hebel. today · 5 November 202615 / 74

Career · Product Management Evolution

Product teams shift from builders to aggregators of verified context

Aggregating enterprise knowledge. PostHog manages 10 AI products serving 100,000 monthly users by prioritizing agent reliability over feature expansion. Users of PostHog’s install wizard send their first event in 1.9 hours, compared to 3.8…

Aggregating enterprise knowledge. PostHog manages 10 AI products serving 100,000 monthly users by prioritizing agent reliability over feature expansion (Ruben Dominguez from The AI Corner). Users of PostHog’s install wizard send their first event in 1.9 hours, compared to 3.8 hours for those who do not, while conversion to paid plans reaches 14.2% against 2.6% for non-users (Ruben Dominguez from The AI Corner). Data leaders at Modern Data 101 define this as the aggregator role, where teams frame business questions, filter data sources, and ground models in first-party signals to maintain system accuracy (Modern Data 101, Modern Data 101).

Automating quality assurance. Anthropic engineers spend 80% of agent session time waiting for human review, leading the company to replace manual oversight with automated harnesses and evaluation metrics (Supercompanies with Greg & Taylor). ElevenLabs provides a technical template for these workflows, tasking developers with testing for latency, interruption, and language switching during agent builds (Martyn Redstone from Own The Conversation). Product teams at ING Madrid now require conversational designers to show evidence of how customer data informs release decisions rather than relying on dialogue design alone.

Refining developer-led roadmaps. Anthropic maintains a ratio of 17 engineers per product manager, delegating feature decisions directly to the engineers building the agent fleets. Developers at PostHog utilize an MCP server architecture specifically for technical users, while reserving in-app agent placement for non-technical audiences. Addy Osmani suggests that teams reduce prompt complexity by defining success conditions and granting models access to verification tools, allowing the system to determine its own reasoning path (Shilpa Mitra from WebAfterAI).

Try today · 30 min

Audit one conversational action in your product portfolio to confirm it functions as a workflow.

List the system of record, required permissions, and audit evidence for the action.

You need

  • A list of your product's current conversational AI actions
  • Access to the system of record for each action
  • Documentation of permission requirements

Steps

  1. Select one conversational action your users perform.
  2. Identify the specific system of record that stores the action's data.
  3. Document the permissions required to execute the action.
  4. List the audit evidence generated when the action completes.
  5. Define the failure route if the model or connected system is unavailable.
LinkedIn X

Sources

  1. Ruben Dominguez, The AI Corner ✎ Written by a personPostHog shares 38 lessons from shipping 10 AI products, emphasizing the performance difference between users who adopt AI wizards and those who do not.

  2. Martyn Redstone, Own The Conversation ✎ Written by a personMartyn Redstone outlines how conversational AI integrates into existing enterprise workflows and provides a checklist for auditing agent-led actions.

  3. Modern Data 101Animesh Kumar explains the aggregator role, focusing on how human-grounded data prevents model drift and maintains accuracy in automated systems.

  4. Supercompanies with Greg & TaylorGreg and Taylor analyze Anthropic's shift toward automated evaluation harnesses to reduce human bottlenecks in agentic engineering.

  5. Cory Blumenfeld, Taking My Time Back ✎ Written by a personCory Blumenfeld provides a manual-first process for identifying problems before implementing AI automation to avoid multiplying existing inefficiencies.

  6. Shilpa Mitra, WebAfterAI ✎ Written by a personShilpa Mitra discusses reducing prompt debt by moving from complex instructions to simple objectives and clear success conditions.

Get this every morning, for your goal.

Tell Hebel your goals and when you want it. Every morning it reads 20,000 newsletters and writes your paper, with one small thing to do today. Free during beta.

Start my paper →

More in Career

Proximity and rhythm define professional growth

Build your professional reputation through evidence of impact

Stacy Muur builds a marketing agency by testing ideas in public

Managers shift from dictating methods to defining outcomes

Process consistency versus output optimization

Elon Musk envisions a post-labor economy driven by humanoid robotics

Product Management Evolution yesterday

AI agents demand deterministic foundations over total rewrites

Sign in to Hebel first

Likes, your collection and done Hebel need an account. We email you a six digit code; no password.