AEON is an AI Workforce Agency — we recruit, train, deploy, and maintain autonomous AI agents for small and mid-sized businesses. Think staffing agency, but the employees are AI. We're not a platform, not a SaaS tool, not a dev shop. We're a managed service that takes accountability for agent performance, with pricing that includes a base retainer plus a performance share tied to measurable outcomes.
We're currently in active build with a pilot client (a 35-location optics retailer running Odoo ERP) and need a fractional CTO to pressure-test our architecture, accelerate the technical build, and help us get from working prototype to a repeatable deployment methodology.
What We've Built So Far
Agent architecture designed: 11 specialized AI agents (6 in Phase 1), orchestrated by a central coordinator agent (Chief of Staff) that routes tasks, manages priorities, and handles cross-agent workflows.
Operations manuals completed for each agent — detailed decision logic, autonomous powers, approval thresholds, escalation rules, daily/weekly/monthly workflows.
Multi-model architecture defined: Claude Opus/Sonnet/Haiku, GPT-4o, and Gemini 2.5 Pro assigned by function (not by cost). Model diversity is an architectural choice, not a budget one.
Brand, pricing, onboarding flow, and command center designed and spec'd.
Technical co-founder handling current architecture and integrations with another developer.
What We Need You For
We need someone who can operate at the intersection of AI engineering and systems architecture — not just advise, but get hands dirty with the team on the hard problems.
Architecture & Infrastructure
Validate and refine the orchestration layer — how the Chief of Staff agent routes tasks across 6+ specialist agents in real time.
Design the memory architecture: episodic (interaction history), semantic (domain knowledge via RAG), and procedural (workflow definitions) — per agent.
Architect the connector/integration layer with circuit breakers, fallback logic, and progressive tool access (agents start read-only, earn write access through validated performance).
Ensure multi-tenancy is right from day one — each client's data fully isolated.
AI Agent Engineering
Define model selection criteria per agent function: temperature settings, token management, context window optimization.
Build the feedback loop infrastructure — how agents evaluate outcomes, score their own performance, and adjust decision thresholds over time (the SPAR "Reflect" phase).
Design structured cross-validation protocols between agents using different model architectures.
Implement decision trails — every agent action logged with input, processing steps, and justification for full auditability.
Production Readiness
Harden the integration layer — we connect to Odoo, Stripe, QuickBooks, bank feeds (Plaid), CRM, POS, email, and Slack. Each connector is its own project with OAuth, webhooks, error handling, and rate limiting.
Build monitoring, alerting, and observability across the agent fleet.
Establish deployment methodology that's repeatable for new clients — not just one-off for our pilot.
Strategic Technical Leadership
Pressure-test architecture decisions with the team. Challenge assumptions. Find the gaps before production does.
Help us prioritize: what to build now vs. what to defer. We have 11 agents spec'd — only 6 ship in Phase 1. Engineering resources are the constraint.
Evaluate build vs. buy for orchestration frameworks (LangChain, CrewAI, custom).
The Tech Stack
AI Models: Claude Opus/Sonnet/Haiku, GPT-4o, Gemini 2.5 Pro
Orchestration: Under evaluation — this is a key decision you'll influence
Memory/Knowledge: Vector DB (Pinecone or Weaviate — TBD), Redis for episodic
Integrations: Odoo API, Plaid, Stripe, QuickBooks/Xero, Slack, email
Communication layer: Slack/Telegram (all agent outputs and alerts)
Frontend: Command center (React-based, 9 views designed)
You're a Great Fit If You Have
Built and shipped multi-agent or LLM-based systems in production (not just prototypes or demos).
Deep experience with API orchestration, integration architecture, and managing external service dependencies at scale.
Hands-on with at least two of: Claude API, OpenAI API, Gemini API. You understand model selection trade-offs, not just prompt engineering.
Experience with RAG pipelines, vector databases, and memory systems for AI applications.
Worked with SMB-facing products — you understand the constraints of businesses that don't have internal tech teams.
Strong opinions on build vs. buy for AI orchestration frameworks, backed by experience.
Comfortable working with a small, fast-moving founding team (2 co-founders + you).
Bonus Points
Odoo ERP integration experience
Experience with Plaid, Stripe, or financial API integrations
Background in multi-tenant SaaS architecture
Familiarity with agent frameworks (LangChain, CrewAI, AutoGen, etc.)
What This Is NOT
This is not an advisory-only role. We need someone who reviews code, makes architecture decisions, and pairs with our developer.
This is not a full-time CTO search (yet). We're looking for a fractional engagement — 10-15 hours/week — with the potential to grow as we scale.
This is not a greenfield project with no direction. We have detailed agent specs, a live pilot client, and a clear product vision. We need someone to accelerate execution, not define strategy from scratch.
How to Apply
Send us:
A brief note on the most complex AI/LLM system you've built or helped build — what was the architecture, what models did you use, and what broke in production.
Your take on this question: If you were building a multi-agent system where 6 AI agents need to coordinate in real time, each connected to different external APIs, what's the single most important architectural decision you'd make first?
Your availability and preferred engagement structure.
We move fast. If your experience matches, expect a technical conversation within 48 hours of applying.