AI Action Plan — Evergreen & Oak Landscaping

Built from your onboarding answers. These are the highest-leverage places AI changes how Evergreen & Oak Landscaping runs, starting with what you'd notice fastest. Each item leads with the outcome, then the tool and likely owner. Mark items [VERIFIED] as you confirm the fit, [SKIP] if one doesn't.

Where AI fits in Evergreen & Oak Landscaping

Evergreen & Oak runs on three revenue lines that each move at a different speed: design-build jobs that close slow and big, maintenance contracts that renew once a year but need steady servicing, and snow removal that has to scale up fast when weather hits. Dana sells and estimates, Sofia designs, Marcus runs the maintenance crews, and Priya keeps the books, but almost everything that touches a customer before or after the actual work still passes through Dana's Gmail by hand. That's the bottleneck AI is best positioned to remove first.

The operational reality is that a nine-person company with a $15k-$60k sales cycle, 74 recurring accounts, and paper timesheets in the field cannot afford to have its owner doing manual follow-up and its bookkeeper re-typing the same job twice. Every hour Dana spends chasing a quote is an hour she isn't designing the next one, and every week Priya spends re-entering Jobber data into QuickBooks is a week where the books lag the business. The single most impressive thing AI could do here: Dana opens Gmail one morning and every quote that's gone quiet for five days already has a follow-up drafted in her own voice, waiting to be sent. That's quick win #1.

Top 5 quick wins (start here)

1. Every stalled quote gets a follow-up drafted in your voice, waiting in your inbox

• The win: Dana opens Gmail and finds a draft reply already sitting under every quote that's gone quiet for five days. No blank page, no "did I already follow up on that." She reads it, tweaks a line if she wants, hits send.
• Why it matters: Quote follow-up is the one thing Dana knows is costing her money and the one thing she has the least time for. This turns a task she avoids into a task she just approves.
• How it works: Jobber shows quote status and dates; the AI tool scans for anything past five days with no reply and drafts a follow-up in Gmail, pulling job details straight from Jobber so nothing sounds generic.
• Tool: Jobber + Gmail + Claude/ChatGPT
• Likely owner: Dana Whitfield
• Effort to start: Low
• First step: Pull last month's stalled quotes from Jobber and have the AI draft follow-ups for those first, so Dana can see the voice match before trusting it on new ones.

2. A season's worth of before-and-after posts, drafted from Sofia's photo folder in one sitting

• The win: Sofia sits down once with a folder of before-and-after job photos and walks out with a stack of Instagram captions ready to schedule, not one post at a time squeezed between site visits.
• Why it matters: Evergreen & Oak's best marketing asset is finished work, and right now it only gets posted when someone remembers to. This turns a backlog of photos into a season of content in one pass.
• How it works: The AI tool looks at each photo pair plus the job type from Jobber and writes a caption in the business's voice; Sofia picks the ones worth posting to Instagram.
• Tool: Jobber + Instagram + Claude/ChatGPT
• Likely owner: Sofia Ramirez
• Effort to start: Low
• First step: Have Sofia pull ten before-and-after pairs from recent design-build jobs and run them through in one batch to see the caption quality.

3. A reply is drafted for every new Google review before Dana even sees the notification

• The win: A new Google review comes in and a thoughtful, specific reply is already drafted, referencing the actual job if the reviewer mentions one. Dana reads it, sends it, and the response window that used to sit open for days closes in minutes.
• Why it matters: Review replies are exactly the kind of task that's easy to keep meaning to do and never get to. Fast, specific replies are also what turn a review into the next referral.
• How it works: The AI tool checks Google Business Profile for new reviews and drafts a reply that references the job type or crew when the review mentions one.
• Tool: Google Business Profile + Claude/ChatGPT
• Likely owner: Dana Whitfield
• Effort to start: Low
• First step: Run it against the last ten reviews on the profile to check tone before turning it loose on new ones.

4. Crew hours go from a photo of a paper timesheet to a QuickBooks-ready entry

• The win: Marcus snaps a photo of the paper timesheet like he already does, and instead of that photo sitting in a text thread, the hours land as a structured entry ready for Priya to check against Jobber, not retype.
• Why it matters: This is the exact seam where Priya's week goes sideways: paper in the field, spreadsheet-shaped work at the desk. Closing that gap gives her back the hours she spends transcribing instead of reconciling.
• How it works: The AI tool reads the timesheet photo and produces structured hours-by-job-by-crew-member output that Priya can check against Jobber before it goes into QuickBooks.
• Tool: Jobber + QuickBooks Online + Claude/ChatGPT
• Likely owner: Priya Shah
• Effort to start: Low
• First step: Have Marcus text a few past timesheet photos and see how cleanly the output maps to a real job and crew member before relying on it for a live week.

5. Your AI knows the whole business from the first message

• The win: Every AI session starts already knowing who Evergreen & Oak is, what it sells, who's on the team, what tools it runs on, and who the customers are, no re-explaining, ever. Every other win on this list gets faster and sharper because the AI has full context from word one.
• Why it matters: Most AI time is lost re-establishing context. Once the brain is connected, the AI works from the real business instead of guesses.
• How it works: The generated CLAUDE.md + knowledge-base/ files load into the AI tool they already use, so it carries the business into every prompt.
• Tool: Claude Projects, ChatGPT custom instructions, or Gemini Gems, whichever they already use.
• Likely owner: Dana Whitfield
• Effort to start: Low · Pays back from the first session
• First step: Drop the generated CLAUDE.md + the knowledge-base/ folder into a Claude Project (or paste system-prompt.md into ChatGPT custom instructions). Re-run one win above with vs. without the brain loaded; the difference is immediate.

Bigger bets

Jobber and QuickBooks stay in sync on their own, with AI checking the exceptions

• The win: Jobs move from Jobber to QuickBooks without Priya re-keying them, and instead of trusting a black-box sync, she reviews a short list of exceptions the AI flags, mismatched amounts, missing job codes, duplicate entries, instead of every line.
• Why it matters: This removes the recurring weekly task entirely instead of just making it faster, and it frees Priya to spend her bookkeeping time on the business's financial picture instead of data entry.
• How it works: A sync tool connects Jobber and QuickBooks; the AI tool reviews the sync log each cycle and surfaces anything that looks off for Priya to check by hand.
• Tool: Jobber + QuickBooks Online + a sync connector (new) + Claude/ChatGPT
• Likely owner: Priya Shah
• Effort to start: Medium

An AI-drafted renewal campaign for all 74 maintenance clients, aimed at 90

• The win: Instead of renewals trickling in as clients happen to call, every one of the 74 maintenance accounts gets a personalized renewal outreach drafted for the season, sequenced so Dana isn't sending 74 emails by hand, with the gap to 90 accounts addressed as a deliberate push, not a hope.
• Why it matters: Growing from 74 to 90 accounts is a stated goal, and right now nothing systematically works that gap. A drafted, sequenced campaign turns a growth target into an actual campaign instead of a number on a whiteboard.
• How it works: The AI tool pulls the maintenance client list from Jobber, drafts a renewal email per account referencing their service history, and stages a follow-up sequence in Gmail for Dana to send in batches.
• Tool: Jobber + Gmail + Claude/ChatGPT, staged as an email sequence
• Likely owner: Dana Whitfield
• Effort to start: Medium

An AI-assisted design-build proposal builder that turns a site visit into a client-ready estimate

• The win: Sofia walks a property, and instead of building each $15k-$60k proposal from scratch in Docs, she starts from an AI-assembled draft that pulls in the plant and material choices, standard line items, and a written scope description, so the proposal goes from site visit to client-ready in one sitting instead of days.
• Why it matters: Design-build is the highest-dollar line in the business, and the proposal is the moment that closes it. Cutting the drafting time doesn't just save Sofia's time, it shortens the sales cycle on the biggest jobs.
• How it works: A structured proposal template lives in Google Docs; the AI tool fills it from Sofia's site-visit notes and a standard pricing reference, and Sofia finishes the design specifics.
• Tool: Google Workspace + Jobber (pricing reference) + Claude/ChatGPT, with a structured proposal template (new)
• Likely owner: Sofia Ramirez
• Effort to start: High

Suggested order — what to do first

Start by connecting the brain (win 5). It costs nothing to set up and every other win below gets sharper once the AI already knows the business instead of being re-explained each time. Do this first even though it isn't the most visible win, because it's the foundation the rest sit on.

From there, start with the stalled-quote follow-up (win 1). It's the single task Dana already knows is costing the business money, the setup is just connecting Jobber and Gmail, and the payoff is immediate and visible every morning. Pair it naturally with the review-reply win (win 3): both live in the same "AI drafts, Dana approves" pattern in Gmail, so once the first is working the second is a short extension, not a new system to learn.

Next, take on the timesheet win (win 4). It's slightly more operational since it involves Marcus and the field crews, not just Dana at her desk, but it directly relieves Priya's worst weekly bottleneck and sets up the data hygiene that the Jobber-QuickBooks sync bigger bet will eventually depend on.

Then move to the Instagram batch win (win 2). It's the most self-contained of the five, doesn't touch money or customer communication, and is a good one to hand to Sofia once she's seen the pattern work elsewhere.

Once two or three of these are running comfortably, take on the maintenance renewal campaign first among the bigger bets. It builds directly on win 1's drafted-outreach pattern, and the 74-to-90 account goal gives it a clear, already-stated reason to exist. Save the Jobber-QuickBooks sync and the proposal builder for after that, once the team has confidence in AI-drafted output that touches money and top-of-funnel sales.

Things to NOT spend AI on yet

• Don't build a customer-facing chatbot before the knowledge base exists. Without a documented brain behind it, a chatbot will guess at pricing and service areas, and a wrong answer to a homeowner costs more trust than no chatbot at all.
• Don't automate the design consultation itself. Sofia's site visits and the trust she builds walking a property with a homeowner are the reason design-build closes at $15k-$60k; that's relationship-heavy work, not a drafting task.
• Don't hand crew scheduling or routing fully to AI while timesheets are still coming in as photos of paper. Automating a decision on top of unreliable underlying data just moves the error upstream instead of fixing it.

How to use this plan

1. Pick one quick win and do the "First step."
2. Once it's in motion, pick the next one. Don't parallelize too many.
3. Revisit this plan after 2-3 wins land. The bigger bets may reshape.

Last updated: 2026-07-12