Cortex

pde-leadership

8 min read

PDE Leadership & Performance

Summary

Tim’s Q1 2026 review cycle with CEO. Core finding: two competing priority systems (product roadmap vs CTO-priority work) cause misalignment and low urgency. CEO follow-on review reinforces four themes: execution system control, PGM performance management, leadership directness, and CTO direct leverage through targeted spikes.

Key Points

Four CEO themes (follow-on review)

  1. PDE execution system & delivery management — need stronger end-to-end control: tighter roadmap awareness, clearer priority flow, less slippage. Critical technical work gets deprioritised when outside main roadmap.
  2. Performance management of PGMs — more direct: clearer expectations, explicit challenge where output below benchmark, stronger accountability. PDE performance only as strong as standard applied to PGMs.
  3. Leadership clarity, directness, decisiveness — avoid elasticity, peacekeeping, admin/chatter. Sharper judgement on what matters, what doesn’t, what happens next.
  4. CTO direct leverage through targeted spikes — 2–3 big unlocks per quarter. CTO role cannot drift into coordination/admin in AI era — needs direct leverage on architecture, agentic engineering, dev velocity.

Two priority systems → one roadmap

  • Product roadmap is the primary system — CTO-priority work (KPIs, type system, verification infra, tech debt, agentic enablement) enters from the side and always loses
  • Fix: merge into a single roadmap so nothing enters “from the side”

Performance management shift

  • Move from passive to explicit: detailed weekly/fortnightly reviews with PGMs
  • Each PGM must report how they’re paying down tech debt and making progress
  • Slow progress highlighted by reporting itself
  • Weekly report from PGMs on product & tech delivery

Recent examples cited by CEO

  • Global Integrations — not aligned to roadmap dates on follow-on release, no comms/coordination, needed prompting via Tammy
  • QA / Angus — unclear improvement curve, not clear if foundational QA infra actually improving with measurable impact
  • Mobile — hangover critical top 1–2 issues from last quarter, needed CEO prompting
  • KPIs project — 80% done but has it been communicated/coordinated with Biz Ops for standard reporting?
  • PDE Talent sessions — not always deciding what makes most sense for business, tries to keep everyone happy/fair

Talent decisions

  • Be more proactive on underperformers — pattern established post-Ulan
  • Restructure Talent Engine meetings: reduce admin/length/frequency, increase structure + actions

CTO time allocation

  • Run 2-week time audit: (a) CTO work, (b) can delegate, (c) can drop
  • Reduce meeting + 1:1 load
  • Enablement team now set up with Jas as CoS

Structural items

  • PDE org structure approaching ceiling — assess future growth structure
  • Tech Lead role reframed — largest impact on Dashpivot
  • Register view changes need to move into a stream (likely Field Reporting team)
  • Agentic engineering under-resourced vs industry benchmarks — solution is internal reallocation, not headcount

Q1 execution review (from Q2 Leadership Kickoff)

  • Dates & milestones approach delivered strong results: 13-14 hard deadlines set in January, teams rose to the occasion
  • Biggest value: forcing focus and constraining scope creep
  • Challenge: platform degradation risk — “we get slower & the platform gets worse over time”
  • Mixed team performance: AI Platform tightest execution, Global Integrations ahead on Claude but behind on comms
  • 80% of KPIs in production, last 20% pushed to Q2
  • Management structure needs refinement — the problem is coordination overhead, not engineering

Q2 execution plan

  • Focus exclusively on June 1st deadlines — minimize mid-quarter complexity
  • Radical simplification of mid-quarter reviews
  • Segment rebounding + full-cycle motion changes launching May 1st
  • Capital fundraising to bolster resources for year-end

AI transformation journey — PDE change management

Chronological record of how PDE moved from grassroots ChatGPT use to agentic engineering as a named strategy, across a remote, distributed team of 70+.

Timeline

  • Late 2022 – 2023: grassroots ChatGPT use, no company plan. Company-wide access should have come sooner.
  • Jun 15, 2025 — Hartley’s blog “On How Sitemate Ships Now, In The Era of AI” published day before rollout
  • Jun 16, 2025 — Cursor + Claude Code licences rolled out to whole team; Nick L led documentation and tech talks
  • Jun – Sep 2025 — full adoption curve observed: innovators → early adopters → laggards came on board faster than expected
  • Oct 10, 2025 — pde-ai-guild formed (Matheus); weekly-ship cadence introduced in late 2025
  • Oct 2025 — Phuket offsite hackathon: dedicated non-production time for every engineer to experiment with AI. Step-change in adoption — moved the org from “some engineers using AI” to “every engineer engaged and excited”; for a remote team the annual offsite is a rare concentrated learning window
  • Late 2025 – early 2026 — guild members ran synchronous sessions inside their teams; everyone saw best practice from someone they worked with
  • Nov 2025 – Feb 2026 — skills explosion; broad experimentation, but PDE Pulse showed engineers overwhelmed, wanting direction
  • Dec 2025 — Multitudes rolled out as first velocity measurement tool
  • Feb 20, 2026 — Hartley’s blog “Waterfall Whitewash, Agile Now An Anchor, A Forward Deployed Future”; shifted conversation from “experimenting” to “delivery framework”
  • Feb 2026 — PDE Metrics replaced Multitudes (drill-down to individual level)
  • Feb – Mar 2026 — written agentic workflow standard + Dev Box vision published (~1 month late vs when team asked)
  • Mar 25, 2026 — Tim’s blog “Agentic Engineering: The System Is the Advantage”; announced to team Mar 30
  • Mar 30, 2026 — PDE Enablement team stood up (Jas leading as CoS CTO first project): Cloud & Reliability + DevEx & AI streams

Patterns that worked

  • Blog-driven strategy at inflection points. Each of the three Hartley/Tim blogs aligned the org at a key moment. Repeatable pattern.
  • Explicit artefacts over osmosis. Remote/distributed team means learning-by-watching-the-next-desk doesn’t exist — every best practice must be deliberately broadcast (tech talks, demos, playbooks, written standards).
  • Deliberate rollout compresses the laggard tail. Access + blog + guild + team demos moved resistant engineers far sooner than expected.
  • Guild as uncertainty absorber. Curious engineers explored together, shared wins, took excitement back to teams.
  • Weekly-ship cadence. Forced the guild to stop perfecting skills before shipping — step-change in value delivered.
  • In-team demos by peers. Everyone saw best practice from someone they worked with, not a central evangelist.

Gaps and learnings

  • No AI-usage measurement yet. Both Multitudes and PDE Metrics measure velocity but not AI usage. Correlation between AI use and velocity remains the missing loop — directly blocks being able to say “invest more in X because it works.”
  • Admin overhead held the CTO back. Transformation work brought a lot of Ops load (coordination, comms, rollout logistics, vendor/licensing, guild running). With no PDE Ops function, it landed on the senior team (CTO, tech leads, staff eng) — the most expensive, highest-leverage technical capacity in the org, and the wrong place for it. Missing role was CTO chief of staff / PDE ops; filled Mar 30, 2026 with Enablement team + Jas as CoS.
  • DevX deprioritised in V4 restructure. Should have been stood up earlier — now compounding fast via Enablement.
  • Agentic workflow doc + Dev Box vision shipped ~1 month late vs when team was asking for direction — noise/skill-explosion period could have been shorter with earlier direction.

Ongoing disciplines

  • Leaders must stay in the tools — can’t delegate curiosity in a space moving this fast
  • Regular checkpoints with peers at other companies cut through the noise of public commentary; worth doing regularly, not ad hoc

Relevance for CEO themes above: this is direct evidence of theme #4 (CTO direct leverage through targeted spikes) — the 3 blog posts and the Enablement team stand-up are exactly the kind of 2-3 big unlocks per quarter the CEO review asked for. Also shows that foundational dev-infra work (pillar 2) and measurement (AI usage) remain the open gaps.

CEO direction signals (2026-05-05 Managers Allhands)

  • PDE = most AI-affected function in Hartley’s mental model — biggest steepest impact, ahead of operational roles and well ahead of GTM. Reinforces theme #4 (CTO direct leverage on agentic engineering).
  • Single-function focus from Q3 (~70% of PDE behind one function — likely safety) is a top-down call from Hartley, not a PDE-internal one. CTO theme: pick the function, build the agent + the SaaS rails it needs in parallel, 3–6–9 months before moving to the next.
  • Resourcing extra into AI Platform team during the 2.0 build (Jul–Sep) confirmed at the all-hands. Implies team-shuffle decisions land in the June Managers Allhands.
  • No risk of wasted Q3 design work — Hartley told Gayuru the dash pivot category work is ~95% likely to align with whichever function is picked. Filter narrows things, not throws them out. Action: keep Q2/early-Q3 work in small chunks so any mid-stream pivot can land cleanly.

Internal agents — leadership accountability

  • The “centralised AI ops nucleus → spread to each team” rollout (osmosis pattern, see Agentic Engineering Strategy) is now an explicit company-level expectation, not just a CTO bet.
  • End-state messaged at all-hands: each team has an engineer managing 1–2 agents’ output alongside their own work. Each PGM should be tracking this for their team.
  • Concrete near-term proof points the CEO is watching: Aaron’s first fix agent (week of May 8), Gearbelt autonomous noise-feed listeners (Will/Carl), 2–3 PDE agents online by end of May.

Sources

Last Updated

2026-05-06

Graph View

Backlinks