Field Guide·How a team actually runs the cycle

A working field guide.

The cycle on one page. The rituals. The maturity model. A week-by-week walk through your first cycle. Read the methodology to understand why; read this to understand how.

01

The cycle, on one page.

Eight beats. Each beat names what the team is doing, what AI is doing, and what gets produced.

  1. 01

    Feed any teammate

    Team: drop Inputs into the Field: call notes, decision memos, tickets, research, interviews.

    AI: ingests, structures, deduplicates.

    Produces: Inputs · classified, indexed, searchable.

  2. 02

    Winnow Winnow

    Team: promote Seedlings, demote weak claims, ask for re-cites.

    AI: proposes Nutrients and Seedlings from Inputs; attaches the evidence trail.

    Produces: Nutrients, Seedlings · Seeds when evidence supports.

  3. 03

    Plant Flourish

    Team: Select a Seed, the recommendation the team commits to growing.

    AI: drafts the Grow Plan, inheriting Soil + the evidence trail.

    Produces: a Grow Plan · context, Constraints, Tasks, Measures.

  4. 04

    Cultivate Flourish + AI Builder

    Team: review draft Tasks; steer the build.

    AI: Tasks handed to Claude Code, Lovable, Cursor or your own agent. Growth comes back with trail attached.

    Produces: Growth · built product change.

  5. 05

    Inspect Assay

    Team: decide what to Prune, what to ship.

    AI: compares Growth against the Seed and Soil it should honour; flags drift, missed Constraints, weak spots.

    Produces: Assay Report · Pruning recommendations.

  6. 06

    Gather Bushel

    Team: shape and approve the launch artefacts.

    AI: drafts the Sheaf, the nine audience-specific artefacts a Harvest needs.

    Produces: a Sheaf, audience-grouped.

  7. 07

    Weigh Tally

    Team: call whether the Yield is honest; identify gaps.

    AI: reads metrics against the Measures set in the Grow Plan.

    Produces: Yield Report · expected vs. actual.

  8. 08

    Glean Glean

    Team: Turn the Soil, fold Gleanings back into the strategic context.

    AI: gathers post-Harvest feedback, surfaces the surprises.

    Produces: Gleanings · next-cycle Seeds.

02

Rituals & cadences.

Three named meetings a Cultivating team holds. Each replaces one ceremony most teams are doing badly anyway. None are mandatory; all tend to emerge after the second cycle.

Weekly · 30 min

Signal review

The team reads the week's Inputs together. Decide what becomes a Nutrient. Promote or demote Seedlings against last week's set. Replaces the standing “research review.”

Reads
Inputs · Seedlings · Refuted
Decides
What gets promoted, demoted, re-cited.
Cycle-start · 60 min

Seed evaluation

The team reviews the Seed slate, reads the evidence trail on each, and selects the one (rarely two) Seeds to Plant this cycle. Replaces the prioritisation meeting where nobody had context.

Reads
Seeds · Soil · last cycle's Yield
Decides
Which Seed becomes a Selected Seed.
Post-Harvest · 45 min

Harvest review

The team reads the Yield Report, looks at Gleanings, and Turns the Soil. The output is a list of revisions to Soil and a fresh Seed slate. Replaces the retrospective and the launch post-mortem in one.

Reads
Yield Report · Gleanings
Decides
What turns back into Soil. Next Seeds.
03

The cultivation maturity model.

Five stages, in plain language. Use this as a diagnostic: where is our team today, where would we like to be in two cycles, and what is in the way?

  1. 01Untended

    Ships from a backlog. Evidence is anecdotal.

    • No explicit evidence trail behind decisions.
    • Success is judged in the rear-view mirror.
    • The wiki rots faster than it is read.
  2. 02Sown

    Has a Field. Captures evidence. Produces Seeds.

    • Inputs, Nutrients and Seedlings are real artefacts.
    • An evidence trail exists on most Seeds.
    • Success measures are still set after the Harvest.
  3. 03Tended

    Measures set before the Harvest. Yield read honestly.

    • Every Grow Plan has expected Yield.
    • Yield gaps are read in public, not buried.
    • The cycle is real, but mostly manual.
  4. 04Cultivated

    Runs the full eight beats. Learnings feed the next cycle.

    • Gleanings are returned to the Field as Nutrients.
    • The Field compounds knowledge across cycles.
    • The team can show the audit trail without preparing for it.
  5. 05Self-renewing

    AI handles synthesis at every beat. Humans hold every cultivation moment.

    • The full agent suite is wired in.
    • Cycle time is no longer governed by hand-offs.
    • The Field learns faster than it forgets.

    Stage 5 is a direction, not a claim. Assay, Tally and Glean are growing: the destination is real, the road is partly built.

04

Your first cycle.

Four weeks. Not prescriptive (a Field can run a cycle in a day or a quarter) but this is the shape most beta teams found.

  1. Week 0

    Set the Soil.

    Two to four hours, once. Fill in Proposition, Market, Lifecycle, Constraints, Vision. Connect Inputs sources: your call recorder, your decision-log, your ticket tracker. Drop in last quarter's research.

  2. Week 1

    Feed & Winnow.

    Winnow proposes Nutrients and Seedlings from the Inputs. The team holds a Signal review: promote, demote, re-cite. By end of week, there should be a Seed slate with an evidence trail on each.

  3. Week 2

    Plant.

    Seed evaluation meeting: pick one Seed (rarely two). Flourish drafts the Grow Plan with Tasks and Measures inherited from Soil. Hand off to your AI builder (or your humans) and let the work begin.

  4. Week 3

    Inspect & Gather.

    Growth comes back, Assay inspects, the team Prunes. Bushel drafts the Sheaf. The team shapes audience-specific artefacts. The Harvest is scheduled with expected Yield locked in.

  5. Week 4

    Harvest, Weigh, Glean.

    Ship the Sheaf. Tally reads the Yield against the Measures. Glean gathers what comes back: the surprises, the confirmations, the misses. Hold the Harvest review. Turn the Soil. The next cycle starts from a wiser Field.

05

Vocabulary, v2.

The surviving cultivation vocabulary, each term with its plain-English gloss. v1 terms that were cut (Furrows, Sunline, Aftergrowth, Regather) are not listed; they are no longer canonical.

Field
A product. The living environment that holds everything the product knows, produces and learns.
Soil
Strategic context. Five parts: Proposition, Market, Lifecycle, Constraints, Vision.
Inputs
Source material as ingested: calls, memos, tickets, research.
Nutrients
Discrete, citable evidence extracted from Inputs.
Seedlings
Working hypotheses. Promoted to Seeds when evidence supports them.
Seeds
Recommendations ready to act on. Each carries its evidence trail.
Grow Plan
An implementation brief: context, constraints, Tasks, Measures.
Growth
Built product change. The output of implementing a Grow Plan.
Sheaf
A launch bundle: the audience-specific artefacts a Harvest needs.
Harvest
The act of releasing. A Harvest is a verb event; a Sheaf is the noun.
Yield
The measured outcomes of a Harvest. Read against Measures.
Gleanings
Post-Harvest feedback. Returned to the Field as new Nutrients.
Growth Rings
The Field's change history. Append-only.

Run one cycle. Decide for yourself.

Closed beta. One cycle is enough to know whether the cultivation model is the right shape for your team.