APPLIED_R&D // TECHNICAL STRATEGY

We turn hard organizational questions into working software.

Comprendo combines strategy consulting and engineering. We build prototypes, pilots, and production systems so you can see whether an AI idea works before you commit to it.

See our work Start a conversation

TRUSTED_BY

  • Gates Foundation
  • Heising-Simons Foundation
  • NCEE
  • Literacy Design Collaborative
  • MGT
  • Educational Service Center of the Western Reserve

PRINCIPLE_01

Strategy that ships

The same people who advise on whether to build something also write the code. Most of our projects end with software people are using.

PRINCIPLE_02

Evidence over opinion

Evaluation runs under everything we ship. Before you commit budget to an AI initiative, someone should have tested whether the thing actually works.

PRINCIPLE_03

We build our own

We also build and sell our own products, so we know what it takes to get software in front of real users.

Capabilities

WE_HELP_CLIENTS_WITH // LISTENING · SENSEMAKING · DECIDING · EVALUATING
  1. CAP_01

    Voice intelligence

    Voice and chat systems that can hold many conversations at once, in several languages. We agree on consent and data-retention rules before the first interview starts.

  2. CAP_02

    Document and knowledge intelligence

    Classification, extraction, and synthesis across thousands of documents, grounded in an organization's own corpus, with human verification wherever the stakes justify it.

  3. CAP_03

    Field and landscape intelligence

    Mapping who works in a field, how they're connected, and where the funding goes, then tracking how that changes over time.

  4. CAP_04

    Decision support and deliberation

    The analysis that sits between your data and a decision: business intelligence, predictive modeling, and structured reasoning. We show how we got to an answer and how confident we are, so the people deciding can explain their choice.

  5. CAP_05

    Evaluation and measurement

    Does the product do what it claims for the people using it, and is the AI system underneath it actually reliable? Error analysis on real traces, domain-specific graders, and judges we validate before we trust them.

Also part of most projects:

  • Data pipelines & integrations
  • Workflow automation
  • Data governance & privacy
  • Hosting & ongoing support

Selected Work

Local Government · School Boards · Commissions

Governly

CHALLENGE

Agenda packets run to hundreds of pages, and neither the staff who build them nor the members who vote on them have time to read every item closely. Decisions get made on partial preparation, and the question that should have changed the vote surfaces afterward.

WHAT_WE_BUILT

One engine that reads every agenda item five ways (summary, fiscal overview, stakeholder impact, policy analysis, and key questions) against the body's own adopted policies. It powers two tools: Agenda Prep, where staff strengthen items before they publish, and Meeting Prep, where members arrive with questions and remarks ready.

OUTCOME

Built by Comprendo for a client who owns and sells it. Bought by school districts and county commissions, and run day to day on our managed operations service.

Research · Philanthropy

Document Classification at Scale

CHALLENGE

A research organization needed specific definitions found and categorized across thousands of documents. Manual review would have taken a team months.

WHAT_WE_BUILT

An AI pipeline that scans, classifies, and flags documents. A person checks every decision that matters.

OUTCOME

Thousands of documents classified in weeks, recovering months of analyst time, with an auditable trail.

K-12 · Philanthropy

Knowledge Assistant Platform

CHALLENGE

School districts and public agencies have dense policy and program documents that their communities struggle to use, especially people who don't read English.

WHAT_WE_BUILT

A white-label conversational platform trained on each organization's own documents, with source citations and multilingual answers.

OUTCOME

Deployed across multiple organizations, from district policy navigation to enrollment support.

Healthcare · Philanthropy

Clinical Trial Protocol Assistant

CHALLENGE

Reviewing clinical trial protocols for gaps is slow, expert-bound work where misses carry real cost.

WHAT_WE_BUILT

A prototype reviewer that reads protocols and surfaces gaps and inconsistencies for expert confirmation.

OUTCOME

Demonstrated that structured AI review can compress expert review cycles.

MORE_ENGAGEMENTS

  1. Qualitative Evaluation Assistant — reviewed a philanthropic knowledge base for evidence of transformation

    Philanthropy · Evaluation and measurement

  2. Product Help Assistant — real-time, in-context help from an EdTech product's knowledge base

    EdTech · Document and knowledge intelligence

  3. Chatbot for Fentanyl Awareness — public education chatbot with a leading national fentanyl organization

    Public health · Voice intelligence

  4. Student Writing Feedback Prototype — meaningful AI feedback on student writing tasks

    K-12 · Document intelligence · Evaluation

  5. School District Policy Assistant — real-time translation and access to dense DCPS policy

    K-12 · Document and knowledge intelligence

  6. Automated Expense Management — intelligent automation for receipts and expenses

    Operations · Document and knowledge intelligence

  7. Text to Speech Coach — a coaching bot answering questions in an author's own voice

    Publishing · Voice intelligence

Leadership

Advisors

  • Brian McDaniel

    CTO, Beech Valley

  • Brad Greenlee

    Dir. Engineering, Medium

  • Noel Doromal

    Dir. Trading Tech, Tower Bay

  • Matt Sack

    GTM Strategy, Informed K12

Working on a hard question? Tell us about it.

Start a conversation