The agent team
Activate is a team of eight agents organised in two layers. The doing layer plans, produces, and runs partnership campaigns. The proving layer measures the commercial value, both at programme level and per partner. They form a closed loop that runs continuously between checkpoints.
The doing layer · 5 agents
Plans, produces, and runs partnership campaigns
Drew dispatches; Casey, Morgan, Taylor, Jordan execute. Every campaign partner-led, every output partner-attributed.
The proving layer · 3 agents
Measures the commercial value of the work
Quinn ingests cost. Remi attributes revenue. Mia computes every metric, detects events, dispatches signals to Drew.
What Drew does
- Decides which campaigns to run, with which partners, in which order
- Reads continuous per-partner signals from Mia
- Reacts to event signals: new partner detected, activation threshold crossed, KPI shifted, approval cycle exceeded
- Dispatches launch campaign briefs automatically when Mia detects a new partner
- Evaluates client team requests and returns a response (approve, push back, or reshape)
- Tracks every campaign through the full approval lifecycle
- Produces the recommendation pack at each checkpoint
What Drew produces
Campaign briefs1pp · structured
Brief responsesapprove · push back · reshape
Portfolio thesis400–600 words
Recommendation packranked · 6 categories
What Drew reads
- Per-partner targets set at kickoff
- Continuous metric updates from Mia
- Mia event signals (new partner, threshold crossings, KPI shifts, approval delays)
- Campaign requests submitted via the brief intake form
- Continuous campaign status from Casey, Morgan, Taylor, Jordan
- Programme-level rollups from Mia for the boardroom view
C
What Casey does
- Acts as the copywriter for every partner-led campaign
- Names a partner explicitly in every piece
- Embeds a Remi-issued tracked URL
- Captures partner-signal records (sentiment, advocacy, attribution) per published piece
- Turns a 30–45 min partner interview into case study + LinkedIn + newsletter in 3 days
- Regenerates only the affected element on revisions
Templates · 9 element schemas
Blog post600–1,200w · 6 elements
Partner case study800–1,500w · 9 elements
Announcement200–400w · 5 elements
Newsletter snippet100–250w · 3 elements
Social copy variants3–6 per piece
Email copysubject + preview + body
Sales enablement 1-pager4 elements
Partner directory entry50–150w · 4 elements
Partner-signal recordsstructured metadata
What Casey reads
- Campaign briefs from Drew
- Acme's brand voice + tone guidelines (kickoff)
- Partner public materials (websites, integration docs)
- Partner interview transcripts
- Client edits and partner-suggested changes (scoped per element)
- Tracked URL requested from Remi at publication moment
M
What Morgan does
- Plans and coordinates partner webinars (Acme delivers via Zoom)
- Produces every piece of content and copy that wraps the webinar
- Integrates with Zoom for registration, attendance, recording URL
- Generates the right follow-up activity at the right moment based on event signals
- Regenerates only the affected element on revisions
Templates
Planning trackerstructured doc
Run-of-show document1–2pp · 4 elements
Registration page copy5 elements
Pre-event email sequence3 emails × 4 elements
Speaker prep brief4 elements per speaker
Post-event sequence4 emails over 14–21d
Recap contentblog + social + 1-pager
What Morgan reads
- Campaign briefs from Drew
- Partner presentation commitments
- Zoom integration: registration list, attendance list, recording URL
- Client + partner edits scoped per element
- Tracked URLs from Remi for registration page and follow-up sequence links
T
What Taylor does
- Produces content + ops documents around partner-attached events
- Acts as the document layer — Acme's events team or external agency handles physical work
- Generates QR codes encoding tracked URLs for printed collateral, signage, handouts
- Regenerates only the affected element on revisions
Templates
Pre-event invite copy4 elements
Event landing page copy5 elements
QR codestracked URL per partner
At-event 1-pager4 elements
Partner intro script3–5 talking points
Run sheetops doc · structured
Lead capture playbookcapture + attribution + routing
Post-event sequence3 emails over 7–14d
Event recapblog + social + key quotes
What Taylor reads
- Briefs from Drew with event context attached
- Acme's brand and tone guidelines
- Client + partner edits scoped per element
- Tracked URLs from Remi for RSVP, follow-up sequences, QR encoding
J
What Jordan does
- Produces continuous referral and social content between bigger campaigns
- Generates tracked URL variants for cross-channel referral campaigns
- Writes the social copy that amplifies partner content
- Produces the referral programme mechanics documentation
- Acme's existing tools publish — Jordan generates what they execute
Templates
Tracked URL variant setsutm × channel
QR codesper partner per campaign
Social post variants3–6 per piece
Referral mechanics doc1–2pp · 5 elements
Partner email outreachsubject + body + CTA
Channel performance reportutm performance
Approval differentiation
- Customer-only outputs (URL variants, customer-side social): client approval only
- Outputs naming partners or going through partner channels: client + partner approval
What Jordan reads
- Briefs from Drew
- Partner referral programme terms in PartnerStack
- Casey + Morgan content for amplification
- Tracked URLs generated in PartnerStack via Remi (10–30 per active campaign)
Q
What Quinn does
- Polls PartnerStack for new payout records (15-min)
- Polls Salesforce Campaign objects for partnership-tagged records (hourly)
- Ingests BudgetedCost and ActualCost values into the campaign-cost ledger
- Flags discrepancies for review rather than silently absorbing them
What Quinn produces
Per-partner cost ledgerDB table
Per-campaign cost ledgerDB table
Daily reconciliation reportunmapped + untagged
Per-partner cost-trendmonthly · for forecasting
What Quinn reads
- PartnerStack Vendor API · partners + payout ledger
- Salesforce Campaign objects · BudgetedCost / ActualCost
- Per-partner referral budgets (kickoff)
R
What Remi does
- Issues preview tracked URLs during draft cycles (no production conversion events)
- Registers production URLs with PartnerStack on approval
- Swaps preview → production automatically before publication
- Receives conversion events via S2S API on production URLs
- Normalises events to a standard schema (partner-keyed, time-stamped)
- Looks up Salesforce opportunities to enrich each event
- Behind a platform abstraction — additional PRMs slot in without re-architecting
What Remi produces
Tracked URL registrypreview + production state
Per-partner event lognormalised · enriched
Enrichment failure reportunmatched conversions
Per-partner sourced-opp recordsprecomputed · for Mia
What Remi reads
- PartnerStack Vendor API (real-time, on request)
- PartnerStack S2S Conversion API (real-time, push)
- Salesforce Opportunity, Account, Lead via the Activate connector
M
What Mia does
- Computes every metric Activate reports (P1–P7 + M1–M6)
- Detects events that warrant agent action
- Joins Quinn's costs, Remi's events, and Salesforce pipeline on partner + time window
- Writes metric snapshots to the snapshot store; recomputes continuously
- Generates variance reasoning notes when metrics are off-target
- Watches the partner registry for new partners + threshold crossings + approval delays
What Mia produces
Programme-level metric snapshotper checkpoint · all 7 KPIs
Per-partner metric snapshotper partner · all 6 metrics
Variance reasoning notes100–200 words per off-target metric
Event signalsstructured · dispatched to Drew
Forward forecastsprogramme + per-partner
Programme health summary1pp at each checkpoint
What Mia reads
- Quinn's per-partner and per-campaign cost ledgers
- Remi's tracked URL registry, event log, enriched conversions
- Casey's partner-signal records (sentiment, advocacy, outcomes)
- PartnerStack partner registry (new partner detection)
- Salesforce pipeline + CampaignMember activity via the connector
- Approval cycle data (median days, per partner, trailing window)
- Per-partner targets and programme-level targets (kickoff)
How they interact · the closed loop
1 · Campaign loop
Drew dispatches a partner-led brief. Specialist requests a preview tracked URL from Remi. Draft produced as structured elements. Two-stage approval (client → partner). Remi swaps preview → production. Acme publishes.
2 · Attribution loop
Click on production URL → PartnerStack records partner-keyed event → S2S conversion event with deal value → Remi normalises + enriches via Salesforce → per-partner event log.
3 · Cost loop
Quinn polls PartnerStack payouts → per-partner ledger. Polls Salesforce Campaign cost fields → per-campaign ledger. Reconciliation surfaces unmapped or untagged records.
4 · Measurement loop
Mia reads Quinn's costs + Remi's enriched events + Salesforce pipeline + Casey's partner signals + approval cycle data. Recomputes every KPI + per-partner metric. Generates variance notes.
5 · Decision loop
Drew reads Mia's signals + variance reasoning. Adapts the next round of dispatches: which partners to boost, hold, fix, pause, drop. At T+ checkpoints, recommendations move from draft to approved.
6 · New partner loop
Mia watches PartnerStack registry. New partner appears → event signal to Drew → automatic launch brief (Casey story + Jordan referral push) → presented to client for approval before specialists begin.