APPLIED AI / ARTS & NONPROFITS

WE SHOULD BUILD IT.

The question your systems can’t answer is a good place to start.

Explore the work

For the people behind the mission.

A DIFFERENT CALCULATION

Experience meets
possibility.

Michael Barker
Principal & CEO, Pursued by a Bear

More about Michael

For most of my career, organizations have had to work around the limits of the systems they could afford. If the accounting system, ticketing platform, donor database, or spreadsheet could not answer the question, the solution was another export, another manual process—or accepting that custom software simply was not practical.

That calculation has changed. Large language models make it easier to create software around an organization’s actual needs. The difficult parts still require judgment: understanding the operation, knowing what the data means, designing the workflow, and recognizing when the output is wrong.

I bring more than twenty years of experience leading mission-driven institutions, along with an MBA from Yale School of Management, to that work. I combine management and fundraising experience with AI-assisted development to build practical tools for arts and nonprofit teams.

THE WORK

Real questions.
Working tools.

Five projects, from executive decisions to everyday coordination. The public demonstrations use synthetic data.

Executive Snapshot

An executive view that doesn't exist in any one system

Bring the numbers and the institutional knowledge into the same conversation.

The problem

At a major regional nonprofit theatre, the financial picture lives across accounting, ticketing, fundraising, spreadsheets, and people's inboxes. There are too many emails for an executive to digest, yet a colleague's note may contain the crucial detail about a payment or commitment. Separate reports cannot explain the whole position.

What I built

I built an executive dashboard that connects available cash, sales, fundraising progress, and the issues requiring attention. It shows the commitments behind the cash position, distinguishes reported actuals from projections, and keeps dates and assumptions visible so the reader can understand what each figure means.

Why it matters

Leaders can understand the current position, see what needs attention, and ask better questions without rebuilding the picture from separate reports and email threads before every conversation.

Where I add value

I bring the judgment of someone who has led these organizations: knowing which numbers belong together, which dates and definitions matter, and when a reassuring total hides a practical problem. The work begins with understanding the organization and testing the interpretation.

Prospect Research

Turn scattered records into a useful fundraising conversation

Help a development team decide whom to engage and what it still needs to learn.

The problem

A donor record rarely tells the whole story. Giving history, attendance, past conversations, board relationships, and research may sit in different places. Preparing for meaningful outreach becomes another round of manual assembly. A ranked list is useful only when a fundraiser can understand the evidence behind it and recognize what is still uncertain.

What I built

I built a searchable prospect workspace with individual profiles, giving and attendance filters, relationship context, and qualification flags. Staff can explore a prospect pool, compare different patterns of engagement, and review the evidence and unanswered questions behind a potential next step. The real project worked with more than 3,000 prospect records; the public demonstration uses 250 fictional profiles.

Why it matters

Research becomes something a fundraiser can use in a conversation and a team can review together. Gaps remain visible instead of disappearing behind a score, and staff can focus their preparation on the relationships they choose to pursue.

Where I add value

My fundraising and executive experience shapes the questions, the interpretation, and the next action. I know that useful prospect development depends on relationships, institutional knowledge, and professional judgment. The tool organizes the information so those skills have more room to work.

Capital Planning

Make the tradeoffs visible before the budget is final

Connect the project list to the years, priorities, and resources needed to deliver it.

The problem

A capital plan has to reconcile urgent repairs, long-term investments, incomplete estimates, and limited funding. Departments may have different priorities and different levels of information. A spreadsheet can hold the details while making it difficult to see what changes when a project moves, its cost changes, or another need takes precedence.

What I built

I built a planning workspace where projects can be edited, prioritized, and scheduled across fiscal years. Annual worksheets and priority views present the same plan for different discussions. Department spending charts show the distribution over time, and Excel and PDF exports support sharing outside the application.

Why it matters

Staff and trustees can examine the consequences of their choices together, identify missing planning details, and carry a clear plan into the budget conversation. The same information can support detailed staff work and a focused discussion about priorities.

Where I add value

I understand the operational and governance questions behind a capital budget. I translate those questions into a tool that fits the organization's planning process, while keeping the assumptions available for review. Deciding what to fund still requires experience, discussion, and judgment.

Volunteer Scheduling

Make a small, essential operation easier to run

Give volunteers a clear next step and coordinators fewer loose ends.

The problem

Think of the food pantry in my hometown and my son's former elementary school: small community organizations depend on people showing up when they are needed. A recurring pickup requires someone to choose a date, understand the instructions, remember the commitment, and communicate when plans change. Those ordinary details deserve thoughtful tools.

What I built

I built a volunteer scheduling workflow for choosing, confirming, and releasing pickup dates, with reminder preferences and a calendar connection preview. Coverage rules guide the available choices. The demonstration lets visitors try a personal signup experience, see their commitments, and make changes using fictional names and dates.

Why it matters

The same approach applies to food pantries, schools, community organizations, and other small teams coordinating recurring work. A focused tool can make participation clearer, help people manage their commitments, and reduce the number of details a coordinator has to chase.

Where I add value

I start with the people doing the work and the exceptions they encounter. My contribution is to understand the operation, simplify the choices, and design a process that volunteers can follow and a small organization can sustain.

Production Budgeter GPT

Turn production assumptions into a budget you can question

Theater Production Budgeter: a custom GPT for research-backed estimates and editable Excel workbooks.

The problem

A production estimate is a starting point for a conversation about the season: what the team is proposing, what it might cost, and which assumptions need a closer look. An estimate needs people who can test it against the actual production and the organization’s circumstances.

What I built

I built the Theater Production Budgeter GPT to help regional-theatre teams develop production-budget estimates and begin season-planning conversations. It is designed to help prepare research-backed estimates in formula-driven Excel workbooks. Starting questions cover an upcoming show, union labor costs, missing or risky costs, and low, base, and high scenarios.

Why it matters

A shared starting point gives the team something concrete to question and revise. Estimates stay subject to human review.

Where I add value

The management work is to challenge the estimate, resolve uncertain assumptions, and decide what belongs in the season plan. A useful conversation still depends on the people responsible for making and delivering the production.

THE APPROACH

The technology changes.
The judgment matters.

The AI part

AI helps me build and revise software. A tool built with AI may use ordinary code for calculations, filters, and scheduling. A tool such as the Production Budgeter GPT uses a language model directly. Each project needs a clear account of what it does and where people must review the result.

The human part

The work begins with the organization: its people, its constraints, and the decisions it needs to make. An email about a payment can matter as much as a financial statement. Knowing how the evidence fits together is management work, and it continues through implementation.

OPEN SOURCE / TESSITURA

A useful method.
Available to everyone.

Tessitura Skill for AI Agents

A free, unofficial skill for reporting, lists, analytics, and reconciliation. Windows-native guidance comes first, with a reusable path for Citrix and remote users. Explore the public core and organization-specific customization.

Also from Pursued by a Bear: Costume Helper, a separate project for costume work.

CONSULTING / AI & OPERATIONS

What should we build for you?

Bring me the question your systems cannot answer, the workflow that takes too much time, or the operational challenge that keeps returning. We’ll work out what is worth changing and where to begin.

Let’s talk