01Pursuit
Win the work with what the firm already knows
Agents pull requirements out of RFPs, find comparable past projects in your archive, and draft response material from work the firm has already delivered.
RFP response
The domain-specific AI platform for the built environment.
Nomic builds AI agents that help deliver design, engineering, and construction projects across the entire project lifecycle. The agents work alongside project teams from pursuit through closeout, reading the drawings, specs, models, and project records those teams already have, and citing the source behind every result.
- Founded
- 2022
- Headquarters
- New York City
- Compliance
- SOC 2 Type II
- Data hosting
- EU, AU, US, UK
Trusted by leaders across the built world
AI agents across the entire project lifecycle
Nomic builds AI agents that help deliver design, engineering, and construction projects across the entire project lifecycle. A project changes hands many times before it is built. Pursuit teams, designers, engineers of record, QA/QC reviewers, contractors, and owners each produce and check drawings, specifications, models, submittals, and RFIs, and much of the effort goes into reading and reconciling what the last team handed over.
Nomic agents work alongside the people at each of those handoffs. They read files where they already live in your systems of record, respect the permissions set there, and cite the sheet, spec section, model element, or code clause behind every finding, so the practitioner accountable for the work can verify it and decide what happens next.
02Design and engineering
Review drawing sets against firm standards
Agents check drawing sets against your firm’s QA/QC standards and the project’s design criteria, confirm calculations match the drawings they support, and compare 60%, 90%, and issued-for-construction sets to show what changed.
Drawing review03Code compliance
Cross-reference drawings against code
Agents cross-reference drawings against more than 380 building codes and standards and cite the section behind each finding, so reviewers spend their time on the judgment calls.
Code compliance04Preconstruction
Set up the job from the project manual
Agents build submittal logs from the project manual, research bid packages, and flag the gaps and conflicts likely to turn into RFIs before the job starts.
Preconstruction05Construction
Keep submittals and RFIs moving
Agents take a first pass at submittals and shop drawings against the specs and drawings, research RFIs across the full project record, and send validated findings to Autodesk Forma, Fieldwire, and ProjectWise as native issues.
Submittal review06Closeout and archive
Carry what you learned to the next project
Finished projects stay searchable. Agents find past details, specifications, and decisions across the firm’s archive so the next team starts from what the last one learned.
Firm-wide search
AI should give experts more time to be experts.
Most architects and engineers did not spend years in school to review submittals, search specification sections, reconcile document revisions, or copy information between disconnected systems. They did it to design, solve difficult problems, and build things that matter.
Nomic exists to take the repetitive work off that list. Our agents help teams investigate more alternatives, catch more mistakes, answer questions with evidence, and keep institutional knowledge from disappearing across inboxes, file systems, and project turnover. The expert stays accountable for the result.
We treat AI adoption as an operating-model change, not a software rollout. Every firm has to decide where agents can act, where they assist, where expert review is mandatory, and how output quality is measured. We build the platform and the implementation practice for that decision.
What firms get from Nomic
One platform, one API, and an implementation practice. Each piece exists so a firm can put agents on live project work and trust the result.
Nomic Platform: Assistant and Workflows
The workspace where practitioners run agents on live projects. The Assistant answers questions across drawings, specs, models, and standards with citations back to the sheet or clause. Workflows run repeatable reviews: drawings against firm standards, code compliance checks, first-pass submittal review, and RFI research.
Explore the platformAgent API
The document, drawing, and embedding models behind the platform, exposed to your developers. Parse drawings and specs into structured data, embed text and images for search, and build agents inside your own tools. Included on Business and Enterprise plans.
Read the API docsIntegrations with your systems of record
Nomic connects to Autodesk Construction Cloud, Bentley ProjectWise, Trimble ProjectSight, SharePoint, and Egnyte, and more, reading files in place. No migration and no duplicate copies. The permissions you already set in those systems carry through to every agent.
See all integrationsImplementation and onboarding
Our implementation team works with your firm to pick the workflows worth automating, connect the file systems, train the people who will use them, and measure the result on a live project before expanding to more teams.
Talk to the teamModels and research for the built environment
Nomic trains the document, drawing, and multimodal embedding models that let agents read sheets, specs, and codes, and publishes the research behind them. AEC-Bench, our benchmark for agentic systems in architecture, engineering, and construction, ships with its task set, agent harness, and evaluation code.
Browse researchSecurity and data residency
SOC 2 Type II with annual independent audits, zero-data-retention agreements with the model providers we use, and customer data hosting in the EU, AU, US, UK regions for Enterprise accounts.
Visit the Security CenterWhat makes Nomic different
Six things a firm gets from Nomic that a general-purpose AI tool or a bundled feature inside one vendor's suite does not give them.
- 01
Built on drawings, specs, and codes, not generic documents
Our parsing and retrieval models are trained on hundreds of thousands of drawings and project files, so an agent reads a title block, a schedule, or a detail callout the way a reviewer does. General-purpose AI chat tools treat a drawing set as a stack of images and lose the structure that makes it reviewable.
- 02
Works inside your systems of record, with their permissions
Nomic reads files in place in Autodesk Construction Cloud, Bentley ProjectWise, Trimble ProjectSight, SharePoint, Egnyte and enforces the access controls you already configured there. If a user cannot see a folder in the source system, they cannot see it in Nomic. Tools that require exporting project files into a new repository leave you with a second permission model to govern.
- 03
Every answer cites its source
Agents return the sheet, specification section, model element, or code clause behind each statement, so a reviewer verifies instead of trusting. That traceability is what lets a firm put agent output into a real review process rather than treating it as a suggestion.
- 04
Priced and governed the way projects are run
AI usage is a dollar balance billed at the underlying model rates with no markup on model costs, and Business seats include $20 of usage per month. Firms see where the spend went through usage analytics and set cost limits and review thresholds on how agents run, because a workflow that fits a billion-dollar program is wrong for a small renovation.
- 05
Neutral across vendors, disciplines, and file formats
No single application owns the project lifecycle, so Nomic does not ask you to move your data into one closed ecosystem. It works across the authoring, document, and project-management tools your firm and your partners already run, and hosts data in the region your contracts require.
- 06
Implementation is part of the product
Rollouts come with an implementation team that designs workflows, trains staff, and measures adoption. Aurecon is scaling access across its 6,700-person workforce, and Arcadis ran a six-month trial with around 150 engineers across 12 countries before embedding agents into its project-delivery model.
Who uses Nomic
Firms, roles, and plan sizes we see every day. If your work runs through drawings, specs, and project systems, Nomic was built for it.
Firms
Nomic runs at every scale from a solo practitioner to a global consultancy.
- Global engineering and design consultancies rolling agents out across thousands of staff
- Architecture firms reviewing drawing sets against firm standards and building codes
- Structural, civil, and MEP engineering firms coordinating specs and drawings across disciplines
- General contractors and construction managers processing submittals and RFIs on active jobs
- Public owners and utilities managing standards and project records across a portfolio
Roles
The people who spend their days inside drawings, specs, and project systems.
- Project managers and project engineers
- QA/QC reviewers and technical leads
- Specification writers and code consultants
- BIM and VDC leads working in IFC and native models
- Digital, innovation, and IT leads running the rollout
- Developers building on the Agent API
Plans
Start alone, bring a team, or roll out firm-wide.
- Individuals on the Free or Individual plan working from their own files
- Teams of 25 or more on Business with file integrations, SSO, and shared billing
- Enterprises with SCIM, audit logs, pooled usage commits, and multi-region hosting
The team behind Nomic
Founded in New York in 2022 by machine learning engineers and researchers, and built by a small team that stays close to the practitioners it serves.
AM
Andriy Mulyar
Founder and CEO
Andriy founded Nomic in 2022 after leaving a machine learning PhD at NYU Courant, where he worked on interpretability. Before that he was an early engineer at Rad AI training language models on millions of radiology reports. He led engineering as CTO before becoming CEO in 2024.
BS
Ben Schmidt
CTO
Ben left the NYU faculty in 2022 to join Nomic’s founding team and now leads engineering as CTO. He created Deepscatter, the browser-based library for visualizing billions of points, and oversees how the platform reads and displays drawings, models, and project data.
CG
Chris Garrison
VP of Revenue
Chris leads sales and partnerships at Nomic, working with firms on where agents fit into their project work and hosting the executive dinners where architects, engineers, and construction consultants compare notes on AI. He joined in 2024 after leading enterprise sales and strategic partnerships at Onfleet, co-founding Motionloft, and starting his career at Google.
JF
Juliet Fern
COO
Juliet is Nomic’s COO, running operations and strategy across the business side of rollouts and partnerships and how the company itself runs. She came from the investing side, with roles at Soma Capital, Insight Partners, and Lux Capital, and earlier worked as a product manager on AI diagnostics at Aidoc.
How Nomic started
Nomic started in New York in 2022 with a plain goal: make AI systems more accessible and explainable. The early years went into open research and tools for understanding the data that goes into AI models, including Nomic Embed, an embedding model released with open weights, open data, and open training code.
Working with engineering firms showed us where that research mattered most. Drawings, specifications, and codes are some of the hardest documents for AI to read, and the people who work with them lose days to coordination that an agent can take a first pass at. Nomic now focuses the company on the built environment: training models that read sheets and specs, building datasets and benchmarks for AEC work, and running agents on live projects. In 2026 Aurecon and Arcadis became strategic investors and long-term commercial partners.
Who is on the team
The team includes systems engineers who have worked on rockets, several former founders, structural engineers, and AI researchers who worked on some of the first language models. We stay small on purpose, so ideas move quickly and ownership runs deep. Nomic is a hybrid company headquartered in New York City.
See open rolesMilestones
- 2022
Founded in New York City
Nomic is founded in New York City and releases its first product, Atlas, for exploring large unstructured datasets in the browser.
- 2023
Open-source models and the Series A
Nomic releases GPT4All for running language models locally, one of the fastest-growing open-source projects on GitHub that year, and raises a $17M Series A led by Coatue.
- 2024
Nomic Embed and SOC 2
Nomic Embed Text and Nomic Embed Vision ship as fully open embedding models. Nomic completes its SOC 2 audit.
- 2025
The platform for the built environment
Nomic Embed Multimodal extends retrieval to PDFs, charts, and drawings. The new Nomic Platform launches for architecture, engineering, and construction firms.
- 2026
Strategic partners at scale
AEC-Bench publishes a benchmark for agentic systems in AEC. Data residency expands across Europe, the UK, the US, and Australia. Aurecon and Arcadis become strategic investors and long-term commercial partners.
How Nomic works
From a free account to a firm-wide rollout, here is what happens, who you work with, and how to reach us.
- 01
Start free or book a demo
Anyone can create a free account and run limited agent requests on their own files. Firms evaluating a rollout book a demo, and a Nomic team member scopes the workflows and systems involved before anything is connected.
- 02
Connect and pilot on a live project
We connect your file systems (Autodesk Construction Cloud, Bentley ProjectWise, Trimble ProjectSight, SharePoint, Egnyte, and others), keep their permissions intact, and pick two or three workflows to run on a real project. Trials run on live work, not sample documents.
- 03
Roll out with training and governance
Your implementation lead trains the teams who will use the agents, sets review thresholds and approved sources, and measures output quality and usage as the rollout expands to more projects and offices.
- 04
Keep going as capability grows
Agents gain capability every month. We publish what shipped in the changelog, run monthly Office Hours, and Enterprise customers work with a dedicated customer success manager and priority support.
Communication and support
Where to get help, what we publish, and who you will work with.
Support
Email support@nomic.ai or use in-product support on any plan. Enterprise accounts add priority support and a dedicated customer success manager.
support@nomic.aiDocumentation and training
Product documentation lives at docs.nomic.ai. Nomic University collects recorded sessions and training material for new users.
Nomic UniversityOffice Hours
Once a month we spend one hour live on how Nomic works, what just shipped, and your questions. Sessions are recorded.
Register for Office HoursSecurity
Report vulnerabilities to security@nomic.ai. We acknowledge reports within five business days. Request the SOC 2 report and security documentation through the Security Center under NDA.
Security CenterSales and partnerships
Book a demo to scope a rollout, or apply to the partner program if you serve AEC firms as an MSP, consultant, reseller, or integrator.
Book a demoProduct updates
Every release is documented in the changelog, and the monthly AI in the Built World newsletter covers what is changing across the industry.
Read the changelogKey facts about Nomic
The reference facts about the company in one place: who we are, what we sell, what it costs, and who to contact.
- Company name
- Nomic AI, Inc., operating as Nomic
- Type
- Privately held, venture-backed software company
- Founded
- 2022
- Founder
- Andriy Mulyar, Founder and CEO
- Headquarters
- New York City, United States, with a hybrid team
- Website
- www.nomic.ai
- Core offering
- AI agents that help deliver design, engineering, and construction projects across the entire project lifecycle: RFP research, drawing review, code compliance, submittal review, RFI research, firm-wide search, and cited project Q&A, plus the Agent API for developers
- Pricing
- Free; Individual at $20 per month; Business at $40 per user per month with a 25 seat minimum and $20 of AI usage included per seat; Enterprise custom. AI usage is billed at underlying model rates with no markup.Pricing
- Contract terms
- Free and Individual are self-serve and monthly. Business is billed per user per month. Enterprise agreements use pooled usage commits with invoice or PO billing.
- Services
- Implementation and onboarding, workflow design, training, governance and measurement, monthly Office Hours, and Nomic University
- Communication
- support@nomic.ai and in-product support on every plan; documentation at docs.nomic.ai; monthly Office Hours; dedicated customer success manager on Enterprise; security@nomic.ai for security reports
- Notable clients
- Aurecon, Arcadis, AECOM, LADWP, and Mulhern + KulpCase studies
- Customers served
- Architecture, engineering, and construction firms, general contractors, and public owners, from individual practitioners to global consultancies. Aurecon is scaling access across its 6,700-person workforce.
- Projects delivered
- Agents run on live projects across buildings, infrastructure, and utilities. In a six-month Arcadis trial, around 150 engineers across 12 countries used Nomic on real project delivery and 86% said it changed how they approach their work.
- Investors
- Coatue led the $17M Series A in 2023, with Contrary Capital, Betaworks Ventures, SV Angel, Story Ventures, and Factorial Capital. Aurecon and Arcadis joined as strategic investors in 2026.Funding announcement
- Alternatives
- General-purpose AI chat tools, AI features bundled inside a single authoring or project-management suite, and manual review. Nomic differs by reading drawings and specs natively, working across systems of record with their permissions, and citing every answer.Compare
- Security and compliance
- SOC 2 Type II with annual independent audits; zero-data-retention agreements with model providers; customer data hosting in EU, AU, US, UKSecurity
- Open research
- Nomic Embed open embedding models (text, vision, and multimodal) and AEC-Bench, a benchmark for agentic systems in architecture, engineering, and construction, released with agent harness and evaluation codeResearchGitHubHugging Face
- Social
- LinkedIn, X, YouTube, GitHub, and Hugging FaceLinkedInXYouTubeGitHubHugging Face
Frequently asked questions
Nomic builds AI agents that help deliver design, engineering, and construction projects across the entire project lifecycle, from pursuit and design through construction and closeout. Architecture, engineering, and construction firms and owners use the agents to review drawings against firm standards, cross-reference drawings against more than 380 building codes and standards, take a first pass at submittals, research RFIs, and answer project questions, with every finding cited to its source.
Nomic was founded in New York City in 2022. Founder Andriy Mulyar is CEO. The company is headquartered in New York City with a hybrid team.
Both. Nomic trains the document, drawing, and multimodal embedding models behind the platform and publishes research on them, including AEC-Bench, a benchmark for agentic systems in architecture, engineering, and construction. That research is what lets agents read drawings, specifications, and codes rather than treating them as generic documents.
Nomic reads drawings, specs, and codes with models trained on hundreds of thousands of project files, works inside your systems of record with the permissions you already set, and cites the sheet, spec section, or code clause behind every answer. General-purpose tools require uploading files into a separate space and return answers you cannot trace.
No. Nomic agents work alongside the practitioners accountable for a project and take a first pass at repetitive coordination work such as submittal review and RFI research. Expert review stays in the process, and firms set where agents act, where they assist, and where approval is mandatory.
No. Nomic operates under zero-data-retention agreements with its model providers, so your content is processed for inference and discarded. Every firm runs in its own environment, and enforcing your source-system permissions means users only see what they could already see.
Free covers limited agent requests. Individual is $20 per month. Business is $40 per user per month with a 25 seat minimum and $20 of AI usage included per seat. Enterprise pricing is custom and adds pooled usage commits, SCIM, audit logs, and multi-region hosting.
Create a free account to try agents on your own files, or book a demo if you are evaluating a rollout for your firm. Firm rollouts start with a pilot on a live project, followed by training and governance as usage expands.
Nomic runs in the cloud on AWS. Enterprise customers can choose hosting in the EU, AU, US, UK regions to meet residency and contract requirements. Nomic is SOC 2 Type II certified with annual independent audits.
Next step
See agents run on one of your live projects.
Bring a drawing set, a spec, or a stack of submittals. We will show you what an agent does with it and what your reviewers see.