Meeting Notes · Thursday, May 28, 2026
With Andrew Gordon Sharpened outreach toward org-focused positioning ahead of the Kia call. Andrew's experience across four orgs surfaced a candidate wedge: standardized skill management for teams running rogue with Claude.
The meeting focused on sharpening Aileron’s outreach and customer discovery ahead of the Kia call on Friday. The core problem: current outreach reads as generic “AI for business” and fails to signal Aileron’s organizational focus — rolling out capabilities across teams, standardizing skills, providing governance (approvals, audit, deterministic execution). The team needs differentiated messaging that earns a conversation without breaking Mom Test principles.
On customer discovery strategy, the bias is toward speed over correctness. The first iteration teaches the second; both validation and rejection are valuable outcomes. The goal is unbiased feedback — even when it contradicts the current business plan — and finding customers who need the solution more than their money.
A potential wedge emerged from Andrew’s experience across four organizations: corporate teams are running rogue with Claude skills on individual instances, lacking standardization, compliance, or version control. API keys are scattered, installations are inconsistent, shared Git folders are filling in for proper tooling. This is a “Homebrew for skills” opportunity. Aileron’s composable architecture — Connectors → Actions → Skills — maps directly to that gap, and the positioning is Docker-analogous: not inventing the primitives, but making them consumable.
What we need out of this call
Two parallel asks: tighten the outreach so it signals organizational focus, and tighten the interview questions so they probe the right substance without leading.
Outreach adaptations
Reads like “using AI for business.” The recipient has no anchor — nothing tells them what specifically this is about, why it would matter to them, or why they should engage.
- Rolling out capabilities across teams
- Sharing skills between teams for efficiency
- Maximizing efficiency and creativity for people
Tight anchor that identifies what this is specifically about — without pitching.
Interview question focus
Conversations should center on organizational AI acceleration — not generic “are you using AI” questions. Specifically:
- Approvals and approval workflows
- Human-in-the-loop where necessary
- Remote timing of agents and triggers
- Auditing and proof of control
- Deterministic actions — agents express intent; execution happens at another layer
- Sharing of capabilities/skills across teams
Tomorrow’s Kia call is the first test case.
Customer discovery strategy
Better to be fast than correct. The first iteration teaches you how to do the second call. Both potential outcomes are valuable:
- Validation — customers recognize the pain and want a solution
- Rejection — learn to pivot in 2 months instead of investing 5 years
What we need is unbiased feedback, even if it contradicts the current business plan. We’re looking for customers who need the solution more than they need their money.
The anchor problem
The current approach feels like a “massive step back” — no earned right to play. We’ve peeled back the customer-discovery framing so far that the pitch now sounds like nothing.
We need a one-sentence description of the unique advantage — concrete enough that a VC would understand it, restrained enough that it doesn’t violate the Mom Test. The balance is: don’t pitch, but have a substantive conversation starter.
Technical architecture insights
The most useful thread of the conversation. Andrew has watched four organizations attempt similar solutions internally, and the pattern is consistent.
- Corporate teams building skills on individual Claude instances — no shared substrate
- No standardization, no compliance posture, no deterministic workflows
- Shared Git folders filling in for proper tooling (and failing at it)
- API keys scattered across machines
- Inconsistent installations across teammates
- No version control, no security review, no compliance signoff
The wedge: “Homebrew for skills.” Standardized skill management as the entry point.
Aileron’s technical approach
The composable primitives stack from the bottom up:
Integrations with services. Google Drive, Gmail, Jira.
Specific capabilities. Send email, create doc, file ticket.
Composed from Actions with declared dependencies.
Goals across the stack:
- Erase complexity, not replace chatbots. The product is the substrate, not the surface.
- Shift workloads to the appropriate model for cost management.
- Abstract away agents. Expose a simple task interface.
- Most organizational tasks are constrained inputs → simple process → known outputs — not what frontier LLMs are necessary for.
Market positioning
The Docker analogy. Docker didn’t invent containers; it made them consumable. Aileron isn’t inventing connectors or actions or skills — it’s making them consumable across an organization.
Adjacent trends supporting the thesis:
- Local LLMs becoming applicable — running 6–8 months behind frontier capability, but adequate for most org-level tasks
- Many tasks don’t need frontier models — classification, summarization, routing
- Token costs are running away for organizations — incentive to shift workloads downward
The positioning isn’t “build a better chatbot” or “replace your AI consultancy.” It’s: erase the complexity that’s blocking adoption.
Next steps
A second variant focused on organizational workflow value — to A/B against the current generic version.
First test case for both messaging variants. Explore the workflow questions live — approvals, sharing, audit.
Track which variant earns replies and which earns substantive conversations. They may not be the same.
Debrief Kia, debrief the messaging A/B, lock the anchor sentence.