Tomok - Construction Intelligence

What the company appears to be: Tomok is a self-described AI-native construction intelligence platform designed to unify fragmented project data from tools like Primavera P6, daily reports, risk logs, and specifications into a single connected model. It claims to help owners and project managers track progress, identify delays, forecast completion dates, and extract submittal requirements using a hybrid of deterministic logic and AI reasoning.

What changed: The description states that the team built Tomok in two days as a hackathon submission, with no prior traction or revenue evidence. It is presented as an experimental prototype, not yet deployed in production.

Single most important open question: Is there any evidence that Tomok has been tested with real construction professionals or used on actual projects? The description offers no data about adoption, usage, or impact beyond a hackathon demo.

Analysis basis: This report is based entirely on the self-reported project description provided by the authors. It contains no independent verification, archived history, or third-party corroboration. All claims are treated as stated by the author and not proven.

The description states that Tomok is a construction intelligence workspace for project owners and contractor teams. Users can upload various file types including:

• P6 XER schedules
• Lookaheads
• Daily reports
• Risk registers
• Specifications
• PDFs
• Spreadsheets

It converts these scattered records into a connected project model, enabling users to:

• Compare baseline schedules, updates, and weekly commitments
• Identify critical and near-critical work using schedule float and logic
• Connect field progress and production rates to P6 activities
• Forecast completion dates and detect emerging warnings
• Extract submittal requirements from massive project specifications
• Generate draft baseline schedule activities directly from those requirements
• Question selected records through a source-grounded AI workspace
• Link every major finding back to its supporting file or exact P6 activity

The system uses Next.js, React, TypeScript, FastAPI, PostgreSQL, and OpenAI tools like the Agents SDK, ChatKit, File Search, and Responses API.

Inference: The product appears to be a hybrid application combining structured data parsing (for schedules) with AI-powered semantic search and reasoning over unstructured documents. However, it is not evidenced that this has been deployed or tested beyond the hackathon environment.

The description positions Tomok as:

• A solution to an 80% delay/overbudget problem in heavy civil construction projects.
• An AI-native OS unifying siloed 1990s software tools.
• A tool that fixes tracking issues by integrating field reality with financial reporting.
• A system that replaces manual, tedious work with AI-driven insights.
• A platform that provides transparency and verifiability, allowing users to trace AI conclusions back to source data.

It claims to be a purpose-built domain agent, not a generic document chatbot.

Inference: The positioning reflects a strong belief in solving a known industry pain point through AI, but lacks evidence of market validation or user feedback. The claim that it fixes an 80% problem is presented as a fact, though no data supports this.

The description identifies the following target customers:

• Project owners
• Contractor teams
• Schedulers
• Field teams
• Document-control staff

It also mentions role-aware dashboards, suggesting that different users have access to tailored views based on their roles.

Inference: The ICP seems to be construction professionals working in large-scale civil projects, particularly those using Primavera P6 and managing multi-million-dollar infrastructure builds. However, there is no evidence of actual customer interviews or real-world usage.

There is no pricing information or business model described in the project write-up.

The description does not mention:

• Revenue streams
• Customer acquisition costs
• Unit economics
• Monetization strategy
• Subscription tiers or licensing models

Not evidenced: No indication of how Tomok intends to generate revenue or whether it has any commercial plans beyond a hackathon prototype.

The system is built using:

• Frontend: Next.js, React, TypeScript
• Backend: FastAPI, PostgreSQL, Drizzle ORM
• AI Tools: OpenAI Agents SDK, ChatKit, File Search, Responses API, OpenAI Codex
• Data Handling: Parsing XER and Primavera XML into normalized tables; semantic retrieval for documents; bounded CSI division grouping for large specs
• Storage & APIs: Neon, Vercel Blob, Upstash

Key technical features include:

• Deterministic schedule comparisons (e.g., date variance, total float)
• AI reasoning focused on specific activities rather than entire datasets
• Source-grounded AI responses with verifiable links to original files or P6 activities
• Default-deny access controls for sensitive data
• Structured extraction of submittals from specifications

Inference: The architecture suggests a focus on domain-specific tools and structured data handling, which may be necessary for reliable performance in construction workflows. However, no evidence exists that this has been tested at scale or integrated into existing systems.

The description states:

• Tomok was built in two days as part of a hackathon.
• It is described as a purpose-built domain agent, not a generic tool.
• Testing showed it could:
  • Identify minor delays
  • Parse real Primavera schedules
  • Apply deterministic comparisons
  • Extract and organize submittals from large specs
  • Provide verifiable AI findings

However, there is no evidence of:

• Real-world deployment or pilot testing
• Customer feedback or adoption metrics
• Performance benchmarks or accuracy claims
• Production usage or integration with existing tools

Not evidenced: No traction data, user base, or performance indicators are provided beyond the hackathon demo.

The description does not mention any competitors. It implies that current solutions in construction project management are fragmented and outdated (e.g., reliance on P6 schedules, emails, and spreadsheets), but does not name specific alternatives or describe how Tomok differentiates from them.

Not evidenced: No competitive analysis or market positioning relative to existing tools is included.

• Unproven market fit: The product is described as a hackathon prototype with no evidence of real-world testing.
• No revenue model: There is no indication of how the company intends to monetize or scale.
• Highly specialized domain: Construction project management requires deep expertise and integration with legacy systems; building a working solution in two days may be unrealistic.
• Trust and confidentiality concerns: The description notes that trust was a challenge, but does not explain how this will be resolved at scale.
• AI hallucination risk: While the team claims to avoid hallucinations by focusing on specific data points, this remains a potential issue without further validation.

Inference: The lack of any commercial or user validation raises significant doubts about viability and scalability.

1. What real-world construction projects have you tested Tomok with? How many users are currently using it?
2. Have you validated the accuracy of AI-generated insights against actual project outcomes?
3. What is your plan for integrating with existing tools like Procore, Outlook, or Primavera P6 in production environments?
4. How do you intend to monetize this platform? Are there any early customers or pilot programs?
5. Can you walk us through how the system handles data consistency and reconciliation across different file formats and sources?
6. What are the key assumptions behind your claim that 80% of projects go over budget/delayed, and where is that data from?