I design and ship agentic workflows using LangChain, LangGraph, and AutoGen to automate complex processes with guardrails, observability, and human oversight. Years of hands‑on delivery means we push boundaries without burning risk budgets.
MakerCore delivers applied AI for real operations. My background spans accounting flows, payroll, billing, and data integrations — now extended with agentic automation, retrieval, and tool use. I also build real‑time media applications: WebRTC calling, live streaming, and on‑the‑fly transcoding for web and native. I translate business goals into reliable systems with clear guardrails, measurable outcomes, and a clean path to production.
Orchestrate tool‑using LLM agents with deterministic fallbacks and audit trails.
Evals, monitoring, prompt/version control, and cost/latency discipline.
Keep people in control where judgment matters; automate the rest.
Accelerate bookkeeping with agent workflows, deterministic checks, and full auditability.
Schedules, compliance checks, and approvals with human‑in‑the‑loop gates.
Usage summarization, invoice explanations, and proactive exception handling.
QuickBooks, banking, and internal APIs; retrieval and tool adapters for agents.
WebRTC, live streaming, and low‑latency transcoding pipelines for web and native.
An approachable marketplace for creators; lessons carried into onboarding and UX flows.
Exploring multi‑agent coordination, tool adapters, and evaluation harnesses.
Tell me the outcome you want. I’ll design the path.
I’m building practical, production‑grade tools that make finance, ops, and growth run themselves — starting with bookkeeping automation and sales analytics, powered by agentic workflows with human oversight.
The next decade belongs to agentic systems: software that can reason, act, and integrate across tools. Organizations don’t want demos — they want outcomes with guardrails. MakerCore ships those outcomes: measurable automation that plugs into finance stacks, CRM, and data lakes with auditable trails, evals, and cost discipline.
A reconciliation and close assistant that handles categorization, matching, exception routing, and narratives — built with LangGraph for deterministic paths and human‑in‑the‑loop checkpoints. Faster month‑end, tighter cash visibility, full auditability.
A decision engine that turns raw go‑to‑market data into live funnels, cohort quality signals, and action suggestions. Think operator‑grade insights: pipeline health, pricing impacts, and prioritized next actions — no more spreadsheet archaeology.
Opinionated patterns: tool‑using agents with deterministic fallbacks, retrieval with confidence thresholds, cost/latency budgets, evals + monitoring, and simple UIs that keep humans in control where judgment matters.
Quote‑to‑cash checks, anomaly alerts, and playbooks.
Policy routing, exception narratives, and evidence packets.
Onboarding flows, payroll validations, and retention signals.
Document triage, status generation, and structured summaries.
RFP drafting aids, criteria checks, and vendor comparisons.
Readable analytics on services usage and impact.
A practical agentic layer that helps councils and city personnel run a city from agenda to action. Clerk‑ and council‑ready assistants compile packets, check policy conflicts, and keep 311 moving — all with human approvals, citations, and audit trails. Start with agenda/packet and service ops, then expand to policy, procurement, and permitting.
Operate from agenda → action with guardrails, audit logs, and clear KPIs across departments.
Agenda/Packet Assistant, Policy Copilot, 311/Service Ops, Permitting/Procurement, Open Data + Transparency.
Role agents (Clerk, Council, City Manager, Dept. Head) with tool permissions and deterministic fallbacks.
If you invest in the systems layer — automation with guardrails and clear ROI — let’s talk. Share your thesis, typical ticket, and timeline; I’ll reply with a concrete plan and demo.
We use all information collected on this site solely for our business operations and do not share it with third parties. If you have questions about this privacy policy, please contact us.