AI agents for research workflows

Funding application

Take over a research team’s project archive. Build a source-linked knowledge base, match a funding call and draft an application for review.

Download this folderChoose another scenario

All data and documents are synthetic. Experiment freely; restore the folder from the ZIP whenever you need to.

Get ready

Open the folder in Codex

Click a step to see a screenshot.

Screenshots show the earlier practice guide and ChatGPT Work interface. Use your separate Codex setup instructions for current controls.

What’s in this folder?

Working with your own data? Keep good, up-to-date backups of all your data and know how to restore them.

Scenario · funding application

Turn a project archive into a new proposal

You’ve joined a research team. Its folder mixes past applications, project updates and a new funding opportunity. Make the evidence reusable, then develop an application for review.

1. Discover the folder

What’s in this folder? Identify the documents, their dates and status, and how they relate. Leave the reveal folder until I ask for it.

2. Build a knowledge base

Build a knowledge base from this folder, with an index and source-linked notes on projects, delivered outputs, team capabilities, partnerships and lessons from past applications. Separate evidence, proposals and unknowns. Flag conflicting versions and cite file paths and headings for every substantive claim. Show your structure first.

3. Match the funding call

Identify the current funding call. Extract its eligibility, budget rules, assessment criteria, required sections and word limits into a checklist with sources. Compare it with the knowledge base. Suggest two feasible application directions, explain the fit and missing evidence, and let me choose.

4. Draft the application

Using the direction we chose, draft the application in the supplied template. Cite supporting sources, stay within each section limit, calculate an itemised budget and make a schedule. Mark unconfirmed commitments, evidence gaps and approvals clearly. Do not turn proposed or previously unfunded work into achievements, invent results or charge again for already-funded work.

5. Review and hand over

Review the draft against every call requirement and the original sources. Check arithmetic, dates, word counts and unsupported claims. Create a compliance table, an outstanding-questions list and a source map. Document how to update the knowledge base and regenerate the application, recording every change.
Check your understanding

After your own review, open funding-application-folder/reveal/notes-funding-scenario-checks.md. Compare the source status, costs and gaps. Several proposal directions can work; traceable evidence matters more than matching a model answer.

This is a fictional call and a draft for review. All people, projects and funding terms are invented. Finish with Document the workflow.

Step 05 Continue afterwards

Document the workflow

For any scenario, leave a workflow someone else can understand and use.

Document our steps and set up a repeatable workflow that tracks every change. Include a run guide, an append-only change log and verification checks. Show me the plan first.
Run guideInputs → commands → outputs → checks
Change logWhat changed, why, when and how
Repeatable workflowRecord each run and revision
What should the log capture?

Old/new paths, input/output hashes, commands, environment, decisions and checks. Include failed attempts and corrections. Append entries; do not overwrite earlier history. Log manual changes too, and compare hashes to catch omissions.

Check your result

Start a new conversation and ask the agent to use the run guide. Ask the agent to repeat a step from the guide. Check the result and confirm the new log entry was added without erasing the history.