Harmonic Security raised $50M to watch what employees paste into AI. Here's the written rule an estimating shop needs first
Harmonic Security closed a $50M Series B to monitor how employees use AI tools. Contractors don't need the product to act on the problem: write down which bid, payroll and owner files may go into which chatbots.
Harmonic Security raised $50 million to monitor and govern how employees use AI tools. A contractor doesn't need to buy that software to act on the problem: write a one-page rule that says which files, such as sub quotes, bid tabs and certified payroll, may go into which AI tools.
What did Harmonic Security announce?
On October 14, 2026, Harmonic announced an oversubscribed $50 million Series B led by Crosspoint Capital Partners, with existing investors participating. Coverage puts its total funding at $76 million. The company says it protects employees at more than 100 companies and will spend the money on engineering and go-to-market teams in the US and UK.
The product started with browser-based AI monitoring and now also covers desktop AI apps and agentic workflows. Harmonic says it uses small language models to classify what an employee is doing and spot sensitive data, with verdicts in about 200 milliseconds. It also reports integrations with Cisco and Anthropic's Claude Compliance API.
Treat the performance claims, including a reported "96% fewer false positives" than traditional data loss prevention, as the company's own. Outlets covering the round note that no valuation, pricing or independent measurements were disclosed.
Why should an estimating shop care?
Estimators and project engineers are often the heaviest informal AI users in a contractor's office, and they handle the most sensitive paper. A typical day touches sub quotes, markup, owner-supplied drawings and spec books, and sometimes worker details for payroll.
Our shared-drive piece covered what an agent can reach. This is the other half: what a person chooses to paste out. Someone cleaning up a bid-leveling sheet in a personal chatbot account never touches your IT system.
Which files need a rule?
| File type | Reasonable default | Why |
|---|---|---|
| Sub quotes and bid-leveling sheets | Company-approved tool only | Pricing is confidential to the sub and to you |
| Owner drawings and spec books | Check the bid documents first | Some owners restrict how documents are copied or shared |
| Your own markup, overhead and margin | Never | Hard to claw back once it is outside your control |
| Certified payroll and worker IDs | Never in general chat tools | Personal data, banking details |
| Public specs and published code sections | Usually fine | Already public |
What should you do this month?
- Write a one-page AI-use rule. Name the approved tools and the plan or account type for each.
- List the "never paste" files from the table above and put it where estimators will see it.
- Ask people what they already use. An amnesty conversation finds more than a ban does.
- Read your owner contracts and bid instructions for clauses on confidentiality and third-party tools. Ask counsel if unclear.
- Check data retention and training settings on any tool you approve, as plan terms differ.
What's still unclear?
We don't know how well tools like Harmonic's work on real contractor workflows, because the evidence is the vendor's. The product is sold to enterprises, and no source says it is aimed at mid-size contractors. Monitoring also doesn't replace judgment: an approved tool can still be used on a file it shouldn't touch.
The takeaway
Before spending on software, spend an hour on the policy. Decide what sub pricing, owner documents and payroll data can touch, tell your estimators, and revisit it when you add a tool.
- What is Harmonic Security?
- Harmonic Security sells software that monitors and governs how employees use AI tools, covering browser-based AI, desktop AI apps and agentic workflows. On October 14, 2026 it announced a $50 million Series B led by Crosspoint Capital Partners.
- Can an estimator paste a bid package into ChatGPT?
- It depends on your company's policy, the AI plan being used, and any confidentiality terms in the owner's bid documents. If your firm has no written rule, that is the gap to close first.
- What is shadow AI in construction?
- Shadow AI is employees using AI tools the company hasn't approved, often on personal accounts, for work such as summarizing specs, drafting RFIs or cleaning up bid tabs. The risk is that sub pricing, owner documents or worker data leave the company's control.
- Does a small contractor need AI monitoring software?
- Not as a first step. A one-page AI-use policy, a list of approved tools and a short list of files that never go into a chatbot cover most of the exposure for a small firm.