Trust

How we build AI responsibly, how we use AI in our own work, and what the AI in TessereX does with your data.

Responsibility & Sustainability

Humagined is committed to building AI responsibly and to using it in ways that reduce waste, for our clients and in our own operations. We guide our AI practices by ISO/IEC 42001, and our founder holds the certification. We right-size the compute we use, apply the same standards to every person on our team, and give our skills to nonprofits and our community. Our largest impact is the work itself: helping manufacturers and logistics companies apply lean and standard work so they produce less waste.

  1. 1.Responsible AI

    Humagined builds and deploys AI under a governance standard, with people accountable for what ships.

    • AI governance guided by ISO/IEC 42001.
    • Published AI diligence statements for humagined.ai and the TessereX platform.
    • Manual review of all code before it ships, alongside internal SOC 2-style security reviews and codebase scans.
    • Client AI use cases documented and ranked before we build, so AI goes where it creates value and risk is understood up front.
    • Client data used only for the engagement it was shared for.

  2. 2.Compute and footprint

    Our largest environmental footprint is compute, so we design to use less of it.

    • The smallest model that does the job, including small CPU-hosted models for small manufacturers.
    • Repeatable questions answered in reporting, with AI reserved for the questions that need it.
    • The same client data is never ingested twice, which avoids redundant processing and storage.
    • AI token usage tracked per engagement beginning in 2027.

  3. 3.Community

    We put our skills to work beyond our clients.

    • Humagined for Good: AI, data, and tool working sessions for nonprofit organizations.
    • Volunteer teaching with Junior Achievement Rocky Mountain.
    • Colorado policy work through the Good Business Colorado Policy Council.

  4. 4.People

    Everyone who works on a Humagined engagement, onshore or offshore, is held to the same standards.

    • The same security, code review, and data handling standards for every team member and subcontractor.
    • Client data access limited to the people working on that engagement.
    • Team capability in AI tools and methods built as part of the work.

  5. 5.Client outcomes

    Our biggest sustainability impact is through our clients.

    • Standard Work Before AI Work: lean at every stage of an engagement.
    • Fewer defects, less rework, less excess inventory, and less wasted energy, through shared definitions and reporting that drives continuous improvement.
    • Every engagement stated in outcomes, not just outputs, and reported against them.
    • Clients own their systems: their teams are taught to run what we build, with no lock-in to a reporting tool.

Reviewed annually. Full policy available on request: trust@humagined.ai

How we use AI

Humagined is an AI firm that uses AI in its own work, the same way we help clients use it. Our core rule is standard work before AI work, and it applies to us too. We use Anthropic's Claude and Claude Code to draft, analyze and build. We keep client data in the client's environment, use synthetic data for demonstrations, and label figures we can't verify. Every deliverable is reviewed by a named Humagined practitioner before it reaches you. AI speeds the work up. Accountability for it stays with us.

Questions about our AI use: trust@humagined.ai

AI in TessereX

TessereX uses AI to surface where your definitions conflict. Your people ratify every definition. Your source data stays in your environment.

AI in TessereX. TessereX uses AI in two places. TessereX Cloud agents interview your technical staff and build Terraform packages that deploy the TessereX Client to your cloud or on-premises hardware. Tess, the interview agent, collects spoken answers from metric owners, matches their definitions against ratified or documented standards, and flags where interviewees conflict so your team can align on standard definitions and calculations. The agents also recommend designs for your data and semantic models. The TessereX Client runs in your environment and uses AI to build data pipelines and deploy semantic models. Once pipelines are built, it monitors them for data compliance: stale or missing data, definition changes, misused fields, and conflicting data points.

Your data. Your source data and query results never leave the TessereX Client. TessereX Cloud receives interview answers (voice), the infrastructure details your technical staff provide for deployment, and field names and business calculations used to build your semantic model and governance documentation. That material is retained in Humagined's AWS environment for three months after the engagement ends, then permanently deleted and cannot be recovered. Everything relevant is transferred to you under your Master Services Agreement. TessereX AI runs on Llama-Tess, a fine-tuned version of Meta's Llama 3.1 8B served through Ollama and hosted by TessereX. It was fine-tuned with AI assistance (Anthropic's Claude) and reviewed by a human. No third-party AI provider receives or retains your data.

Who decides. AI proposes and your people decide. Conflicts the platform surfaces are settled in a facilitated ratification session by the owner of each value stream. No definition enters your semantic model without a named human approving it.

What you keep. Every definition and calculation is stored in your own database. At the end of an engagement you can export everything and switch TessereX off, or keep running it yourself.

How it was built. TessereX was developed with AI coding assistance (Anthropic's Claude Code). Before each major release, at least two AI agents (currently Claude and Cursor) independently run an internal SOC 2-style security review and AI-assisted penetration testing. Where the agents disagree, the issue is logged as a bug and resolved before release. Since October 29, 2026, every change also receives a manual code review by an engineer before release. Since November 2, 2026, TessereX also enforces structured exception handling and token limits on its AI agents, so a failed or over-limit task stops and is reported rather than returning partial results. TessereX intends to pursue a SOC 2 Type II audit once its controls have operated long enough to complete the audit's observation period.

Questions: TessereX, PO Box 630773, Highlands Ranch, CO 80130.

Developed with Claude Code · Runs on Meta Llama 3.1 · Built with Llama.