Working with RISR's AI Team

Your guide to Maya, Sam, Quinn, Charlie, Hugo, and Archie — the AI agents who work alongside us.

🚀 Quick Start

Six AI agents, each with their own specialty and Slack presence.

🦕
Maya
Marketing
#marketing-ai Works with Alex

Shared Tools (All Agents)

Notion HubSpot Fireflies PostHog Google Workspace Web Search

Good For

  • Content drafts & blog posts
  • Campaign planning
  • Brand voice questions
  • Marketing research
🏌️
Sam
Sales
#sales-ai Works with Dan, Jason

Shared Tools (All Agents)

Notion HubSpot Fireflies PostHog Google Workspace Web Search

Good For

  • Prospect research
  • Pipeline reports
  • Deal prep & call prep
  • Competitor analysis
🧙
Quinn
Product
#product-mgmt-ai Works with Alex

Shared Tools (All Agents)

Notion HubSpot Fireflies PostHog Google Workspace Web Search

Good For

  • PRDs & feature specs
  • Product metrics
  • User research synthesis
  • Roadmap planning
🧞‍♂️
Charlie
Customer Success
#customer-success-ai Works with Cory

Shared Tools (All Agents)

Notion HubSpot Fireflies PostHog Google Workspace Web Search

Good For

  • Customer health reports
  • QBR preparation
  • Onboarding help
  • Account history summaries
🎅
Hugo
Engineering
#eng-ai Works with Megan, Brandon

Shared Tools (All Agents)

Notion HubSpot Fireflies PostHog Google Workspace Web Search

Engineering Extras

Claude Code GitHub CLI Heroku CLI Sentry CLI AWS CLI Postgres

Good For

  • Bug investigation
  • Code questions
  • Deployment help
  • Database queries
🏗️
Archie
Infrastructure
#archie-ai (DM only) Works with Brandon

Shared Tools (All Agents)

Notion HubSpot Fireflies PostHog Google Workspace Web Search

Infrastructure Extras

System Config Agent Management Gateway Control

Good For

  • "What can the agents do?"
  • Technical issues
  • Questions about capabilities
  • Agent system explanations
How to reach them

@mention in their channel, DM them directly, or @mention wherever they're present.

💬 Just Ask

The agents are more capable than you might expect. Don't overthink it — if you're wondering "can they do X?" just try it.

Try prompts like these:

Need prospect research before my call with ABC Advisory tomorrow — pull their info from HubSpot and research their firm
Draft a LinkedIn post about AI data capture for financial advisors
What's our activation rate looking like this month? Check PostHog.
Acme Corp renewed but downsized — summarize their history from HubSpot and suggest a save strategy
There's a weird bug in the reports — can you check Sentry?

If they can't help, they'll say so. If they need more context, they'll ask. You won't break anything by asking.

🛠️ Tools & Capabilities

What each agent can access and do.

🔧 Shared Tools — Available to ALL Agents

Notion HubSpot Fireflies PostHog Google Workspace Web Search File Creation

Notion — Read/write pages & databases  •  HubSpot — CRM data, contacts, companies, deals  •  Fireflies — Meeting transcripts & summaries  •  PostHog — Product analytics, events, feature flags  •  Google Workspace — Gmail, Calendar, Drive, Docs, Sheets  •  Web Search — Brave API for current info  •  File Creation — Save deliverables for team access

📧 Agent Email Security

Each agent has their own @risr.com Google account. For security, they can only send and receive emails to/from @risr.com addresses — no external emails. They can also view (but not edit) the team calendar.

Agent Additional Tools Best Used For
Maya 🦕 Shared tools only Content drafts, campaign planning, brand questions
Sam 🏌️ Shared tools only Prospect research, pipeline reports, deal prep
Quinn 🧙 Shared tools only PRDs, feature specs, metrics, user research
Charlie 🧞‍♂️ Shared tools only Customer health, QBR prep, onboarding help
Hugo 🎅 Claude Code, GitHub, Heroku, Sentry, AWS, Postgres Bug hunts, code questions, deployment help
Archie 🏗️ System Config, Agent Management "What can agents do?", technical issues

📁 Files & Images

What agents can receive and create.

📂 Access All Agent Files

Each agent has their own folder. When they save something, it's there instantly.

Access Files →

📥 They Can Receive

  • Images dropped in Slack (they'll see and analyze them)
  • PDFs and docs uploaded to chat
  • Links to Google Drive files (if they have access)
  • Links to Notion pages

📤 They Can Create

  • Markdown docs and reports
  • Spreadsheet-style data exports
  • Save anything to their files/ folder
  • Research compilations

🧠 Memory

Agents don't just forget everything between conversations. This is one of the most powerful features.

The Short Version

Agents keep notes. When you tell Maya something important about brand voice, she writes it down. Next time she drafts copy, she remembers.

How It Works

Agent Memory

MEMORY.md — Their long-term memory. Key decisions, preferences, important context.

memory/YYYY-MM-DD.md — Daily notes. What happened, what they learned.

Shared Memory

risr-shared/ — Company-wide context ALL agents can read. Company decisions, customer info, product updates, process docs.

Add to Their Memory

Tell them directly — they'll write it down:

Remember that we're focusing on enterprise deals this quarter
Add to your notes: our ICP has shifted to larger RIAs
Update the product roadmap doc with this decision

Share Across All Agents

Need ALL agents to know something? Tell any agent to update the shared context:

All RISR agents should know: we've rebranded the Exit Planning module to "Succession Planning"
Update the shared context: Acme Corp is now our largest customer and they have a dedicated CSM
This is company-wide info — add to risr-shared: Q2 focus is renewals over new logos

The agent will update risr-shared/ so every other agent sees it next time they work.

✨ Getting Better Results

Small changes in how you ask can make a big difference.

Let Them Use Their Tools

Instead of giving agents every detail, ask them to look things up. They have powerful tools — let them be agentic!

❌ Instead of
"Here's everything about the prospect: company is XYZ, revenue is $5M, they're in Ohio..."
✅ Try
"Pull the info on ABC Advisory from HubSpot and research their firm before my call tomorrow"
❌ Instead of
"Write something about our product"
✅ Try
"Draft a 200-word LinkedIn post about AI data capture, targeting financial advisors who serve business owners"
❌ Instead of
"Help with this customer"
✅ Try
"Acme Corp renewed but downsized. Pull their history from HubSpot and suggest a save strategy."

Pro Tips for Power Users

🎯 Give Context Up Front

I'm drafting a proposal for a 50-advisor firm. They currently use Orion and are frustrated with reporting. Help me emphasize our reporting strengths.

⚡ Ask for Sub-Agents for Parallel Work

Spawn sub-agents to research each of these 5 competitors in parallel, then compile the findings

📝 Ask for Drafts, Then Iterate

Give me a rough outline first, then we'll refine together

🔄 Ask Them to Self-Review

Draft this email, then review it yourself before showing me — fix any issues you spot

✅ Ask Them to Double-Check

Double-check those numbers before you finalize

⚡ Sub-Agents

For big tasks, agents can spawn helpers to work in parallel.

How It Works

When you give an agent a big task, they can spawn "sub-agents" to work simultaneously.

You don't need to ask for this — agents do it automatically when it makes sense. But you can be explicit:

Spawn sub-agents to research each of these in parallel

EXAMPLE

"Research these 5 competitors"

Behind the scenes:

  • Agent spawns 5 sub-agents
  • Each researches one competitor
  • Results compiled and returned

🆘 Escalation

When agents point you to someone else.

Sometimes an agent will say:

"I can't change that setting — you'll need to ask Archie or Brandon"

This happens for config changes, new tool access, or capability requests.

Tag @Brandon Directly

He manages agent infrastructure and can make changes. Archie can answer questions about what's possible, but Brandon approves changes.

⚠️ Limitations

The real constraints and workarounds.

Why: Safety — we want human review on external communications.

Workaround: Ask them to draft, you review and send (or they send after you approve).

Why: Privacy — agents have their own accounts.

Workaround: Forward or paste relevant threads to them.

Why: No payment access.

Workaround: They can draft the request for you.

Why: Privacy between sessions.

Workaround: Ask them to check with another agent or share context yourself.

Why: They're not psychic.

Workaround: Share context, ask them to research.

⚡ They Can Make Mistakes

Review anything important, especially numbers, customer names, and external-facing content.

Architecture

Slack → Gateway Server → Agent
                           ↓
                      Workspace
                     (files, memory)
                           ↓
                       Tools
                  (APIs, search, etc.)

Each agent has:

  • Slack account — Their identity in Slack
  • Workspace folder — Their files, memory, instructions
  • Tool access — What systems they can use
  • Session — Current conversation context

Sessions

A "session" is a conversation thread. Within a session, the agent remembers everything said. Between sessions, they rely on memory files.

Sessions can get long. When they do, older context gets summarized to make room for new messages.

The Gateway

All messages route through a gateway server that:

  • Routes messages to the right agent
  • Manages tool permissions
  • Handles session state
  • Runs on our infrastructure (not a third-party service)

Memory Files

agent-workspace/
├── SOUL.md          # Personality, role definition
├── AGENTS.md        # Operating instructions
├── MEMORY.md        # Long-term memory
├── memory/
│   ├── 2026-01-30.md
│   └── 2026-01-29.md
├── files/           # Deliverables
└── tools/           # Tool-specific notes

Shared Context

risr-shared/
├── company/
│   ├── biz-summary.md
│   ├── team.md
│   └── team-profiles/
├── product/
├── customers/
└── docs/

All agents read from risr-shared/ but write to their own workspace.

📋 Quick Reference

Agent Channels
  • #marketing-ai (Maya)
  • #sales-ai (Sam)
  • #product-mgmt-ai (Quinn)
  • #customer-success-ai (Charlie)
  • #eng-ai (Hugo)
Need Help?
Drop questions, feedback, or links in #ai-agent-feedback