Overview
Iradaty’s team does not use Notion as their daily workspace. That constraint shaped the system.
Instead of trying to force adoption of a new internal tool, I built Notion as the backend layer behind the organization’s existing habits: daily field updates, project knowledge, funding opportunities, proposal drafts, team structure, and reporting needs. The result is an AI-native workspace that Iradaty benefits from without needing the whole team to operate inside Notion directly.
The workspace serves two connected purposes. First, it is the knowledge and reporting backend for Iradaty Bot: daily updates from the field are stored, structured, and turned into weekly reports, monthly reports, and project profiles. Second, it is the business development workspace I use as Iradaty’s business development specialist: funding opportunities are captured, assessed, and turned into proposals and reports using the organization’s own context.
This is the important separation: Iradaty Bot is the reporting interface and workflow; the Notion workspace is the institutional backend that makes the reporting, proposal development, and knowledge reuse possible.
The challenge
Iradaty works in a high-pressure humanitarian context where information changes quickly and lives close to the field. Project managers, field workers, media staff, and leadership all hold pieces of the truth: daily activities, urgent needs, beneficiary stories, blockers, team capacity, donor requirements, and project history.
Before the workspace, that knowledge was difficult to reuse consistently. If someone needed to write a proposal, produce a report, prepare social media content, or understand a project’s current status, the easiest path was often to ask people directly. That meant calling project managers and field workers repeatedly to reconstruct information that already existed somewhere.
The problem was not lack of knowledge. The problem was that the knowledge was not organized as a live, queryable system.
For proposal development, the same issue appeared in another form. A good go/no-go decision depends on more than reading a call for proposals. It requires knowing Iradaty’s real work, previous proposals, project profiles, team structure, daily field updates, available evidence, and institutional tone. Without a connected backend, every new opportunity risks becoming a fresh scramble.
What I built
I built the Notion workspace as an AI-native backend around two main flows.
1. Field intelligence and reporting
Daily reports from the field flow from the chatbot into structured databases. Those updates can then be used to generate:
- Weekly reports.
- Monthly reports.
- Project profiles.
- Management summaries.
- Evidence for proposals and donor reports.
- Context for social media and communications.
The key design principle is simple: field workers should not need to learn a complex reporting platform before the organization can benefit from structured knowledge. The system starts from the reporting behavior that already exists, then uses Notion as the layer where updates become searchable, reusable institutional memory.
2. Business development and proposal work
I also use the same workspace to manage Iradaty’s business development pipeline. Funding opportunities are captured and assessed in context, not as isolated links or PDFs.
The workspace supports:
- Intake of funding opportunities.
- Fit assessment and go/no-go decisions.
- Donor and opportunity tracking.
- Proposal planning.
- Proposal drafting.
- Report drafting.
- Reuse of previous proposals and project profiles.
- Alignment with Iradaty’s tone, team structure, and current field updates.
Previous proposals and project profiles are not treated as dead files. They become reference material for future fit assessment and proposal generation. A new proposal can start from Iradaty’s actual history, current activities, team capacity, and language patterns instead of starting from a blank page.
How the system connects
The workspace sits between several layers:
- Iradaty Bot: the reporting workflow that collects and structures updates.
- Postgres on Railway: the bot’s operational database.
- Notion: the curated institutional workspace where reports, project profiles, opportunities, and knowledge records become usable.
- NexDonor: the broader infrastructure I am developing for funding intelligence, proposal workflows, and nonprofit knowledge systems.
Notion is not the only database in the architecture. Its role is different: it is the human-readable, AI-curated workspace where records can be reviewed, connected, enriched, and reused. It gives the system a place where institutional memory is not just stored, but organized into forms that support decisions and writing.
Why Notion mattered
Notion is useful here because it can hold different kinds of organizational knowledge in one connected environment:
- Daily reports.
- AI-generated summaries.
- Project profiles.
- Previous proposals.
- Team and role information.
- Funding opportunities.
- Donor notes.
- Reports and public-facing content.
- Reference knowledge and reusable language.
The value is not simply that these records sit in one tool. The value is that they can be connected, filtered, summarized, and used by AI agents.
That matters for business development. A funding opportunity can be assessed against the organization’s actual profile. A proposal can draw from previous language and current field updates. A report can pull from daily activity across teams and time periods. Social media content can be grounded in recent work instead of vague institutional claims.
Custom agents and curation
The workspace is curated by custom agents and connected to the NexDonor infrastructure. That means the system is not just a static database waiting for someone to maintain it manually.
I am also experimenting with custom agents that maintain the knowledge base itself: cleaning incoming records, keeping project profiles current, organizing daily updates, and making the workspace easier to query over time.
A second layer of experimentation focuses on funding intelligence. Two custom agents work together around Iradaty’s donor ecosystem: one constantly researches direct and indirect signals to find unconventional funding and collaboration opportunities; the other takes those candidates and enriches them with research that enables a more confident fit assessment. The goal is not to flood the team with links, but to surface a short list of understandable opportunities with enough context to support a real go/no-go discussion.
Agents help structure and reuse the knowledge in the workspace: turning raw updates into usable summaries, supporting opportunity assessment, and making proposal work faster and more consistent. The human role remains important: I still review, decide, and shape the final outputs. The agents reduce the cost of retrieval and drafting; they do not replace judgment.
This distinction matters. The system is AI-native, but not AI-autopilot. It is designed so that human decisions are better informed, faster, and less dependent on chasing scattered information.
Impact
The biggest practical change is that information no longer has to be begged for every time someone needs to produce something.
When developing reports, proposals, or social media content, the team does not need to repeatedly call project managers and field workers to reconstruct what happened. The knowledge base already contains daily updates, project context, previous proposals, team structure, and relevant institutional language. The chatbot and workspace can retrieve high-coverage context across teams, projects, and time.
That changes the work in several ways:
- Reports can be drafted from structured daily updates instead of memory.
- Proposals can be grounded in previous proposals, current field activity, and real team capacity.
- Fit assessment becomes more consistent because opportunities are compared against known organizational context.
- Social media and donor-facing content can draw from actual field updates.
- Project profiles become living records instead of static documents.
- Institutional memory becomes available to the people writing and deciding, even if the field team never opens Notion.
The honest result is not “the organization adopted Notion.” The result is stronger: Iradaty gained an AI-native knowledge backend without needing the whole team to change tools.
What this case proves
This case shows a different kind of Notion implementation. The goal was not to make everyone use Notion. The goal was to build a backend where organizational memory, daily updates, funding intelligence, and proposal development could meet.
For Iradaty, Notion became the layer where field reality turns into usable institutional knowledge: reports, project profiles, opportunity assessments, proposals, and content. It connects the chatbot, the organization’s daily work, and the broader NexDonor infrastructure into one operating memory.
That is the deeper value of the workspace: it lets people ask the system before interrupting the team, and it lets new writing start from what the organization already knows.