EU, ECHO, FCDO and the Dutch Ministry of Foreign Affairs, including on consortium applications with several partners.
What You'll Learn
In the workshop: a 3-hour, hands-on session built around one real grant application, worked from eligibility check through to a submission-ready draft. You leave with a reusable profile and templates for your own organisation, not just a demo. Everyone who attends also takes home their own starter pack: the skill files and templates from the session, ready to reuse in Claude.
Stop Chasing Dead-End Proposals
The Pain: Investing days into a proposal only to realise halfway through, or worse after rejection, that you missed an eligibility rule buried on page 40 of the guidelines, or that your project never really aligned with what this funder wants.
The Fix: Instantly extract eligibility criteria, required sections, and evaluation rubrics from the RFP, then run a rigorous gap analysis and fit matrix against your own profile, so you know whether a funder is worth your time before you write a single word.
Kill the Blank-Page Paralysis
The Pain: Staring at a blinking cursor for hours, trying to remember how you phrased your organisational impact stats last time while rekeying boilerplate for the tenth time.
The Fix: Upload your past wins, theory of change, and backgrounders once. Let your system handle the structural heavy lifting so you never start from zero again.
Build the Narrative From Your Past Wins
The Pain: Manually rebuilding your narrative, work plan, and log frame from scratch for every funder: retyping objectives, indicators, and assumptions into a new template and hoping the timeline still adds up.
The Fix: Turn your project description into a reusable narrative and a populated log frame matrix, objectives, indicators, means of verification, and assumptions, plus a work plan and timeline, retailored to each funder's own template and terminology without losing your voice.
Build the Financial Model With Every Formula Intact
The Pain: Building the budget in a separate spreadsheet that never quite talks to the narrative, so cost justifications, cash flow, and cofinancing maths get reconciled by hand at the last minute.
The Fix: Feed in your work packages and the funder's eligible-cost rules to generate a full budget spreadsheet, personnel costs, overheads, cofinancing, with every formula intact and ready to adapt, plus the justification narrative to match.
Reconcile Everything
The Pain: Submitting a proposal with a knot in your stomach: praying the narrative, the log frame, and the budget still agree with each other, and that you haven't missed a logical gap, a weak claim, or a strict character limit that gets you automatically disqualified.
The Fix: Run an automated, strict reviewer simulation that cross-checks your narrative, log frame, and budget against each other and the funder's own rubric, flags weak claims, and trims your text to exact limits effortlessly, so nothing contradicts before it goes out the door.
Why This Is Different
The traditional grant writing process is sequential, manual, and built around the heavy lifting of drafting from scratch. Bringing deep-context AI into that process shifts the work from manual writing to strategic directing and rapid iteration.
The Old Way
Blank-page dread: Hours lost hunting down stats and manually rekeying boilerplate text.
The Practical AI Way
Instant assembly: Reusing structured knowledge bases to draft structural arcs in minutes.
The Old Way
Manual compliance panic: Cross-referencing line-by-line against dense, shifting guidelines.
The Practical AI Way
Automated auditing: Instant extraction of rubrics, limits, and eligibility checklists.
The Old Way
Siloed spreadsheet stress: Disconnected financial justifications built under deadline pressure.
The Practical AI Way
Unified modeling: Work packages and cost justifications generated side-by-side.
The Old Way
Late-stage surprises: Finding logical gaps close to the deadline with no time to fix them.
The Practical AI Way
Proactive stress-testing: Running automated reviewer simulations days before submission.
The Old Way
Generic “AI Slop” Fear: Writing lazy, unedited prompts that reviewers immediately spot and penalize.
The Practical AI Way
Ground-Truth Rigor: Using AI strictly as a mechanical engine for your own verified institutional data and past wins.
Beyond a Single Application
The application is only part of the work. With Consulting Support we extend the same workflow to the work around it.
Consortium applications
Partners send their inputs in one agreed format. We pull them into a single narrative, results framework and budget, with one version everyone works from and one review before it goes to the donor.
Reporting after the award
Narrative reports, indicator results and financial figures drafted from the same content library and checked against each other, so what you report matches what you promised and what you spent. Modification and no-cost extension requests start from the same record.
Knowledge that stays
Grants staff move on. Your past applications, reports, donor rules and lessons learned stay in your content library, so the next person starts from your last application.
Two Ways to Work With Us
Take Bid Desk free and set it up yourselves, or have us set it up and build your information repository with you. Everything runs in your own Claude account and stays with your organisation.
Free: set it up yourselves
Bid Desk for grants, free and open source on GitHub, ready to publish in your own Claude account.
- Bid Desk: eligibility, funder rules, AI or team drafting, results framework, budget checks and the funder's own forms
- A setup guide for grant writers and a sample fit check
- Open source under the MIT licence, free for any organisation
Best for teams comfortable setting up AI tools themselves.
Consulting Support
We set everything up with you and stay on hand while you use it.
- Setup: Bid Desk and the workflow installed in your own Claude account, with your funders' rules and templates.
- Your information repository: We turn your past applications, reports, budgets, theory of change and cost norms into a content library the AI can draft from.
- Training: We train your grants team on a live application, plus the programme staff and partners who feed into your applications and reports.
- Support: Help on your next applications and reports, a review before you submit, and updates when funders change their rules.
Remote, with an in-person start available in the Netherlands.
Tell us which funders you apply to, and we'll recommend where to start and send a fixed quote.
FAQ
General Workshop Questions
Will this replace our grant writers?
It handles the bulk of the drafting and checking work. The judgment on which funders to pursue, how to position a project, and when something needs a human read stays with your team.
Can our whole team work on one application?
Yes, and there's no shared Claude account. One person publishes your Bid Desk and shares it from Claude's Share menu, with edit access for the people who work on applications. Everyone then works in their own Claude account on the same board, library and applications. If two people edit the same application at the same moment, the last save wins, so agree who owns each section. Export a backup from the content library regularly, and before the person who owns it leaves, so you can move it to a new owner.
Is Bid Desk really free?
Yes. Bid Desk is open source under the MIT licence, free for any organisation to use and adapt. Download it from GitHub and publish it in your own Claude account. Each person who uses it needs their own Claude plan. If you'd like it set up with your funders' rules and your own content, that's our Consulting Support.
Do we need to use Claude specifically?
The techniques apply to most mainstream AI tools. It was built to work on Claude.
Will AI make this easy?
Grants still take real work, and they should. But you'll have a lot of help doing it, from Claude handling the heavy lifting to catching problems before a reviewer does.
Can AI get things wrong?
Yes, and every draft still needs a human check, especially for numbers and outcome claims. The quality-check step in the workshop is built specifically to catch mistakes before a funder does.
How does this workflow differ from standard online AI tutorials?
Most online tutorials cover generic prompts or basic prose polishing. This workflow focuses strictly on advanced, multi-document workflows: feeding entire RFPs into Claude, building a permanent organisational knowledge base, linking complex work packages to financial models, and running automated compliance audits.
Do we need an active paid Claude Pro account?
You can participate with a free account, though a Claude Pro account is recommended to make full use of features like Projects, Artifacts, and larger file upload limits during the exercises.
Technical & AI Capability Questions
How do we handle funder character and word limits using Claude?
Claude is strong at multi-constraint formatting. You will learn specific prompting techniques to trim drafts down to precise character or word counts without losing critical data, logical structure, or required grant terminology.
Can Claude reliably handle financial calculations without introducing errors?
In Bid Desk, the app does the arithmetic. Claude drafts budget lines from your own cost norms, and Bid Desk calculates the totals, co-financing and indirect costs, then checks them against the funder's rules. You still review every number before you submit.
What about donor data privacy and confidentiality?
Data privacy is a top priority when handling sensitive institutional information. We cover setting up private workspace environments, turning off data training permissions, and masking confidential details before feeding past bids or financial figures into the model.
Practical Application Questions
Will applications written with AI sound like generic, low-quality “slop” to reviewers?
Yes, if you use AI the lazy way, by asking a chatbot to write a grant from scratch based on a generic prompt. Grant reviewers are already penalizing obvious, unedited AI output. This workshop teaches the exact opposite approach: building a permanent, secure organisational knowledge base so the tool acts strictly as a mechanical assembler of your own verified impact data, past wins, and core voice. You keep complete editorial control, ensuring it reads authentically and respects the reviewer's time.
Can these workflows be adapted to other AI tools like ChatGPT or Gemini?
Yes. We use Claude live because its context window and document handling suit long-form grant writing well, and the core strategy, knowledge base structure, gap analysis, and budget alignment, applies directly to other mainstream AI tools.
How much time does this workflow actually save on a typical grant application?
The biggest time savings tend to come from initial document synthesis, boilerplate tailoring, and compliance checking. Automating that heavy lifting can cut manual assembly time by roughly 50 to 60 percent, freeing up time for strategic positioning and partner alignment.