Grovework
Talk to us

AI champion networks for companies of any size

AI champion networks that take root.

Grovework builds and runs the network of employee AI champions that turns licenses into daily use, across every AI tool you pay for, without reading anyone's prompts. Start with a 90-day pilot, then run it yourself or keep us running it.

See how it works Try it on your own export Ask about a 90-day pilot

Usage metadata only. We never collect prompts.

A grove of trees above ground. Below ground their roots join into one network. Champions sit at the junctions, and fine lines reach out from each one to many colleagues.

Fig. 1A grove in cross-section. Above ground, separate trees. Below, one network. Champions sit where the roots meet, and each one reaches the colleagues around them.

  • Champion
  • Colleague
  • The network

The problem

Licenses are not adoption.

Most rollouts follow the same arc. Seats get assigned, the launch webinar fills up, and a handful of enthusiasts take off. Everyone else tries it twice and goes back to the old way of working.

What changes habits is a trusted colleague on each team who shows how AI fits the work in front of them, and keeps showing up after launch week. That takes a network with a calendar, a selection process and a report. Grovework builds and runs all three.

85%

of enterprise employees aren't using AI to drive business value, according to Section's January 2026 research.

Source: Section via Business Wire, Jan 2026 ↗

60 to 90days

is how quickly champion networks stall without a steady rhythm, in the experience of a CISO writing for CSO Online.

Practitioner view: CSO Online, Nov 2025 ↗

Why a grove

Named for what happens underground.

Trees in a grove are linked through their roots and shared fungal networks that move water, nutrients and even warning signals from tree to tree. Ecologists sometimes call it the wood wide web.

A champion network does the same work for AI know-how. It connects people across teams in a way the org chart never shows.

How it works

What Grovework runs, from launch to the quarterly report.

Five parts of one cycle, plus two that run alongside it. Champions work in Slack or Teams. Your program lead works on the web platform.

  1. Set up

    Pick your AI tools and goals, then choose programs from a menu.

    • Charter and champion role description, generated for your company
    • Launch announcement drafted and ready to send
    • Programs picked from a tested menu instead of a blank page

  2. Select and rotate

    Nominations come in, and a recommended slate comes out each quarter.

    • Self-nominations plus peer and manager nominations
    • About 1 champion per 20 to 50 users, weighted toward low-adoption teams
    • Terms of 1 to 2 quarters, with an alumni pool

  3. Run programs

    The steady rhythm that keeps a network alive past launch.

    • Onboarding cohorts, monthly working sessions and office hours
    • Use case competitions, demo days and recognition
    • A shared prompt and use case library, with calendar and reminders built in

  4. The Grove Loop

    Ideas from teams come back to them as tools they use.

    • Five stages take a team's request from seed to a planted solution
    • Cedar matches each request to a champion builder or a technical builder
    • Review scales with risk, from personal tools to system connections
    • Tickets sync with Jira and ServiceNow, so builders keep one backlog
    See the Grove Loop

  5. Measure

    Show leadership what changed, team by team.

    • Teams with an active champion compared with similar teams without one
    • Adoption, proficiency, network health, the Grove Loop and sentiment
    • A quarterly report drafted for you

Steps 2 to 5 repeat every quarter

Plus two parts that run alongside the cycle

Every week

News feed

What changed in your AI tools, with a badge that says whether you can use it here yet.

See the News feed

Every quarter

Sustainability and the Grove Fund

Energy, carbon and water estimates for your AI use, habits that cut waste, and verified restoration you can fund.

See sustainable adoption

The Grove Loop

From a team's idea to a tool they use.

Networks stick when ideas from teams ship back to those teams. Grovework connects teams, champions and builders, sends each idea down the shortest path, and closes the loop with the person who raised it.

The Grove Loop. Five stages run left to right: Seed, Shape, Build, Plant and Tend. At Build the path splits between champion builders, who work no-code in approved tools, and technical builders, who handle systems and data. A dotted shortcut runs from Shape straight to Plant when a solution already exists and a Try this guide answers it. A return path runs from Tend back to Seed, because feedback and new requests become seeds.01SeedAnyone names a problem02ShapeA champion picks the path03BuildCedar matches a builder04PlantShips with a launch kit05Tend30 and 90 day check-insFeedback and new requests become seedsAlready exists: Try this guideChampion buildersNo-code, in approved toolsTechnical buildersSystems and data The Grove Loop, top to bottom. Seed, Shape, Build, Plant and Tend. Build splits between champion builders, no-code in approved tools, and technical builders, for systems and data. A dotted shortcut runs from Shape to Plant when a solution already exists and a Try this guide answers it. A return path runs from Tend back to Seed, because feedback and new requests become seeds.Feedback and new requests become seedsAlready exists: Try this guideChampion buildersNo-code, in approved toolsTechnical buildersSystems and data01SeedAnyone names a problem02ShapeA champion picks the path03BuildCedar matches a builder04PlantShips with a launch kit05Tend30 and 90 day check-ins

Fig. 2The Grove Loop. Anyone plants a seed from Slack, Teams or the web by saying what slows them down and how often. Each seed ends as a solution the team uses, or as a quick answer when one already exists.

  • Main path
  • Shortcut
  • Back to Seed

Three paths an idea can take

  1. Already exists

    The champion finds a feature or solution you already have and answers with a Try this guide. Nothing gets built.

    Who answersThe team's champion

  2. Champion build

    No-code work inside tools you already approved, such as custom GPTs, Copilot agents, Gemini Gems and Claude projects.

    Who buildsChampion builders

  3. Builder request

    Anything that needs company systems, data or code. The shaped request goes to the people who build those.

    Who buildsTechnical buildersCentral AI team, IT, data, engineering

Every builder has a profile with platforms, skills, approved data access and monthly capacity. Cedar matches each request on those, and Linden groups similar seeds from different teams so one build serves all of them.

Review that scales with risk

  • Personal

    One person's tool. No review, and it follows your AI use policy.

  • Team

    Shared with a team. The champion lead checks data and ownership.

  • System

    Connects to company systems, uses sensitive data or faces customers. The builder team reviews it, and IT security, privacy and risk sign off.

One idea, seed to tend

Supply Chain · purchase order discrepancies

Example

  1. Seedfrom Teams

    An analyst answers two questions. What slows you down? Explaining purchase order discrepancies to vendors. How often? Every day.

  2. Shape3 days

    The team's champion finds nothing in the solution catalog that fits, estimates the hours at stake and picks a champion build.

  3. BuildTeam review

    A champion builder makes a Copilot agent that drafts each explanation from the purchase order, receipt and invoice.

  4. Plant2 teams

    Ships to 2 teams with a launch kit, joins the solution catalog, and Aspen posts it to the news feed.

  5. Tend30-day check-in

    Confirms 18 hours a month saved. Aspen tells the analyst who raised it.

Fictional team and figures, for illustration only.

The person who raised it hears back at every stage

Aspen writes them an update from seed to check-in. Rowan flags seeds that wait too long, against targets of a champion reply within 5 business days and a build decision within 2 weeks.

Technical builders keep working in their own tools

They stay in Jira, ServiceNow or Azure DevOps. Grovework creates the ticket with the shaped request and syncs status back, so nobody keeps a second backlog.

Grovework stores solution metadata only, such as name, owner, tool, risk tier and link, and never prompts or code.

Champion selection

Choose champions on evidence. Refresh them every quarter.

Most programs take whoever volunteers first. Cedar weighs six signals and recommends a slate that covers the teams that need help most. Selection never looks at the content of anyone's prompts.

Interest
Self-nomination and a short application
Peer pull
Nominations from peers and managers
Activity
Team-level usage, or individual usage for people who opt in
Coverage gaps
Teams with few champions or low adoption
Contribution
Use cases submitted, sessions led, people helped
Capacity
Manager confirms 2 to 4 hours a month

Each slate is checked for a mix of function, level and location before it reaches you.

Cedar's recommended slate

Q1 2027 · Operations division · 4 open seats

Example
  1. Priya NairAccounts Payable
    • Peer nominations
    • Capacity confirmed

    Three peer nominations. Her manager confirmed 3 hours a month.

  2. Marcus BellCustomer Support
    • Coverage gap
    • Capacity confirmed

    Support has 140 licensed users and no champion yet.

  3. Dana OkaforProcurement
    • Team activity
    • Peer nominations

    Her team's weekly use is climbing. Two colleagues named her.

  4. Luis OrtegaField Service, Denver
    • Coverage gap
    • Team activity

    Adds a champion outside headquarters.

Fictional names. Covers 3 teams with no champion today.

You make the final call.

Terms

1 to 2 quarters

Each champion serves a set term with a clear end date, which makes the role easier to say yes to.

Rotation

About a third

of the network is refreshed each quarter. New people arrive while most of the group carries on.

Alumni

Alumni pool

Past champions stay on call for sessions, mentoring and the next pilot.

Meet the helpers

Four helpers handle the busywork.

A network runs on a hundred small tasks. Aspen, Cedar, Linden and Rowan, each named for a tree that grows in groves, draft, sort and chase, so your program lead can spend time with people.

Aspen

Program assistant

Drafts charters, launch announcements, session agendas, reminders and the weekly news digest. Updates whoever raised an idea at each stage of the Grove Loop.

Sample · draft announcement

“Starting in March, every team will have an AI champion: a colleague with time set aside each month to help you get more from the tools.”

Cedar

Selection

Recommends each quarter's champion slate, explains why, and manages rotation. Matches each seed to a builder by platform, skill and capacity.

Sample · slate note

Suggest Marcus B. for Customer Support. No champion on the team yet, and his manager confirmed 3 hours a month.

Linden

Insights

Writes the quarterly report, including the sustainability page, and spots trends. Groups similar seeds so one build serves several teams.

Sample · report outline

Q3 report drafted: adoption by team, solutions planted, hours saved, and two teams to watch next quarter.

Rowan

Alerts

Flags stalled teams, missed sessions, coverage gaps, overdue rotations, seeds waiting too long, and usage spikes that push the footprint up.

Sample · alert

Customer Support hasn't held a session in 6 weeks.

Helpers draft, recommend and flag. Your program lead approves what goes out.

Stay current

What's new in your AI tools, and whether you can use it here.

Champions hear one question more than any other. “I read this tool can now do X. Can we use it here?” The News feed answers it in one place.

Every outside item carries a badge

  • Released by vendorAnnounced by the vendor. Nobody here has checked it yet.
  • Under reviewIT or the program lead is checking it now.
  • Enabled hereSwitched on and approved. Go ahead and use it.
  • Not approvedOff limits here, with the reason and what to use instead.

Vendors announce features faster than companies approve them. Your program lead or IT sets the badge, so champions stop guessing and nobody builds a habit on a feature that hasn't been cleared.

New features come with a short “Try this” exercise, and the “I tried it” count feeds the proficiency layer of your report. Each week Aspen drafts a Slack or Teams digest by tool, department and role, and your program lead approves it before it goes out.

News feed

This week · Finance · 4 tools

Example
  1. Microsoft 365 CopilotEnabled here

    Example release: meeting recap options. Recaps can group action items by owner. IT switched this on for Finance last week.

    Try thisOpen last week's close meeting recap and ask for the action items grouped by owner.

    Source link: vendor's official note

  2. ChatGPT EnterpriseUnder review

    Example release: shared project spaces. Teams can keep files and instructions in one place. IT is checking where those files are stored.

    Source link: vendor's official note

  3. ClaudeReleased by vendor

    Example release: longer document uploads. Larger files fit in a single conversation. Announced by the vendor and not yet checked here.

    Source link: vendor's official note

  4. GeminiNot approved

    Example release: personal cloud storage. Would let people attach files from personal accounts. Use the company drive instead.

    Source link: vendor's official note

  5. Internal · Champion winApproved by program lead

    Priya N., Accounts Payable. Built a prompt that drafts vendor follow-up emails from the aging report. Her team now runs it every Monday.

Demo data. Generic example items, not real vendor announcements, and source links are placeholders. Names are fictional.

Measurement

Measure what changed, team by team.

The headline view compares teams that have an active champion with similar teams that don't. Usage comes from the admin reports your AI tools already produce.

Weekly active share of licensed users

By month since the network launched

Illustrative
  • Teams with an active champion
  • Similar teams without one

Fig. 3Illustrative data showing the shape of the headline view. Not a client result.

Show the numbers
MonthWith a championWithout one

Five layers in every report

  • Adoption

    Active users and depth of use by team, trended against teams without champions.

  • Proficiency

    Pulse checks and training completion.

  • Network health

    Coverage, attendance, people helped, and rotation on schedule.

  • The Grove Loop

    Seeds planted, time from seed to plant, reuse across teams, and confirmed hours saved.

  • Sentiment

    Confidence and trust in AI, from short pulse surveys.

Using the Grove Fund? Linden adds a sustainability page with footprint ranges and certificates retired. See how it works

Your data stays yours

Usage metadata, never content.

Grovework never collects prompts, responses or files. Adoption comes from the aggregate usage reports your AI tools already give admins, shown by team. Every client starts without giving us any system access.

Fig. 4Data tiersYou decide how much to share. Each step up is optional.
Tier 0

Program only

Champion names, work emails, departments, survey answers, seeds and solution metadata.

No connection to your AI tools

Tier 1

Usage reports by upload

Aggregate usage exports from your AI tools' admin consoles, uploaded monthly.

No system access

Every client starts on Tier 0 or Tier 1

Tier 2Later

Read-only reporting connections

Read-only connections to your tools' admin reporting, in place of manual uploads.

Read-only, reporting APIs only

Tier 3Later

Customer-hosted

Grovework runs in an environment you host and control.

Data never leaves your environment

Minimum group of 5

Adoption is reported by team, and groups smaller than 5 people are never shown on their own. A report never reads as monitoring individuals.

No prompts collected

Prompts, responses and files never reach Grovework. Champion selection never looks at what anyone typed.

Pseudonymized IDs

Employee IDs can be pseudonymized with keys your company controls.

Delete anytime

You can delete everything Grovework holds about your company, at any time.

Opt-in onlyIndividual usage data is used only for champions who choose to share it.

Works with the tools you already use.

  • Microsoft 365 Copilot
  • ChatGPT Enterprise
  • Gemini
  • Claude
  • Your internal tools

Grovework runs the people side of adoption. Your AI tools stay as they are, and you can mix them or switch later without rebuilding the network.

Sustainable adoption

Adopt AI sustainably, with numbers you can stand behind.

Sooner or later someone asks what your AI use costs in energy, carbon and water. Grovework estimates it from the usage data we already have, helps champions cut the waste, and runs the Grove Fund so your company can pay for verified restoration. Your sustainability team gets an answer for AI in their Scope 3 reporting.

  1. Measure

    Energy, carbon and water, estimated from usage you already share.

    Interactions, tokens or spend by tool and team. Every figure is a low, central and high estimate with its data quality labeled.

  2. Reduce

    Champions teach habits that cut waste without cutting use.

    A right-sized model for each task, with reasoning, image and video generation saved for work that needs them. Rowan flags spikes, and Linden shows footprint per hour saved.

  3. Contribute

    The Grove Fund turns a rule you set into verified restoration.

    Per 1,000 AI interactions, per licensed seat each month, or a share of AI spend.

  4. Report

    A quarterly sustainability page in Linden's report.

    Method, ranges and certificates retired, with figures ready to export for Scope 3 reporting.

AI footprint for one quarter

Example company · 1.2 million AI interactions

Illustrative
  • EnergyData quality: Usage-based

    Central estimate 1,200kWh

    360 to 3,600 kWh10x spread

  • CarbonData quality: Usage-based

    Central estimate 480kg CO2e

    36 to 1,440 kg CO2e40x spread

  • WaterData quality: Usage-based

    Central estimate 12,000Labout 3,200 gal

    360 to 54,000 L150x spread

Covering the high water estimate takes 15 Water Restoration Certificates at 1,000 gallons each.

Fig. 5Illustrative. Demo factors per interaction, applied to 1,150,000 text interactions and 5,000 images, each image counted as 10. Bar length shows the spread from low to high on a log scale. Not a client result.

Spend-basedUsage-basedProvider-reported

Why every figure is a range

Published per-prompt figures differ by more than 100x on water, because they count different things.

0.26mL

of water for a median Gemini text prompt, as Google reports it, counting data center cooling. Google also reports about 0.24 Wh and 0.03 g CO2e. Not independently verified.

Google Cloud blog ↗

45mL

of water for a 400-token response in Mistral's lifecycle study, which also counts electricity generation and hardware. It reports 1.14 g CO2e.

Mistral AI ↗

Both can be right about what they measure. So Grovework never shows one precise-looking number. Every estimate carries a low, central and high value and a data-quality label, and the method is published and updated as AI providers disclose more.

Step 3 in detail

The Grove Fund

Your company sets a contribution rule, and the fund pays for verified restoration against it. Grovework tracks every purchase and reports what was retired.

Pick a rule

  • 01Per 1,000 AI interactions
  • 02Per licensed seat, each month
  • 03A share of AI spend

  • Water Restoration Certificates

    For water claims

    1 certificate = 1,000 gallons restored. Third-party verified and registered.

    What a certificate is, BEF FAQ (PDF) ↗

  • Verified carbon removal credits

    For carbon claims

    Credits for carbon removed from the atmosphere, verified and bought from vetted partners.

  • Tree planting

    For your people

    A visible layer for employees, with one tree per solution planted through the Grove Loop. Kept separate from any carbon claim.

Guardrails

Claims your sustainability team can sign off on.

Contribution language only

Say

“We funded restoration of 48,000 gallons.”

Avoid

“Our AI use is carbon neutral.”

Water certificates for water claims

Water Restoration Certificates back water claims only. Carbon claims need carbon removal credits.

We never hold your money

You buy from vetted partners directly or through a partner's purchasing API. Grovework tracks and reports what was retired.

A published method

Our footprint method is public, and we update it as AI providers disclose more.

Ways to work with us

Start with a 90-day pilot.

We run your champion network with you for one quarter, across every AI tool you pay for, against measures agreed at kickoff. After that, keep us running it or take it over yourself.

Where most clients start

90-day pilot

Grovework runs the network with your program lead for one quarter.

  • A launch plan built from your own usage exports in the first meeting
  • Champion slate, sessions and a first use case challenge, run by us
  • A weekly scorecard against five measures agreed at kickoff
  • A results report for your sponsor at day 90
  • Every module, the full Canopy plan, for the 90 days

Fixed price, shared on a call

Ask about a pilot

Managed network

After the pilot, Grovework keeps running the network alongside your program lead.

  • Selection and rotation every quarter
  • Sessions, challenges and demo days, run by us
  • Monthly build clinics, where champions bring shaped ideas to builders
  • Quarterly reporting for your leadership
  • Runs on any platform plan

Pricing on request

Ask about managed

Self-serve platform

Your own program lead runs the network on Grovework.

  • All five parts, from setup to measurement
  • Aspen, Cedar, Linden and Rowan
  • Seedling, Grove or Canopy, sized to your company
  • Slack or Teams for champions, web for program leads

Pricing on request

Ask about the platform

Platform plans for any size of company

Priced by the number of people with AI licenses. Bundles cost less than adding modules one at a time, and sustainability is free in every plan.

Seedling

The core network. Small enough for teams under 250 AI users.

  • Champion selection and rotation
  • Programs, sessions and challenges
  • Slack or Teams app and the News feed
  • Weekly scorecard and results report
  • Starter library of agendas and messages
  • Grove Fund, free

Grove

Seedling plus the Insights pack.

  • Everything in Seedling
  • Linden's quarterly report for leadership
  • Adoption across every AI tool, side by side
  • Quarterly one-pager for your sponsor

Costs less than adding Insights on its own

Canopy

Everything, including the Grove Loop.

  • Everything in Grove
  • The Grove Loop, from a team's idea to a built solution
  • Builder matching and a home for builders
  • Ticket sync with Jira, ServiceNow or Azure DevOps

Insights pack included at no extra cost

The Grove Fund is free in every plan. Footprint estimates and contributions never cost extra, because sustainable AI adoption should be the default. Purchases go to vetted partners with no markup.

Run it yourself

Playbook

Our starter library of session agendas, launch messages, challenge briefs, role descriptions and checklists, plus a launch plan built from your own exports. Pricing on request.

Ask about the Playbook
Add-on

Skills add-on

For mature networks. Turn proven solutions from the Grove Loop into reusable AI skills that teams can run in their own tools. Pricing on request.

Ask about skills

Who we are

A small company in Chicago.

Our founder built and scaled AI champion programs inside a large enterprise, and learned what keeps them going after launch week: a steady calendar, fair rotation, and reporting that leadership believes.

Grovework turns that experience into a platform and a service that run on any AI tool. He draws and paints, and he drew the Grovework logo.

Questions

What buyers ask us first.

Which AI tools do you work with?
Microsoft 365 Copilot, ChatGPT Enterprise, Gemini, Claude and internal tools. Grovework runs the program around them, so you can mix tools or switch later. Adoption data comes from each tool's admin usage reports.
Do you see our prompts?
No. Grovework never collects prompts, responses or files. We work with usage metadata, reported by team with a minimum group size of 5. Individual data is used only for champions who opt in, and you can delete everything at any time.
What does a 90-day pilot include?
A launch plan built from your own usage exports, a champion slate, the session calendar, a first use case challenge and a weekly scorecard against five measures agreed at kickoff. At day 90 your sponsor gets a results report. The price is fixed and agreed before you start.
How is this different from Microsoft's champion app?
Microsoft's free Champion Management Platform handles sign-ups, points and events for Microsoft 365, and champions log their activity by hand. Grovework runs the program across every AI tool you use: it proposes and rotates champions, keeps the calendar going, records activity automatically and reports adoption by team. If you use Microsoft's app today, we import your champion list.
We already have an adoption analytics tool. Do we need this?
Keep it. Analytics tools show where adoption is low. Grovework runs the program that raises it: champions, sessions, challenges and the path from a team's idea to a built solution. It works from your AI tools' own admin usage reports, so the two sit side by side.
Is this a training program?
No. Training gets people started, and champions keep them going after it ends. Grovework runs the network that does that. If you also need courses, we'll point you to a training partner.
Can our security team review it first?
Yes. We share a security overview covering the data flow, what we collect and never collect, how long data is kept, and the status of each control, including what is still planned, such as SOC 2. It includes answers your team can drop into a security questionnaire.
How long does it take to launch?
Setup generates your charter, champion role description and launch announcement, so most of the calendar goes to your nomination window and internal approvals. We map the timeline with you in the first conversation, before you commit to anything.
Who runs it day to day?
On the self-serve platform, your program lead does, with Aspen, Cedar, Linden and Rowan drafting, recommending and flagging. With a managed network, Grovework runs selection, sessions, competitions and reporting, and your lead approves each slate. Champions give 2 to 4 hours a month, confirmed by their manager.
Do our engineers need to use another tool?
No. When an idea needs a technical builder, Grovework creates the ticket in Jira, ServiceNow or Azure DevOps with the shaped request and syncs its status back. Engineers keep one backlog, and Grovework stores only solution metadata, never prompts or code.
How do we prove ROI?
The quarterly report compares teams with an active champion against similar teams without one. The Grove Loop adds hours saved for each solution, confirmed at the 30 and 90 day check-ins, so the report ties adoption to work that actually changed.
How do you estimate the footprint of our AI use?
We start from the usage metadata Grovework already receives, such as interactions, tokens or spend by tool and team, and never from prompts. Per-interaction factors turn that into energy, carbon and water, shown as low, central and high estimates with a data-quality label. The method is published, and we update it as AI providers disclose more.
Is this carbon offsetting?
The Grove Fund is a contribution, and we help you describe it that way. You can say you funded restoration of 48,000 gallons. We don't support claims like carbon neutral AI. Water Restoration Certificates back water claims only, verified carbon removal credits cover carbon, and tree planting is an engagement layer kept apart from any carbon claim.

Talk to us

Grow the network that grows your AI adoption.

Tell us where your rollout stands. We'll set up a first conversation about your goals, your tools, and whether a managed network or the platform fits.

  • Based in Chicago, working with teams anywhere
  • Any AI tool your company already licenses
  • Usage metadata only, never content

We only use this to reply to you.