Global Nature Watch
World Resources Institute · Development Seed
People ask questions about forests, farms, and other land changes. An agent selects datasets, runs geospatial analyses, and returns maps and charts grounded in satellite data.
AI project patterns
Find a starting point for your mission.
A pattern is a repeatable approach to building a successful project.
These six patterns and their case studies are a selection from our growing research into other organizations’ AI projects.
We draw on that broader expertise to find a pattern that fits, then plan a project around your goals.
Pattern #5
Explore data through conversation
Let people ask questions in their own words. Fits when you have useful data, but people need specialist help to get answers from it.
Pattern #4
Guide people with personalized advice
Turn specialist knowledge into practical next steps. Fits when people need advice tailored to their situation, and your experts can’t help everyone individually.
Caddy
Citizens Advice
Citizens Advice’s Caddy researches trusted guidance for frontline advisors. A supervisor checks each draft before it is used with a client. Doncaster reports less time searching for information and more time focused on clients.
Pattern #10
Help people complete complex applications
Help people get through complicated processes. Fits when complicated forms and repeated data entry keep people from accessing support they qualify for.
Form-Filling Assistant
Nava Public Benefit Corporation
In a California pilot, caseworkers used Nava’s agent to prepare benefit applications across multiple programs. It gathered existing information and filled forms; caseworkers checked and submitted them.
Pattern #8
Gather and understand community input
Give more people room to be heard. Fits when you need many people’s experiences to inform a decision, but gathering and reviewing their input overwhelms your team.
Sensemaking AI
Jigsaw · City of Bowling Green
Jigsaw’s Sensemaking AI helped Bowling Green hear from 8,000 residents. Google reports that their input informed the city’s 25-year growth plan.
Pattern #6
Gather and maintain shared information
Bring scattered facts into a resource people can use. Fits when people need a reliable reference, but the information is scattered across sources that keep changing.
Who belongs to parliamentary groups?
mySociety
mySociety uses agents to find parliamentary groups’ websites and extract membership lists. It checks the results and repeats the process when a new register appears.
Pattern #7
Prepare specialized reports and studies
Work through studies that take specialist time and several tools. Fits when specialist reports or studies hold up decisions, and your experts spend too much time preparing the material.
Agentic grid planning
Duke Energy · Amazon Web Services
Duke Energy’s engineers use agents developed with AWS to prepare grid studies and coordinate work across simulation tools. Engineers direct the work and make the final decisions.
From pattern to project
What fits your mission?
You don’t need to choose before we talk. Bring your goals and what gets in the way. We’ll explore the fit together and shape a project worth building.
Apply for an invitation