Civic technology · applied AI · public systems

Auditable AI for complex public systems.

I’m La Shara Cordero, a court operations leader and applied AI product builder who believes in learning by doing. I design working prototypes that make evidence, decisions, and unknowns easier to inspect.

Read the field notes View selected work
Portrait of La Shara Cordero
La Shara Cordero Courtroom to codebase · California

§ 01 · Selected work

A thesis, tested in public.

These projects are the result of learning by doing—separating what a system observed from what it concluded, while keeping human authority visible.

Flagship · civic AI · readiness intelligence

Verity Lex

The AI reads the record. It never assigns the score.

A cited public record readiness surface for California superior courts. A bounded agent locates and extracts evidence. A deterministic registry calculates the result. Every claim can be checked and every unknown remains declared.

GPT 5.6 · Tavily · Next.js · deterministic TypeScript engine · recomputable audit bundle · published methods

Live product Methods Source
Companion field note · Civic AI “Open AI Build Week: Foundation First, Software Second” →

Why foundation comes before software when introducing AI to court and government systems.

Flagship · empirical product research · agent observation

Agentis Lux

Shows developers what retrieval agents experience when they read a website.

Deterministic checks, a published methodology, and a fifty site benchmark make the second audience visible. Built to inspect how autonomous agents parse structure, permissions, and latency across the modern web.

AWS Lambda · Amazon Bedrock · DynamoDB · Vercel · fifty site benchmark

Live product Source
Companion field note · The Second Audience “I built it twice. Didn't place twice. Then Google and Vercel shipped it.” →

The web has a second audience: autonomous agents. Inspecting what retrieval bots experience.

Heimdall logo
⭐ Sponsored Evaluation

AI governance · access control

Heimdall

Reproducible least-privilege test of RBAC at an AI gateway. Three scoped roles against three model tiers, nine calls, plus a falsification test and a negative control.

HARNESSBifrost Local · 3 Scoped Roles
EVALUATION9 Calls · Falsification Controls
RECEIPTAppend-Only Audit Log
Source code
“Scoping Permissions with RBAC at the AI Gateway” →

Nine calls, three refusals, and a falsification test by a court access administrator.

Clew Directive logo
Multi-Agent AWS

Education · Personalized AI learning

Clew Directive

A stateless, privacy-focused tool providing personalized learning paths through verified free resources. Formulates custom curricula based on learner goals and style.

AGENTSScout + Navigator + Curator
INFRAAWS Amplify · Nova · Lambda · CDK
PRIVACYStateless · Zero Retained Data
Live product Source code
AWS Builder Deep Dive · AIdeas “Clew Directive: Free AI Learning Navigator” →

Stop following the hype, start directing the search. Multi-agent curriculum synthesis on AWS.

Themis Lex logo
Judicial Readiness

Court service · policy alignment

Themis Lex

Role-based AI readiness guidance for court staff. Evaluates specific court workflows against sensitivity guidelines to determine where AI can safely assist and where human authority must remain.

DOMAINCalifornia Court Operations
ENGINEClaude Haiku · Bedrock · PDF
BOUNDARYHuman Authority Preserved
Live product Source code
Civic tech field note · Build Club “The Part of Themis Lex I Never Explained” →

Why California court staff need role-based AI readiness self-checks that preserve human authority.

§ 02 · Operating philosophy

Structure earns trust.

Trust is not a tone of voice layered over a product. It is the result of assigning evidence, calculation, uncertainty, and authority to the right parts of the system.

01

Evidence before conclusion

Consequential claims remain connected to their sources. Retrieval and inference are not treated as the same thing.

02

Structure before intelligence

Rules, boundaries, failure behavior, and deterministic calculations belong in the architecture before a model is invited in.

03

Human authority remains visible

AI can retrieve, extract, explain, and prepare. It should not quietly acquire the power to decide.

Unknown is a status, not a verdict.

§ 03 · Selected writing

The field notes matter.

Documenting the architecture, failure modes, real costs, and engineering decisions behind working systems.

Systems judgment 💎 DEV Community Gem · Sep 11

My Agents Never Get Tired. I Do.

Live excerpt feed Pause on hover

“I approved a prompt on a Tuesday morning and went to make a cup of Irish breakfast tea. By the time I sat back down, the agent had refactored forty-two files... My agents never get tired. I do. When generation cost drops to zero, human verification capacity becomes the bottleneck. At some point you have to know when a system is good enough—and when you’re just satisficing.”

“I approved a prompt on a Tuesday morning and went to make a cup of Irish breakfast tea. By the time I sat back down, the agent had refactored forty-two files... My agents never get tired. I do. When generation cost drops to zero, human verification capacity becomes the bottleneck. At some point you have to know when a system is good enough—and when you’re just satisficing.”

Read full field note →
Engineering craft Sep 7 · 8 min read

Future AWS Agent Engineer?

Live excerpt feed Pause on hover

“A few weeks ago I hit ReAct in the coursework and had a record scratch moment. I was building autonomous agent loops on Bedrock, but every line of plumbing was generated. If the agent writes the tools that the agent calls... does the engineer still understand the system? Yes—because authority and architecture belong to the human.”

“A few weeks ago I hit ReAct in the coursework and had a record scratch moment. I was building autonomous agent loops on Bedrock, but every line of plumbing was generated. If the agent writes the tools that the agent calls... does the engineer still understand the system? Yes—because authority and architecture belong to the human.”

Read full field note →
Systems volume 🏆 Netlify Dual Winner · Prolific & Consistent

Netlify’s Hot AR Summer

Live excerpt feed Pause on hover

“I wasn’t going to enter. 33 apps built in 31 days. One self-contained HTML file each. No build step, no dependencies, no Math.random in any selection logic. Same input, same output, with the derivation shown. The whole month cost $44, and the write-up shows the receipts.”

“I wasn’t going to enter. 33 apps built in 31 days. One self-contained HTML file each. No build step, no dependencies, no Math.random in any selection logic. Same input, same output, with the derivation shown. The whole month cost $44, and the write-up shows the receipts.”

Read full field note →
Read all writing

§ 04 · About

The courtroom came first.

The code is newer. The systems judgment is not.

2015–Present Court & Government Operations
→
2025 Started Building in Public
→
2026 Active NACM Speaker · 40+ Systems Shipped

La Shara Cordero

I spent over a decade inside court operations and government before I began building software. I started this journey with a simple philosophy: learn by doing. Rather than analyzing AI from a distance, I began building tools in 2025 to test my questions through working systems.

“Who has authority? What evidence supports the claim? What happens when the data is incomplete? Who carries the consequence when an implementation fails?”

— The four inquiries governing every Clew Labs deployment

I build through a directed, review-intensive AI workflow: I define the problem, architecture, boundary constraints, acceptance criteria, and design. AI accelerates implementation. I inspect, falsify, and verify the result.

Civic & Judicial AI

Frontline court readiness, sensitive workflow evaluation, and audit bundles.

Gateway Governance

Least-privilege RBAC, append-only provenance logs, and falsification suites.

Public Rights Engines

Plain-language deconstruction of complex notices and municipal agendas.

Advisory & Keynotes

NACM workshop presenter, technical writing, and civic AI architecture pilots.

Directed AI Systems Pipeline Statutory boundary ➔ Working audit bundle
01
Statutory Boundary

Define operational authority, evidentiary requirements, and declared unknowns first.

02
Deterministic Logic

Encode scoring registries, least-privilege RBAC, and schemas before inviting models.

03
Bounded Synthesis

Models locate and extract evidence within strict prompt contracts and verified tools.

04
Human Falsification

Adversarial edge tests, negative controls, and recomputable audit bundles for sign-off.

Systems Archive & Laboratory Record 25 Shipped Projects
Explore full repository index +
Civic & Judicial Tech Agent Governance Deterministic Audits Multi-Agent AWS Plain-Language Engines
Porch Light

Weekly plain-language synthesis of Ventura city agendas with receipts on every claim.

Charitas Clew

Public rights engine deconstructing bureaucratic notices with Gemini Flash, backed by an honest post-sprint security audit.

Nova Trivium

Amazon Bedrock AgentCore triage router directing queries across DynamoDB tools, FAQs, and human handoff.

Sobremesa

DEV Frontend Challenge entry with WCAG AA computed contrast and zero runtime dependencies.

Hot AR Summer logo
Hot AR Summer

33 apps built in 31 days during Netlify’s Agent Runner build challenge.

Janus Clew

Measures coding growth from shipped work.

Memoria Clew

Local first research memory with transparent reasoning.

Metis Clew

Plain language code learning for career switchers.

Lumen Clew

Repository insights without judgment.

Hermes Clew

The proof of concept that preceded Agentis Lux.

Anton Clew

Deterministic policy controls for AI code changes.

README Clew

Audits documentation claims against code.

Justicia Clew

Plain language jury information grounded in the court’s own public record.

Athena Clew

Autonomous debugging with reusable principles.

Argus Clew

Agent readiness review for AWS commits.

Clew Quest

Intelligent open source project discovery.

RealDoor

A renter controlled application readiness copilot for affordable housing.

Hung Jury

A finding that identified data quality as the blocker.

Steep

Tea leaf reading for code, steeped in evidence.

Vigil Crest

A challenge triage agent that hedges its judgment.

Open for Opportunities & Pilots

§ 05 · Connect

Let’s build something auditable.

Whether you are exploring public sector AI readiness, architecting least-privilege gateway controls, or looking for a product leader who bridges court operations with working systems—let’s talk.

🏛️ Civic AI Product Leadership 🛡️ AI Gateway & RBAC Governance ⚖️ Frontline Court Capacity Workshops 🔬 Responsible AI Pilots & Speaking

Direct inquiries typically answered within 24 hours.