Siew Ling Wong · Engineering Lead · Media / Agency · SMB · 60 employees
Rentak Media is a 60-person agency buying a roughly 14-seat engineers-only cohort (RM 53,000 proposal, procurement form out 4 days), and the deal record carries no evidence of an existing AI tool standard, licence footprint, or agent guardrail policy. Those are the two biggest gaps: Day 1 would land into a vacuum of unmanaged editors and ad hoc prompting, and Day 2's agentic-development content has no PR-review convention or blast-radius rule to attach to. The highest-leverage 90-day move is a one-week pre-workshop audit that inventories Cursor/Claude Code/Copilot licences, secrets handling, and a cycle-time baseline, then run Day 1 and Day 2 back-to-back for the full cohort and ship a written shared-rules plus agent-allowlist policy within 30 days, with weekly active-usage and escaped-bug tracking to justify a second cohort.
CRM shows a RM 53,000 proposal (about 14 seats at RM 3,500 plus 8% SST) but zero evidenced Cursor, Claude Code, or Copilot licences, MCP servers, or IDE rule files — Day 1's hands-on labs assume a stack that has not yet been confirmed to exist. The score sits at at_risk because intent is real (procurement engaged) while the environment Day 1 depends on is unverified; it drops to critical if the pre-workshop audit finds no licences in place.
The 'Engineers-only cohort' tag plus the complete absence of any PR-review convention, prompt-hygiene standard, or agent guardrail in the deal record means Day 2's agentic-development module currently has nothing to anchor to. For a media agency working to client deadlines, the default failure mode is review discipline collapsing under delivery pressure. Score is critical because the workshop must first teach a baseline convention before it can teach AI woven into it.
The Engineering Lead, Siew Ling Wong, is the named champion and has carried the deal to Proposal with procurement already engaged — a genuine positive signal of leadership appetite. But the cohort is engineers-only with no manager or leadership track, so workflow change rests on one person, and blast-radius tolerance for autonomous agents is entirely unrecorded. Score at_risk: sponsorship exists but is single-threaded.
No cycle time, PRs per dev per week, or deploy frequency appears anywhere in the deal record — the only throughput signal is the roughly 14-seat single cohort, implying an engineering team small enough that one workshop reaches most of it. The score is deliberately mid-band: unmeasured, not proven poor. The pre-workshop baseline is what lets the workshop prove a gap rather than assert one.
The procurement form went out 4 days ago, which means AI seats are not yet deployed — current daily and weekly usage is effectively zero-by-default and unmeasured across sub-teams. With 60 total employees and a ~14-seat rollout, one cohort covers a large share of the engineering org, so utilisation will be highly visible within 30 days either way. Score critical because nothing has shipped yet, not because demand is absent.
Deal health sits at 62/100 and the record supplies no defect rate, escaped-bug count, time-to-feature, or dev-satisfaction data — for a media agency the real exposure is client-facing rework rather than internal bug counts. Score at_risk because the workshop's value is unprovable until these are measured pre- and post-engagement, and the 90-day follow-up should own that instrumentation.
No proposals yet. The AI writes one from the readiness check.