Cortex

q2-2026-team-roadmaps

9 min read

Q2 2026 Team Roadmaps

Summary

Synthesized from all Q2 kickoff meetings (Apr 9–21, 2026). Eight product teams presented Q1 results and Q2 plans. Common threads: June 1st hard deadline driving focus, safety emerging as lead category post-June, AI form fill as cross-team priority, and feedback loops / agent-based monitoring becoming standard practice.

Key Points

Company-wide themes

  • June 1st launch is the forcing function across all teams — Agent V1, form fill, MCP app stores, Gearbelt work orders, mobile releases
  • Safety-first pivot post-June: 80% of PDE resources reallocating to safety as “king category”
  • Feedback loops everywhere: Uber-style UX ratings for AI features, agent-based triage of feedback (50+ pieces/day on AI templates alone), MCP intent-vs-outcome tracking
  • All three GTM regions missed Q1 expectations (Oceania orange, EU strong red, NA red with March bounce-back)
  • Potential for first back-to-back MoM lead growth in almost 2 years

Dashpivot (Gayuru, Peggy, Steph, Alvaro + team)

  • Q1 wins: location field, access module export, actions module, counter field (15% adoption), table header colors (1 in 5 table editors)
  • Metrics: forms created 15% growth, actions usage 63% surge, NPS 36 → 42
  • Q2 focus: weather field, photo support, relationship field (linking forms), team-level actions, safety features
  • Customer signal: 45 min saved per report generation, critical need for relationship field to link NCRs within checklists
  • Strategic: self-improving systems with coding agents reviewing feedback, roadmap flexibility needed post-Q3

AI Platform (Nick, Steven, Fariha, Cai Li + team)

  • Q1 wins: AI templates launched with GTM, prompted generation, form fill private beta
  • Q1 miss: hiring targets (freeze to focus on June 1st)
  • Recognized as tightest execution team in Q1
  • Form fill insights: multiple fill methods needed (current for diaries/reports, different approach for checklists, hands-free for compliance)
  • Beta feedback: speed confirmed but trust issues from content deletion — users need visibility into what AI changed
  • Non-field users expect broader AI beyond form filling (workspace data querying → MCP opportunity)
  • Q2 focus: expand context window, additional file types, POC for guided/conversational fill
  • Late May team session to move from preset roadmaps to flexible vision pathways

Mobile (Hannah, Cameron, Shaun + team)

  • Q1 wins: Sitemate ID scanning, counter field, Gable integration shipped
  • Q1 miss: Sitemate web (no engineer due to hiring freeze, likely Q3)
  • Actions usage only 13% on mobile vs web — buried in profile settings, nav bar revamp should fix
  • Android crash rate warnings from Google Play — risk to discoverability
  • Q2 focus: camera uploads, search/filtering, background uploading, AI form fill (voice priority), push notifications scoping
  • Hard deadline April 22nd for 26.4 release
  • React native upgrade enabling Q3 mono repo transition

Global Integrations (Ibrahim, Phong, John An + team)

  • Q1 wins: MCP apps beta for ChatGPT/Claude/Copilot, Retool apps, Sitemate MCP live (last blocker for app stores)
  • ARR crossed $1.5M despite no new Flow Site integrations
  • MCP usage extremely low but early — onboarding is the gap, not interest
  • Q2 focus: MCP in app stores, packaging V1 (account-level add-ons), Flow Site freemium
  • Packaging vision: account-level toggles (Open API, MCP, exports), Sitemate Start + Flow Site UI merger, Global Data module for enterprise
  • Technical: Open API VTL tech debt approaching hard size limits, contract testing framework (76 tests across 37-38 endpoints), MCP data structure redesign for LLM efficiency (120k tokens choking ChatGPT)

Global Platform (Jamie Van Stone, John Rey, Kenzo + team)

  • Q1 wins: workspace switcher V1, account layer UI, user export, tablet improvements, MCP permissions
  • User export killed Windmill exports (dropped to zero in April)
  • Q2 focus: account layer as billable Sitemate Start V1 feature (first revenue-generating add-on for the team)
  • Top customer pain points: granular permissions (between member/controller), global user management across workspaces, moving forms between teams (70% of Windmill usage)
  • Moving forms design: single form initially, org-level templates only, permission-tiered access

Gearbelt (Will, Kierian, Glen, Karjack + team)

  • Q1 wins: schedules refactor, Sitemate mobile integration (online/offline), issues refactor
  • MoM growth 4.4% → 5.1%, churn improved 0.9% → 0.6%
  • Sprint completion at 50-60% — product backlog becoming bottleneck (tickets prepared day before delivery)
  • Hired 2 engineers (Negan, Kelly) for Q2
  • Q2 focus: Gearbelt Mobile (mechanics CRUD workflow), work orders refactor with Kanban view, editable templates
  • Work orders becoming core of maintenance workflow — moving from read-only to full CRUD
  • Future (Q3+): costing, inventory management, telematics integration

Dashpivot Platform (Edward Ko, Jenny Zhou, Cerino, Jeff Shillitto, Tim Edwards, Oliver Dunn + team)

  • Q1 wins: front-end component refactor, search item ordering, list filtering, deployment features, legacy nav migration for Dashpivot V3, middle/auth for team-level lists
  • GraphQL endpoints nearly complete — dependency for account layer functionality and team-level lists roadmap
  • Roadmap adjusted mid-quarter to support account-level lists for Dune One (prevented additional tech debt)
  • Usage trends: list page visits +30% vs 6 months ago (peaked March), list deployment +14% YoY, list creation +30% vs April 2025, list property table cells +42% vs 6 months (6x vs 12 months)
  • Template analysis: yes/no checkboxes dominate, multi-sigs > single sigs, median 17 fields/template (outlier at 3,400); Q2 targets templates with 200+ fields
  • Customer priorities: child list property uploads, PDF attachments in list property tables, teams/folders for list organization, list-template relationship visibility
  • Q2 focus: Lists (temporal dates, GraphQL completion, property expiry, account-level) · Templates (form loading, field/logic limits, sections/pages) · Performance (table cell lazy-load)
  • Key concepts in development: list view sorting, child list editing (explicit enable), manual parent-child syncing V1, text field character limits (found 8M-char rich text field), team-level lists module
  • New hire: Keen starting as PDM in Melbourne
  • AI platform feedback expansion: star ratings, agent-reviewed feedback with automated fixes, UX flow-based triggers (vs time-based), team-configurable trigger points

Growth (Hartley, Victoria, Renata, Eric, Myron + team)

  • Q1 wins: sign-up restart trials, user invites during setup, phone formatting, Google indexing fix for sign-up/login, watch demo revamp, UTM params for Gearbox tracking, CSS auto-versioning, company logo upload via templates editor
  • Metrics: sign-up conversion held 25–28% despite declining page visits; activation (5 forms in 30 days) stable at 15–17%
  • Q1 misses: social sign-up delayed (Sitemate auth layer complexity); team departures (Anton, trunk transitions) caused product decision delays
  • Core Q2 problem: 63% drop-off between sign-up and first form creation — users must navigate folders, find templates, create forms before getting value
  • Q2 New User Experience project:
    1. Single folder restriction for new users (no workspace switching)
    2. Pre-loaded default templates by vertical/category
    3. New home screen with action-driven shortcuts
    4. Recent templates section showing vertical-relevant options
  • Design considerations: toggle for enterprise users needing folder structure, Sitemate Chef workspaces need advanced structure by default, potential competency-based unlock system, dynamic checklist replacing trial guide
  • Immediate releases: social sign-up launch (pending Victoria’s return), company logo removal from sign-up flow
  • Approach: loose design direction over pixel-perfect specs, engineer freedom to build/test/adjust, early prototyping, consider enterprise/POC account requirements in feature toggles

Cross-team dependencies

  • AI form fill touches: AI Platform (core), Mobile (voice/photo input), Dashpivot (web integration)
  • Account layer: Global Platform builds it, Global Integrations packages it, all teams affected
  • MCP: Global Integrations owns, AI Platform and Dashpivot consume
  • June 1st: all teams must coordinate completion for marketing launch

Updated timeline & milestones (2026-05-05 Managers Allhands)

  • Week of May 11: internal launch preview
  • Week of May 18: GTM training week — not full-day sessions; one overall briefing + SDR/AE/CSM deep dives + per-rep mocks. Customer meetings stay open.
  • May 12 (Tuesday): Tim lands in AU for ~4 weeks (2 weeks lead-up + 2 post-launch). No agenda beyond launch presence and feedback loops.
  • End of May: first 2–3 internal agents online per pillar (marketing, revenue ops, PDE). Marketing has 3 agents targeted by month-end.
  • End of this week (~May 8): Aaron ships first “fix agent” within AI Platform team, then hands to Sam → AI platform / world-knowledge engineering team.
  • June 1: 1.0 launch (form fill + AI templates). No direct commercial changes; Sitemate Start launches ~end of first week of June as a billable item.
  • June Managers Allhands: locks in team shuffle/reorg + which function 2.0 doubles down on (likely safety).
  • July–September: 2.0 build (digital employee/functional agent). Extra resourcing into AI Platform team.
  • October: 2.0 launch — likely larger commercial changes.

Marketing/Growth update (post-Q2 kickoff)

  • Tracking to 5 consecutive months of MoM lead growth for the first time in 2+ years (could be 3+ years).
  • Volume: ~5K leads/mo (close to 5.5K). Website ~1K, generators ~750.
  • Eric running marketing team; Renata managing growth engineers.
  • Two new levers coming online: (1) new organic content production engine, (2) marketing ops = programmatic warm outbound. Triggered on events (e.g. Bob signs up for a generator → identify his company, role, safety/quality/MD stakeholders → tailored outbound). Multiplier on top of volume.
  • ~30–40% of generator leads currently not routed to GTM (Gmail-only, low quality) — marketing ops should fix this.
  • Quality lift through funnel: enterprise reps will be able to Slack-request custom marketing videos (Sitemate Chef-style UX) for specific accounts using generative video.
  • Reset story: in Dec 2025, growth team was optimising a decaying engine. Required workflow rebuild, not team changes — re-engineered pages for AEO/SEO, new operating principles. No IC-level changes were needed.

Internal agents rollout (osmosis pattern)

  • AI Platform team = guinea pig for the agent build pattern.
  • Each pillar (marketing, revenue ops, ops team, PDE) gets 2–3 agents online by end of May.
  • Named agent owners: Aaron — deck infrastructure agent (Sitemate Chef-style custom decks). Jordan — core review agents.
  • Pattern: feedback stream → listeners running → extract signal → autonomous system handles low-level work, loops human in where needed.
  • Centralised AI ops function as nucleus → spreads to teams. End-state: each team has an engineer responsible for managing 1–2 agents’ output alongside their own work.
  • Gearbelt is the second team adopting this (after AI Platform) — going live with autonomous noise-feed listeners.
  • Prediction (end of 2026): end-to-end loop where Slack feedback → review agent scopes → fix agent generates options → designer picks one → PR → prod, for small frontend-only features. “Bet a very large amount of money this happens at least once before end of year.”

Four safety agent features (post-June roadmap)

  1. Induct — conversational induction review and grading
  2. Review — document review via email with iterative feedback
  3. Audit — continuous monitoring vs annual auditing (easiest, lowest value)
  4. Overwatch — real-time site monitoring using photos, scan-ons, location data

Sources

Last Updated

2026-05-06

Graph View

Backlinks