Internal Only

FRIDAY CALL

Jason Kirk Call — Team Reference and Script

Friday Call Script

45 MINUTES WITH JASON

Keep this open on a second screen. Follow the flow. Clear handoffs between speakers. Close on Phase 1.

Roxy Opens
5 minutes
Hero
Welcome Jason and Matt. Warm, personal. Brief intros of the Kelly team and their roles.
SITREP
"We took everything you shared seriously. The team's been heads down building something we think you'll find valuable." Set expectations, then hand off to Treasure.
"Jason, thanks for making time. I'm going to hand it to Treasure, who's going to walk you through what we've put together. Then Richard and Wade are going to get into the approach. And we want to leave plenty of time for your questions."
Treasure Kimp — Solution Overview
10 minutes
Terrain
Show the tech stack table. "We mapped every product and every language. PHP, .NET, Python, Node, Ruby — all under one roof." Key line: "We're not standardizing your code. We're standardizing the dispatch system."
Methodology
Hit the 5 principles: dispatch over code, system over individual, agents as employees, protect production, M&A playbook. Don't read them — summarize.
"Jason, we identified every product in your portfolio and the tech stack each one runs on. We're not here to tell you to rewrite anything. We're here to wire AI agents into the processes that already work — through Jira, GitHub, and AWS. Let me hand it to Richard and Wade to show you exactly how."
Richard + Wade — Technical
10 minutes
Agent Pipeline
Show the flow: Product Agent → Architecture Agent → Dev Agent → QA Agent. "This is exactly what you described on our call."
Arsenal
MCP as the backbone. CodeRabbit, Snyk, Cursor. Security posture. Don't go too deep — the podcast covers the detail.
  • Claude MCP — connects AI to Jira, GitHub, Confluence natively. Stack-agnostic. The backbone.
  • Agent pipeline — Product Agent, Architecture Agent, Dev Agent, QA Agent. "This is exactly what you described."
  • How it works across different languages — MCP reads the repo context regardless of PHP or Ruby
  • CodeRabbit for multi-stack code review, Snyk for security, Cursor IDE for dev experience
  • Security: enterprise data isolation, no training on DaySmart code, GDPR for UK
  • Wade: agile transformation parallel, AT&T Query Probe, healthcare AI — real proof points
  • Don't go too deep — the podcast covers the technical detail. Hit the highlights.
Richard: "Jason, you said your teams are using Claude Code and Copilot. Here's how we take that from individual tools to a system. Claude's Model Context Protocol connects directly to your Jira and GitHub — regardless of whether the code underneath is PHP or Ruby. That's what makes the agent pipeline work across all 10 products without touching the code."

Wade: "What you're describing with Nova — agents working in the process of creating software — is exactly where we live. We've done this at AT&T where we built an agentic application that optimizes queries at runtime. This isn't theoretical for us."
Podcast Moment — Optional
3-5 minutes
  • Option A: Play a 3-minute clip (0:00-3:00 or 5:00-7:30) — "We built this using AI from our call notes"
  • Option B: Share the link for Jason to listen later — "21-minute deep dive, we'd love your take"
  • Option C: Skip it if the conversation is flowing — more Q&A time is better than a demo
  • Roxy and Treasure read the room and make the call
Q&A / Discussion
10 minutes
  • Let Jason and Matt react. Listen more than you talk.
  • Matt's questions tell you exactly what Phase 1 should focus on
  • See the full Q&A reference below for anticipated questions and answers
Treasure Kimp — Phase 1 Close
5 minutes
Phase 1
$15K, 2 weeks, implementation-ready playbook. "What does your timeline look like to get started?"
Phase 2 & 3
"Scoped after Phase 1 findings. We show you what we find, then we scope the next step together."
Treasure: "Jason, Phase 1 is two weeks, $15,000. We map your SDLC, assess AI readiness across your teams, and deliver an implementation-ready playbook. Not a study. A battle plan. What does your timeline look like?"
Treasure Kimp + Roxy Close
5 minutes
Comms Log
"Here's how we built this — full transparency." Quick scroll through the 5 steps. This IS the proof we use AI.
Close
Ask for Phase 1. Timeline. Who needs to be involved. SOW within the week.
  • Treasure leads the ask: Phase 1, $15K, 2 weeks, implementation-ready playbook
  • "What does your timeline look like to get started?"
  • Roxy backs it up: "Jason, we've known each other a long time. This is different from last time."
  • Next step: SOW within the week. Who from DaySmart needs to be involved?
Treasure: "Jason, based on everything we've discussed, Phase 1 — two weeks, $15,000 — is the right starting point. We map your SDLC, assess AI readiness across your teams, and deliver an implementation-ready playbook. Not a study. A battle plan. What does your timeline look like?"

Roxy: "Jason, I wouldn't bring this team to you if I didn't believe in what they can do. Let's get started."
Prepared Answers

TOUGH QUESTIONS

Jason will push back. Here are 12 questions he's likely to ask and exactly how to answer each one.

"What does Phase 1 actually look like day-to-day?"
Richard answers
Week 1: Working sessions with your teams — product, engineering, QA. We map how each team builds software today, where time is lost, and how AI tools are currently being used. Week 2: We deliver the playbook — specific tool recommendations wired to your Jira and GitHub pipelines, top 3-5 automation opportunities ranked by impact, and a phased rollout plan. You'll have something actionable, not theoretical.
"How do you handle our different tech stacks?"
Richard answers
Claude's Model Context Protocol connects to Jira and GitHub at the operational layer — it reads the repo context regardless of whether the code is PHP, .NET, Ruby, or Node. We don't need to standardize your languages. We standardize the orchestration on top of them. CodeRabbit parses the abstract syntax trees of each language specifically, so it understands legacy PHP as well as modern Node.
"What is Kelly actually doing with AI internally?"
Whole team — be honest
We're building our AI practice right now — and that's actually an advantage. We're not selling a legacy framework. We're building in real-time using the same tools we're recommending to you. This playbook we just showed you? Built with AI. The podcast? AI-generated from our call notes. We practice what we preach every day. The companies ahead on AI hired partners who learned alongside them. We're offering to be that partner.
"You lost to an offshore competitor last time. Why should I trust you now?"
Roxy + Treasure
We learned from that. Last time we came with staffing — headcount, not strategy. This time we're bringing a specific technical approach built around your exact situation. The playbook you just saw proves we did the homework no one else is doing. We're not throwing bodies at this. We're bringing practitioners who build AI agent pipelines and embed with your teams.
"I don't need a $15,000 PDF telling me what I already know."
Treasure
Neither do we. Phase 1 isn't a study — it's an implementation playbook. You get exact wiring diagrams for how AI agents connect to your Jira and GitHub pipelines via MCP. You get specific tool configurations for your stacks. You get a ranked list of automation opportunities with estimated cycle time impact. You can use it with us or without us — that's how confident we are that you'll want us for Phase 2.
"How does this integrate with what Nova is already doing?"
Wade
We augment, not replace. Your Nova teams are in rapid experimentation right now. We come alongside them — shoulder to shoulder — and help them evaluate what's working, fill the gaps, and build the orchestration layer that connects the individual tools into a system. We're not asking them to stop what they're doing. We're helping them do it faster and make it repeatable.
"What about product managers? They can't keep up with faster engineering."
Richard
You identified this on our last call and it's exactly right. If we only accelerate engineering, we just move the bottleneck to product. Our approach includes the product agent specifically — AI that helps generate and refine requirements so product keeps pace with engineering. The whole pipeline has to move together or you get no benefit.
"What about security? Our code is proprietary."
Richard
All AI tools we deploy are configured for enterprise data isolation — no training on your code. We use enterprise tiers with data segregation. For your UK operations like Slick and TeamUp, we ensure GDPR compliance on all AI tool usage. API keys and secrets never enter AI prompts. And all AI-generated code goes through human security review before it touches production. This is baked into our approach from day one, not bolted on later.
"Can you actually deliver in 2 weeks or is that optimistic?"
Treasure
Two weeks is realistic because we've already done significant homework. We're not starting from zero. We know your tech stacks, your tooling, your processes, and your goals. Phase 1 is about validating our assumptions with your teams and turning our research into specific implementation plans. The heavy lifting of understanding DaySmart has already been done.
"What if Phase 1 doesn't show us anything we don't already know?"
Wade
If all we delivered was a summary of your current state, that would be a failure and we'd deserve the criticism. Phase 1 delivers the HOW — the specific wiring, the tool configurations, the agent pipeline architecture mapped to your processes. You know what needs to change. We show you exactly how to change it without disrupting what works.
"Who specifically would be working with our teams?"
Treasure + Roxy
Richard leads the AI architecture and tool integration. Wade brings the agentic workflow expertise — he builds 2-3 agents a week and has done this for AT&T and healthcare clients. Treasure is your day-to-day point of contact for execution. Roxy stays involved at the relationship level. These are senior people, not junior consultants. You'll be working directly with the people on this call.
"How is this different from what Accenture or Deloitte would pitch?"
Richard
They'd send you a 100-page assessment over 3 months with a team of 20 juniors. We're four senior practitioners who embed with your teams for 2 weeks and deliver an implementation playbook you can execute immediately. We don't sell frameworks — we build working systems. And we built our own pitch using the exact tools and approach we're proposing for you. No big firm is doing that.
Reminders

DON'T FORGET

Do
  • Use Jason's numbers — 35 days, 16-17, 50%
  • Reference his exact words from the call
  • Listen to Matt — his input shapes Phase 1
  • Be direct and specific — Jason is a Marine
  • Show confidence but stay honest
  • Close with a clear ask and timeline
Don't
  • Talk over each other — clear handoffs
  • Lecture Jason about his own business
  • Get deep into Phase 2/3 pricing yet
  • Dodge the "What is Kelly doing with AI" question
  • Oversell — he hates vaporware
  • Forget: Matt is day-to-day, get his buy-in