AI at CommunityForce
Run your awards program with an AI that already knows how.
Ask how many applications are still unassigned and get the real number back, from your live data, right now. Have your Monday status report land in your inbox without anyone building it. Open an application and find it already summarized for your reviewers.
That’s Phoenix. Built for awards programs, not adapted from a chatbot, it answers when you ask and works while you don’t. It uses your data for your team’s work, and nothing else.
Illustration of a Phoenix Copilot conversation.
Here’s your briefing for today. I checked the whole organization, every queue.
Stalled applicants
- 793 applications started but not submitted. 1,129 of 1,922 started are in. Review applications
- 118 submitted applications have no reviewer assigned (186 awaiting review in total). Assign reviewers
Forms left active past their close date
- The Future Leaders Fellowship closed 226 days ago but is still active. Fix in General Information
All clear
Forms closing soon in the next 7 days: none.
Ask Phoenix Copilot…
Phoenix Copilot can make mistakes. Verify important information.
Phoenix Copilot
Ask Phoenix a question. Or hand it the job.
Type the way you’d talk to a colleague who knows awards programs inside out. Which reviewers are behind? How does this year’s pool compare with last? Who hasn’t finished their application? Phoenix Copilot goes and looks, in your live data, every record, not a sample, and comes back with the answer and a report you can keep. Then you keep asking, and it keeps up.
It doesn’t stop at answering. Ask it to build the report you’ve been meaning to make, fix a form across every question at once, or draft the email you keep rewriting, and it shows you exactly what will happen before it happens. Nothing lands until you say so.
It knows where you are. Open a form, and it’s about that form. Open an application, and it’s about that applicant. And it only ever sees what your role can see, so the answers are yours, and only yours.
It knows the product, too. Ask how to do something and it answers from the documentation and shows you the article it used, so nobody has to leave what they’re doing to go hunting.
Try it with the question every administrator has by 9 AM: “What needs my attention today?”
See how reporting and Copilot work
Phoenix Agents · Now rolling out
Phoenix doesn’t just answer anymore. It shows up for work.
Until now, an awards platform waited to be told what to do. That changes here. Phoenix Agents do the routine work of running an awards program on their own, and report back to you.
Some work while you sleep.
Tell Phoenix what you need, in plain words: “Send me the Applicant Summary for our Spring Scholarship every Monday morning.” It shows you what it’s about to set up, you say yes, and from then on it just arrives, file attached, every week, without anyone remembering to run it. Your Daily Digest Agent greets each administrator every morning with the five things that actually need them.
Some work beside you.
Inside the screens your reviewers already use, three agents take the grind out of reviewing: the Summary Agent reads a whole application and hands you the person inside it; the Anti-AI Agent flags essays that may not be the applicant’s own; the Reviewer Agent drafts a first pass at the scoresheet from what the applicant actually wrote. Nothing runs until someone asks.
And you can see all of them, on one page.
Every agent your organization uses, overnight or on demand, lives on the Agents page. See what each one does, how often it’s used, and switch any of them off in a click.
Phoenix Agents only see what you can see, never contact an applicant, never decide an award, and keep everything on the record. Some tasks are just too important to be done by anyone but you.
Meet Phoenix AgentsIllustration of a Phoenix Copilot conversation about creating an agent.
I can help you set that up. A scheduled agent re-runs an instruction on a schedule and emails you the result. It’s read-only, so it can report on things but can’t change data or email applicants.
To propose one, I need a few details:
- What should it do? “Send me the applicant status breakdown,” “email me the forms closing this week”…
- How often, and at what time? “Every Monday at 8 AM.”
- Who should get the email? Defaults to you if you don’t say.
Ask Phoenix Copilot…
Phoenix Copilot can make mistakes. Verify important information.
On-demand Phoenix Agents
Help for the people doing the reviewing
Reviewing is where the hours go. Three agents work beside your reviewers, inside the screens they already use, only when asked, and never making the call. Each one gives a head start; the judgment stays human.
Summary Agent
Read 200 applications before lunch.
An application opens, forty answers scroll by, and you are hunting for the person inside them. The Summary Agent reads the whole thing first and gives the reviewer a plain-language picture of who this applicant is and what they are asking for. The full application is one click below, so nothing is skipped.
Powered by Phoenix
Anti-AI Agent
Know when the essay isn’t theirs.
Committees are already asking whether the essay was written by the applicant or by a chatbot. The Anti-AI Agent checks every essay and flags the ones worth a closer look, sentence by sentence. It is a data point for the committee. A person makes the call.
Powered by GPTZero
Reviewer Agent
The scoresheet, already half thought through.
Twenty scoring questions per application, dozens of applications in, is where consistency slips. The Reviewer Agent reads the application and suggests an answer to each question, with its reasoning shown. The reviewer accepts it, changes it, or ignores it. The score is always theirs.
Powered by Phoenix
And the arithmetic takes care of itself
The parts of a rubric that are just arithmetic, like GPA thresholds and eligibility fields, score themselves. Reviewers spend their attention where judgment is actually needed.
Built to be trusted with your data
Your team can lean on Phoenix because of what it is not allowed to do.
How we govern AIDesigned in alignment with the NIST AI Risk Management Framework and OWASP guidance for AI applications.
Never trained on your data
Phoenix reads your data to answer you. Your data is never used to train AI models, ours or anyone else's.
Your view, nothing more
Phoenix only sees what the person using it is allowed to see. Ask it something outside your role and it doesn’t know, because it can’t.
People decide
Awards, eligibility, and every decision that touches an applicant belong to your team. AI proposes and reports.
Off by default, on the record
AI is off for your organization until an administrator turns it on; from there every agent has its own switch, and everything is logged, so you can always answer who did what, and when.
Phoenix Copilot and Phoenix Agents can make mistakes. Verify important information.
See it working on programs like yours
Ask Copilot a question, watch an agent get created from one sentence, and judge it on your own terms.